Jensen Huang

Rayuwa a Injiniyanci, Kasuwancin Ƙirƙira da Basirar Wucin Gadi [Artificial Intelligence]

Daga yarinta a Asiya zuwa jagorancin NVIDIA Tarihin rayuwa bisa shaida da kwatanci mai nazari da Steve Jobs

Bugu na Hausa An rufe bincike a 11 Nuwamba 2026

Abubuwan da ke ciki

Gabatarwa

Fahimtar mutumin da ke bayan dandali [Platform]

Muhimmancin Jensen Huang yana cikin wani sauyi da ya taimaka a samu: kwamfutocin da aka ƙera don zana hotuna sun zama injunan da za su taimaka wa masu bincike su horar da basirar wucin gadi [Artificial intelligence]. Kamfanin da ya kafa tare da wasu a 1993 ya girma daga ƙaramin kamfanin zane [Graphics start-up] mai makoma marar tabbas zuwa babban mai samar da ababen more rayuwa [Infrastructure] da ke bayan AI na zamani [AI]. Wannan nasara ta cancanci bincike na kusa. Ba ta buƙatar labarin cewa ya san makoma, ya yi aiki shi kaɗai ko bai taɓa kuskure ba.

Wannan littafi yana bin rayuwarsa daga yarinta a Asiya, ta ƙaura, ilimi, aikin injiniyanci, kafa kasuwanci da faɗaɗar NVIDIA. Yana duba iyali da cibiyoyin da suka tsara damarsa, abokan aiki da haɗin gwiwar da suka ba kamfanin damar wanzuwa, da wuya-wuyan shawarwarin da suka sauya alkiblarsa akai-akai. Haka kuma yana tambayar wani abu da ya ƙara jan hankali yayin da ya ƙara shahara: shin ya dace a kira Jensen Huang Steve Jobs mai rai?

Amsar da aka gina a nan tana da iyaka. Dukansu sun haɗa fasaha, imanin dogon lokaci da wata bajinta ta bayyana makomar da wasu ba su iya gani a fili ba tukuna. Dukansu sun zama suna da alaƙa ta kusa da kamfanoninsu. Amma Huang ba madadin Jobs ba ne. Iliminsa na injiniyanci, ci gaban jagorancinsa ba tare da katsewa ba, abokan cinikinsa na kasuwanci da tsarin gina dandalin ƙididdiga [Computing platform] suna bambanta shi da wanda ya kafa Apple tare da wasu. Kwatancin yana amfani idan yana nuna takamaiman halaye. Yana ruɗarwa idan ya mayar da mutane biyu masu sarƙaƙƙiyar rayuwa zuwa alamomin da za a iya musanyawa.

Rayuwar da ake ci gaba da yi ba za ta sami tabbataccen babin ƙarshe ba. Wannan bugu ya tsaya a 11 Nuwamba 2026. An ɗaura kwanan wata ga alkaluman kuɗi, mukaman aiki, bayanan hukumar gudanarwa da sanarwar kayayyaki maimakon ɗaukar su a matsayin gaskiya ta dindindin. An nuna bayanan waiwayen baya a matsayin tunawa inda ya dace. Bayanin kamfani game da fasaharsa shaida ce mai amfani ga abin da ya sanar, amma ba hujja ce mai zaman kanta ga kowace da'awar aiki [Performance claim] ba.

Haka kuma wannan ba tarihin rayuwa na almara ba ne. Babu ƙirƙirarrun tattaunawa, tunanin sirri ko wuraren da aka sake ginawa aka gabatar a matsayin gaskiya. An haɗa bayanan iyali da aka rubuta a fili inda suka taimaka wajen bayyana rayuwar. Shekarun haihuwa marasa tabbaci, zato game da dalilai da bayanai masu tayar da hankali ba sa zama gaskiya saboda maimaitawa. Manufar ita ce labari mai sauƙin karantawa wanda ɗan kasuwa, ɗalibi ko ƙwararren fasaha zai iya amfani da shi ba tare da sai ya riƙa ware yabo daga ƙarin gishiri a kowane shafi ba.

1 Yarinta da ta ratsa iyakoki

Taiwan, Thailand da shawarar iyali

An haifi Jen-Hsun Huang, wanda aka fi sani da Jensen Huang, a 17 Fabrairu 1963 a Taiwan. Bayanai sun bambanta kan garin haihuwa: wasu tsoffin bayanan sun ambaci Tainan, yayin da rahotannin baya suka ambaci Taipei. Taiwan ce tabbatacciyar mafarar wurin. Bambancin yana da muhimmanci domin tarihin rayuwa bai kamata ya mayar da bayanin da ake maimaitawa zuwa tabbatacciyar gaskiya cikin shiru ba. [1]

Yarintarsa ta ratsa iyakokin ƙasa tun kafin sana'arsa. Iyali sun zauna a Taiwan, daga nan Thailand. A hirarsa da Condoleezza Rice, Huang ya tuna cewa mahaifinsa ya ƙaura zuwa Thailand don taimakawa kafa matatar mai. Ya bayyana mahaifinsa a matsayin injiniya, mahaifiyarsa kuma ta yi aikin shara da hidimar gida a makarantar Katolika bayan iyali sun koma Amurka. Waɗannan tunawarsa ne na jama'a, ba cikakken tarihin aiki da wani bincike mai zaman kansa ya sake ginawa ba. Iyayensa sun nemi damar ilimi ga 'ya'yansu maza, kuma yaran sun je Amurka kafin iyayen su kammala ƙaura. A hirarraki na baya Huang ya haɗa wannan rabuwa da fatan da iyayensa suka yi wa 'ya'yansu. [2]

Sau da yawa ana rage labarin ƙaura zuwa taƙaitaccen saƙon ƙarfafa gwiwa: talakan yaro ɗan gudun hijira ya zo, ya yi aiki tuƙuru, ya zama biloniya. Irin wannan jeri yana ɓoye yawancin abubuwan da suka sa tafiyar ta yiwu. Yara ba sa zaɓar tsarin ƙaura, canjin kuɗi, shiga makaranta ko yardar dangi su karɓe su. Iyaye suna yin shawarwari masu sakamako da bayanai marasa cika. Yaro yana gadon damar da rashin tabbas da shawarwarin suka haifar duka.

Ga Huang, motsawa tsakanin ƙasashe ya kuma nufi motsawa tsakanin harsuna da tsammanin mutane. Ƙalubalen aikace-aikace ba koyon kalmomi kawai ba ne. Ya haɗa da gane yadda mutane ke magana da malamai, kafa abota, mayar da martani ga rikici da tantance ƙwarewa. Ɗalibi na iya fahimtar lissafi kafin ya fahimci ƙa'idojin zamantakewar aji. Irin waɗannan nau'o'in koyo ba dole su yi tafiya da sauri ɗaya ba.

Zato ne a ce ƙaura ce ta jawo wata takamaiman ɗabi'ar gudanarwa kai tsaye bayan shekaru da dama. Ƙarshen magana mai hankali shi ne sabawa da sabon yanayi ya zama wani ɓangare na rayuwarsa tun farko. Sana'arsa ta baya ta ƙunshi shiga filayen da bai sani ba akai-akai, gano ƙa'idodinsu da yanke shawarar waɗanne zato za a iya ƙalubalanta. Kamanceceniya tana ba da tunani, ba gwajin halin ƙwaƙwalwa ba ce.

Ya kamata a fahimci rawar iyayensa ba tare da ƙirƙirar cikakken rumbun tarihin iyali ba. Ingancin bayanan jama'a kan sunayensu da ayyukansu ba ɗaya ba ne. Shaidar da ta fi ƙarfi kuma take maimaituwa ta shafi jajircewarsu ga ilimi da tashin hankalin ƙaura da suka yarda da shi don nemansa. Muhimmancinsu bai dogara da cike kowace gibi da bayani mai yiwuwa ba. Shawararsu ta ba Huang damar shiga cibiyoyi da hanyoyin hulɗar da suka zama ginshiƙin ci gaban fasaharsa.

Haɗarin labarin nasarar da dole sai ta zo

Sanin ƙarshen labari yana sauya yadda masu karatu ke fassara yarinta. Nasarar aji ta yau da kullum na iya zama alamar hazaka; abin da bai daɗi ba na iya zama shiri na dole ga jagoranci. Babu ɗayan sakamakon da yake zuwa kai tsaye. Ɗimbin 'yan ƙaura masu baiwa suna jure wahala ba tare da gina manyan kamfanoni ba. Yawancin 'yan kasuwa masu nasara suna da yarinta dabam sosai.

Saboda haka tambayar da ta dace ba ita ce ko an riga an ƙaddara wa ƙaramin Huang ya kafa NVIDIA ba. Ita ce yadda damar da ya samu ta faɗaɗa, da abin da ya yi da ita daga baya. Wannan tsarin yana ba ƙoƙari nauyinsa ba tare da watsi da shawarwarin iyali, ilimin jama'a da yanayin tarihi da suka sa ƙoƙarin ya yi amfani ba.

2 Kentucky da sanin yadda ake zama dabam

Zama a Oneida

Wani muhimmin gyara ya shafi zaman Huang a Kentucky. Oneida Baptist Institute ya bayyana cewa shi da ɗan'uwansa sun zauna a ɗakin kwanansa yayin da suke zuwa makarantar gwamnati ta Oneida Elementary School a hayin South Fork na Kentucky River. Saboda cibiyar makarantar sakandare ce a lokacin, yaran suna ƙetare gada mai rawa don zuwa azuzuwansu. Ba daidai ba ne a ɗauki tsarin kwana da rajistar makaranta a matsayin abu ɗaya. [3]

Labarin ya sami salo mai ban mamaki da lokaci. Huang ya tuna fara rayuwar Amurka cikin tsanani da sanya shi tare da manyan samari a yanayin da iyalinsa ba su hango sosai ba. Bayanin Oregon State ya rubuta yadda ya bayyana wannan farkon mai wahala, tare da rubuta ƙaunarsa ga ƙwarewar makarantar. Dukansu suna cikin labarin: rashin jin daɗi da godiya ba sa kore juna. [4]

Lakabin cibiyoyi yana buƙatar kulawa. Kalma kamar makarantar gyaran hali na iya isar da yadda mai magana ya tuna yanayin, amma ba lallai ta bayyana cikakken tarihin cibiyar ko kowane ɗalibi yadda ya dace ba. Abubuwan da suka fi tabbas su ne tsarin kwana, halartar makarantar firamare da babban sauyin al'ada da yaro mai nisa da iyayensa ya fuskanta.

Daga baya ana iya tuna cibiyar yarinta da takamaiman abu guda mai fitowa fili. Gada mai rawa irin wannan abu ce: a zahiri take, ana tunawa da ita kuma an rubuta ta ta hanya mai zaman kanta. Ta fi amfani da ƙirƙirar bayanin abin da Huang ya yi tunani yayin ƙetare ta. Tsarin wurin ya isar wa masu karatu da abin da ya isa game da tazarar gida, makaranta da duniyar da bai saba ba a kewaye da shi.

Bayan shekaru Jensen da Lori Huang sun tallafa wa sabon gini a Oneida. Cibiyar ta rubuta tallafin da ya yi daidai da gudummawar wasu [Matching grant] na US$2 miliyan, kuma an buɗe Huang Hall da ya biyo baya a 2019. Kyautar ta haɗa babban ɗan kasuwan da wurin da yake ɓangaren yarintarsa, ba tare da zato cewa kowane fanni na yarintar ya kasance mai daɗi ba. [3]

Juriya ba tare da ƙawata wahala ba

Bayanai kan masu nasara sukan mayar da wahala zuwa umarni: saboda wahala ta riga nasara, ya kamata shugabannin gaba su fuskanci wahala su ma. Wannan ra'ayi ne marar ƙarfi. Matsanancin hali na iya haifar da dabarar warware matsala, amma kuma yana iya iyakance dama da kawo illa mai ɗorewa. Sakamako ya dogara da tallafi, yanayi da mutumin da abin ya shafa.

Nacewar Huang daga baya a fili kan jajircewa tana da alaƙa da jagorancinsa. Bai kamata a yi amfani da ita wajen yabon cin zarafi, rashi ko azabar da za a iya kaucewa ba. Darasin mai gina abu ya fi ƙanƙanta: mutane na iya koyon amsa wa yanayi mai wuya, cibiyoyi kuma na iya taimaka musu su farfaɗo, su sami ƙarfin gwiwa da aiki mai amfani.

Cikakken bayani kan juriya ya haɗa da taimako. Saurayin da ya saba da sabon yanayi har yanzu yana buƙatar malamai, dangi, abinci, masauki da hanyar ƙarin ilimi. Shugaban gaba da ya bi ta rikici har yanzu yana buƙatar injiniyoyi, abokan ciniki da kuɗi. Tuna wannan ci gaba yana taimakawa hana tarihin rayuwa zama labarin ƙarfin zuciyar mutum shi kaɗai.

3 Oregon, makaranta, aiki da buri

Hanyar ilimi mai ƙarin natsuwa

Iyali daga baya sun zauna a Oregon, inda Huang ya yi sakandare a yankin Beaverton. Bayanin Oregon State ya kuma nuna shi a matsayin matashin ɗan wasan ƙwallon tebur da yake cikin jerin ƙasa. Waɗannan bayanai suna nuna saurayin da yake gina ƙwarewar karatu da gasa, ko da yake ba su ba da cikakken tarihin makaranta na kowace shekara ba. [4]

Wasu taƙaitattun labarai masu shahara suna ambaton ainihin jerin tsallake aji ko shekarar shiga jami'a da ta saɓa wa wasu bayanai. Irin wannan takamaimai ba ta zama dole don fahimtar hanyar ba. Abin da yake da tabbas shi ne ya ci gaba a makaranta, ya shiga fannin injiniyan lantarki [Electrical engineering] a Oregon State University kuma ya kammala digirinsa na farko a 1984. Ranar digirin da aka rubuta ta fi ƙarfi a matsayin ginshiƙi fiye da sake gina kowace shekarar ƙaramar sakandare bisa zato.

Wasanni na gasa suna ba injiniyanci kwatanci mai amfani, amma kada a ruɗe kwatanci da hujjar sanadi. Ɗan wasan ƙwallon tebur yana daidaita kansa da abokin hamayya akai-akai, yana samun amsa nan take kuma yana atisayen motsi har sai ya zama abin dogaro. Injiniya ma yana koyo ta amsa. Ba za a tantance ko Huang ya ɗauki wasu halayen wasanni cikin gudanarwa da gangan daga sanin cewa ya yi wasa kawai ba.

Aikin hidimarsa na farko a Denny's wani ɓangare ne da ake maimaitawa a labarinsa na jama'a. Ya yi magana kan wanke-wanke da yi wa abokan ciniki hidima, da darussan aikin yau da kullum. Aikin gidan abinci yana da amfani domin yana sanya ci gabansa a wajen keɓaɓɓiyar duniyar manyan fasaha. Kafin ya jagoranci injiniyoyi, ya san wurin aikin da ake ganin hidima, sauri da daidaita ƙananan ayyuka da yawa kullum. [2]

Abin da aikin hidima zai koya

Gidan abinci tsari ne. Abokin ciniki yana ganin abinci, amma isar da shi ya dogara da shiri, wankewa, oda, girki, biya da sadarwa. Gazawar sashe ɗaya tana kaiwa ga abokin ciniki ko da sauran sassan sun yi kyau. Wannan hanya ce mai amfani ta fahimtar dalilin da ya sa kyakkyawan ɓangare [Component] na fasaha shi kaɗai ba ya haifar da samfur mai nasara.

Wannan ƙa'ida ta bayyana daga baya a girma mafi yawa a kasuwancin NVIDIA. Na'urar sarrafawa [Processor] na iya zama mai ƙarfi amma da wuya a rubuta mata shiri [Program]. Manhaja [Software] na iya zama mai kyau amma ba ta da hanyar rarrabawa [Distribution channel]. Cibiyar bayanai [Data centre] na iya samun isassun na'urorin sarrafawa amma ba isasshiyar hanyar sadarwa [Networking], wuta ko sanyaya [Cooling] ba. Waɗannan daidaitawa ne na nazari, ba da'awar cewa wani zangon aiki a gidan abinci ya ba Huang taswirar kamfanin ƙididdiga [Computing company] ba.

Akwai kuma darasi kan ganin aiki. Babban shugaban kamfani na iya rabuwa da mutanen da suke sa samfur ya zama mai amfani. Aikin hidima yana nuna sakamakon ƙananan gazawar gudanarwa da sauri. Amfaninsa ga jagoranci yana cikin kiyaye sha'awar sanin waɗannan sakamako bayan matsayi da dukiya sun sa guje musu ya yiwu.

4 Jami'a, injiniyanci da haɗuwa da Lori

Oregon State da tushen sana'a

Huang ya sami digirin Bachelor of Science a injiniyan lantarki [Electrical Engineering] daga Oregon State University a 1984. Jami'ar ta rubuta cewa Lori Mills ce abokiyar gwajinsa a kwas na farko na injiniyan lantarki, kuma sun yi aure shekara biyar bayan haɗuwarsu. Dangantakarsu ta fara cikin ilimin fasahar da yake tsara sana'arsa ma. [4]

Injiniyan lantarki [Electrical engineering] ya ba shi fiye da takardar shaida. Ya horar da shi yin tunani kan tsarin zahiri [Physical systems] masu iyaka. Da'ira [Circuit] dole ta cika bukatun lokaci [Timing], wuta [Power], farashi da dogaro [Reliability] a lokaci guda. Tsarin ƙira [Design] da ya yi nasara a ma'auni ɗaya na iya gaza a wani. Wannan horo ya bambanta da kallon ƙirƙira [Innovation] a matsayin ra'ayin farat ɗaya da aiwatarwa ba tare da wahala ba.

Ɗakin gwajin jami'a yana sa waɗannan zaɓuɓɓukan sadaukarwa su zama na zahiri. Lissafi dole daga ƙarshe ya yi daidai da tsarin sassa [Components] mai aiki. Dole a binciki abin da bai dace da tsammani ba maimakon kawo masa uzuri. Waɗannan siffofi ne na ilimin injiniyanci gabaɗaya, ba jerin takamaiman ayyukan Huang da aka sake ginawa ba. Suna taimakawa bayyana irin gina ƙwararren mutum da digirinsa yake wakilta.

Kada a rage Lori zuwa mai taimakawa a labari wadda ta haɗu da biloniyan gaba bisa sa'a. Ta yi karatun injiniyanci kuma daga baya ta zama mai bayar da gudummawa tare da wani ga manyan ayyukan ilimi. Bayanai na jama'a suna goyon bayan wannan rawa. Ba su tabbatar da jingina mata kowace shawarar iyali ta sirri ba, kuma ba su ba da hujjar cewa ta riƙe mukamin NVIDIA na hukuma ba sai idan amintaccen tarihi ya tabbatar.

Stanford yayin aiki

Huang ya kammala digirin masters na injiniyan lantarki [Electrical engineering] a Stanford University a 1992 yayin da yake aiki na cikakken lokaci. Tarihin Stanford kansa ya tabbatar da digirin, yayin da Oregon State ya bayyana haɗa aiki da karatun gaba da digiri. Wannan yana sanya karatunsa na gaba cikin bunƙasar sana'arsa maimakon gabaɗaya kafin ta. [4, 5]

Wannan bambancin yana da amfani. Mutumin da yake karatu yayin aiki na iya kwatanta kayan aikin ka'ida da matsalolin kasuwanci. Karatun jami'a na iya bayyana dalilin da ya sa ƙira [Design] take aiki yadda take; aiki kuma na iya nuna iyakokin da gwajin aji bai haɗa ba. Babu muhalli ɗaya da yake maye gurbin ɗaya. Amfaninsu na iya fitowa daga motsawa tsakaninsu.

Kwanakin kuma suna ƙalubalantar labarin kasuwanci mai salo amma mai ruɗarwa. Huang ba ɗalibin da ya bar jami'a ya ƙi ilimin hukuma don gina kamfani nan take ba ne. A lokacin da aka kafa NVIDIA, yana da digirin farko da na gaba na injiniyanci da gagarumin gogewa a masana'antar na'urorin rabin jagora [Semiconductor industry]. Labarinsa yana nuna muhimmancin koyo da shirin sana'a aƙalla kamar yadda yake nuna ƙarfin gwiwar ɗaukar haɗari.

Ilimin hukuma bai ba da cikakkiyar girke-girken kafa kasuwanci ba. Ba zai tabbatar cewa abokan ciniki za su sayi guntun zane [Graphics chip], masu zuba jari za su ba da jari ko sabon dandalin manhaja [Software platform] zai jawo masu haɓakawa [Developers] ba. Ya ba da kayan aikin binciken waɗannan matsalolin. Nasarar baya ta dogara da amfani da faɗaɗa kayan aikin a cikin rashin tabbas.

5 Koyon kasuwancin na'urorin rabin jagora [Semiconductor]

AMD da LSI Logic

Tarihin NVIDIA na takardun doka ya rubuta aikin Huang a Advanced Micro Devices daga 1983 zuwa 1985, da LSI Logic daga 1985 zuwa 1993. Ranar fara AMD ta cancanci a kiyaye ta domin wasu shahararrun bayanai suna sauƙaƙa ta zuwa 1984, shekarar digirinsa na farko. Tarihin aiki na hukuma ya ba da mafi kyawun tushen jerin lokaci fiye da kyakkyawan jeri da aka zato daga kammala karatu. [6]

A AMD ya yi ƙirar ƙaramin na'urar sarrafawa [Microprocessor design]. A LSI Logic ya riƙe nauyin injiniyanci, tallace-tallace da gudanarwa, ciki har da aiki da ya shafi kasuwancin CoreWare. Waɗannan mukamai sun faɗaɗa gogewarsa bayan ƙirar da'ira [Circuit] guda. Sun nuna masa dangantakar ƙarfin fasaha, bukatun abokan ciniki da dabarun kasuwanci. [6]

Na'urorin rabin jagora [Semiconductors] suna ɗora buƙatu na musamman kan sabon kamfani. Ƙirar guntu [Chip] tana da tsada; masana'anta ta dogara da cibiyoyi na ƙwararru; kuskure na iya buƙatar sake ƙira [Redesigns] mai tsada; samfur kuma na iya zuwa bayan kasuwa ta riga ta canza. Tazarar shawarar injiniyanci da martanin kasuwa na iya tsawaita. Mai kafa kamfani a wannan masana'anta yana buƙatar hikimar fasaha da hanyar samar da kuɗi ga rashin tabbas.

Aikin LSI Logic na guntu na musamman [Custom chips] shiri ne mai dacewa don fahimtar yadda abokan ciniki ke fassara buƙatar tsari [System requirement] zuwa silicon [Silicon]. Haka kuma ya nuna bambancin da ya ci gaba da zama ginshiƙi ga NVIDIA: abokan ciniki ba kasafai suke son guntu saboda yana da ban sha'awa a fasaha kawai ba. Suna son sakamako, kamar ingantattun zane [Graphics], ƙarancin farashi, gajeren lokacin ƙididdiga [Computation time] ko samfurinsu mai ƙarin ƙarfi.

Zama fiye da ƙwararren sashe

Sauyin injiniya zuwa mai kasuwanci wani lokaci ana bayyana shi kamar barin aikin fasaha a baya. Sana'ar Huang tana nuna salo dabam. Fahimtar fasaha ta ci gaba da zama ginshiƙi, yayin da da'irar shawarwari ta faɗaɗa ta haɗa mutane, farashi, kuɗi da zaɓen kasuwa.

Ɗan kasuwa dole ya zaɓi matsalolin da suka cancanci kulawar ƙungiya. Injiniya na iya inganta [Optimise] ƙirar da aka kayyade [Design]; mai kafa kamfani dole ya kuma yanke shawarar ko ya kamata wannan ƙira ta wanzu. Tambayar tana canzawa daga ko za a iya gina abu zuwa ko mutane isassu za su ba shi daraja, ko kamfanin zai iya isar da shi da ko abokin hamayya zai amsa da sauri.

Wannan sauyi ya fi sauƙin bayyanawa a baya fiye da aiwatarwa. Ƙwarewar fasaha na iya zama tarko idan mai ita ya nace cewa ƙira [Design] mafi kyau dole ta yi nasara. Abokan ciniki na iya fifita dacewa [Compatibility], samuwa ko sauƙin amfani. Mai kafa kamfani dole ya yarda ya gano cewa fifikon kasuwa ya bambanta da zato na asalin ƙungiyar injiniyanci.

Saboda haka shekarun Huang kafin NVIDIA suna cikin tsakiyar tarihin rayuwarsa. Ba ɗakin jira kawai ba ne kafin muhimmin labarin ya fara. Sun ba da gogewar fannin da ba a iya raba shawarwarin samfur, hulɗar masana'anta da amincewar abokin ciniki.

6 Kafa NVIDIA

Masu kafa uku da kasuwa mai wuya

Jensen Huang, Chris Malachowsky da Curtis Priem sun kafa NVIDIA a 1993. Huang ya zama babban shugabanta kuma ya ci gaba a mukamin yayin da kamfanin yake girma. Shahararrun tarukan Denny's suna cikin labarin kafawarsa, amma gidan abincin bai kamata ya ɓoye tarin ƙwarewar waɗanda suke zaune a ciki ba. [7, 8]

Damar farko tana cikin zane [Graphics] ga kwamfutocin mutum [Personal computers]. Wasanni [Games] sun ba da amfani mai buƙatu masu yawa: masu amfani suna son duniyoyin gani masu wadata, yayin da na'urorin sarrafawa na gama-gari [General-purpose processors] suke fama da samar da zane masu ƙara sarƙaƙiya cikin inganci. Na'urar sarrafawa ta musamman [Specialist processor] za ta iya inganta ƙwarewar ta yin wasu ayyuka da sauri sosai.

Wannan mafarar kasuwanci ce, ba cikakken shirin tattalin arzikin AI mai samarwa [Generative AI] na yau ba. Yana da sauƙi a mayar da nasarar yanzu zuwa baya a yi tunanin masu kafa sun riga sun ga kowace amfani ta gaba. Fassarar da ta fi ƙarfi ita ce sun gano babbar matsalar ƙididdiga [Computing problem] kuma suka shiga ta kasuwar da za ta iya riƙe kasuwanci.

Bambancin babban ra'ayi da samfur na farko yana da muhimmanci. Ra'ayin na iya ci gaba da amfani ko da aiwatarwa [Implementation] ta gaza. Kamfani na iya gaskata cewa ƙididdiga ta musamman [Specialised computation] za ta yi amfani amma ya yi kuskure kan tsarin gine-gine [Architecture], abokin ciniki ko lokacin yunƙurin farko. Tsirarsa sai ta dogara da sauya yunƙurin ba tare da watsi da kowane amfani na ainihin fahimtar ba.

Dalilin muhimmancin abokan kafa

Tarihin rayuwa da ya ta'allaka ga shugaban kamfani na iya goge kamfanin da yake kewaye da shi ba da gangan ba. Malachowsky da Priem abokan kafa ne, ba kayan taimakon wata cikakkiyar manufa ba. Injiniyoyi, masu zuba jari da abokan ciniki na farko su ma sun taimaka sauya ra'ayi zuwa ƙungiya. Ya fi sauƙin tantance jagorancin Huang daidai idan gudummawarsu ta kasance a bayyane.

Malachowsky ya kawo gogewar injiniyanci da shugabancin fasaha daga Hewlett-Packard da Sun Microsystems. Tarihinsa a tsarin zane [Graphics systems] ya sanya shi cikin tushen fasahar kamfanin, kuma ya ci gaba da alaƙa da NVIDIA a matsayin mai kafa da babban masani [Fellow]. Kamfanin ba misalin ɗan kasuwa marar ilimin fasaha ne da kawai ya ɗauki injiniyoyi su aiwatar da cikakken shiri ba. [32]

Priem ma ya yi aiki a Sun, inda ya tsara gine-ginen kayayyakin zane [Graphics products] na GX. Rensselaer Polytechnic Institute ya rubuta cewa ya yi aikin babban shugaban fasaha na NVIDIA daga 1993 zuwa 2003. Aikinsa na baya ya haɗa IBM Professional Graphics Adapter. Waɗannan gogewa sun sanya masu kafa cikin jerin injiniyancin zane [Graphics engineering] da ya riga NVIDIA, kuma suna taimakawa bayyana dalilin da ya sa tallata GPU [GPU] na farkonsa ba za a ruɗe shi da asalin fannin gabaɗaya ba. [33]

Gudummawar Huang tana tare da wannan ƙwarewar. Ya yi aiki tsakanin ƙirar na'urorin rabin jagora [Semiconductor design] da gudanarwar kasuwanci a AMD da LSI Logic. Haɗin ya ba ƙungiyar kafa ra'ayoyi kan aiwatar da zane [Graphics implementation] da aikin gina kasuwancin na'urorin rabin jagora [Semiconductor business]. Bai kare su daga kuskure ba, amma ya ba su kayan gane kuskure da mayar da martani.

Aiki na musamman na babban shugaba shi ne haɗa shawarwari. Damar fasaha dole ta zama shirin samfur; shirin samfur yana buƙatar kuɗi; kuɗi suna buƙatar amincewa; amincewa kuma dole ta tsallake cikas. Babu mutum guda da zai iya yin kowane aikin injiniyanci ko kasuwanci. Jagoranci ya haɗa da kiyaye alkibla mai ma'ana a tsakaninsu.

Kamfanin na farko ya kuma yi aiki cikin tsarin haɗin gwiwar [Ecosystem] da bai sarrafa ba. Tsarukan aiki [Operating systems], ƙa'idojin zane [Graphics standards], masu haɓaka manhaja [Software developers] da abokan masana'anta za su iya sauya darajar ƙirarsa. Wannan dogaro ba raunin NVIDIA shi kaɗai ba ne. Yanayin gina sabon sashi [Component] ga babban tsarin ƙididdiga [Computing system] ne.

Saboda haka darasin kafawa ya fi fara da ƙarfin hali kawai tsauri. Shi ne zaɓen matsalar da ta cancanci warwarewa, tara ƙwarewa masu cika juna da gina kasuwancin da zai iya koyo kafin kuɗinsa su ƙare. Ƙarfin hali yana da amfani, amma ba zai maye gurbin waɗannan tsare-tsare ba.

7 Lokacin da kamfani ya kusa durƙushewa

Matsalar NV1 da Sega

Kayayyakin farko na NVIDIA ba su samar da hanya mai santsi zuwa nasara ba. NV1, wanda aka ƙaddamar a 1995, ya samu wahalar kasuwanci. Aikin da kamfanin ya yi da Sega daga baya ya nuna matsala mafi zurfi: alkiblar fasaharsa tana ƙara rabuwa da ƙa'idojin [Standards] da kasuwar zane [Graphics market] mafi faɗi take tsarawa a kansu. [7, 9]

Matsalar tsarin gine-ginen [Architecture] ta kasance takamaimai. NV1 ya yi amfani da taswirar siffa ta quadratic [Quadratic texture mapping], yayin da tsarin haɗin gwiwar [Ecosystem] Direct3D da yake tasowa ya ta'allaka kan alwatika [Triangles]. Goyon bayan hanya dabam ya ƙara wa masu haɓaka wasanni [Game developers] aiki kuma ya iyakance amfanin na'urorin NVIDIA [Hardware]. RIVA 128 ya ɗauki alkibla mafi dacewa. Darasin wani ɓangare ya shafi ƙa'idoji: aiwatarwa [Implementation] mai kyau na iya faɗuwa idan masana'antar manhaja [Software industry] da take kewaye ba za ta iya amfani da ita cikin sauƙi ba. [34]

Wannan ba gasa kawai ba ce kan wace siffa ta fi a lissafi. Tsarin gine-ginen zane [Graphics architecture] dole ya yi aiki da kayan ƙirƙira [Authoring tools], hanyoyin rubuta shiri [Programming interfaces] da dabarun da masu haɓakawa suka riga suka fahimta. Mai samarwa da yake neman abokan ciniki su ɗauki hanya dabam dole ya bayar da amfanin da ya isa ya tabbatar da ƙarin aikin. Tsirar NVIDIA ta dogara da karɓar alkiblar tsarin haɗin gwiwar [Ecosystem] maimakon tsammanin tsarin zai sake shirya kansa a kan ƙirarsa ta farko.

A bayaninsa na baya kan rikicin, Huang ya bayyana gane cewa ci gaba da kwangilar zai iya barin NVIDIA da samfurin da ba zai goyi bayan makomarta ba, yayin da barinta zai iya barin kamfanin ba tare da kuɗin tsira ba. Ya tuntuɓi shugabancin Sega ya nemi taimako. Tallafin da ya biyo baya ya ba NVIDIA damar sauya alkibla. [9]

Bayanai kan tsarin sun bambanta yadda suke bayyana biyan kuɗi da dabarar zuba jari. Ƙarshen magana mafi aminci shi ne Sega ta samar da tallafin kuɗi mai muhimmanci, wanda aka fi rahoto a kusan US$5 miliyan, kuma Shoichiro Irimajiri ya taka rawa mai muhimmanci. Kada a sauƙaƙa shi zuwa labarin cewa shugaban Nintendo da ba shi da alaƙa ya ceci kamfanin, ko cewa kuɗin dole kyautar mutum ce marar sharaɗi. [8, 10]

Muhimmancin ɗabi'a takamaimai ne. Dole Huang ya yarda da babban kuskure ga abokin haɗin gwiwa mai muhimmanci. Wannan ya bambanta da nuna kansa a matsayin mai hangen nesa da ya yi daidai tun farko. Ceto mai gamsarwa ya buƙaci wani haɗin faɗin gaskiya, amincewa da yardar abokin haɗin gwiwa ya karɓi sakamako marar sabo.

RIVA 128 da dawowar begen kasuwanci

RIVA 128, wanda aka gabatar a 1997, ya zama samfurin juyin nasara mai muhimmanci. Jadawalin tarihin kamfanin NVIDIA ya rubuta sayar da raka'a miliyan ɗaya a watanni huɗu. Wannan adadin yana nuna yadda samfur zai iya sauya matsayin sabon kamfani da sauri idan ya fi dacewa da bukatun kasuwa. [7]

Tallafin Sega da nasarar RIVA 128 suna cikin labarin tsira mai alaƙa. Ɗaukar kowane mataki a matsayin fatarar kuɗi dabam yana ƙara gishiri ga tarihi. Ɗaya ya samar da sararin yin aiki; ɗayan ya taimaka sauya sabuwar alkibla zuwa kuɗin shiga. Dukansu sun dogara da aikin injiniyanci da aiwatar da kasuwancin ƙungiya mafi girma.

Wa'adin gaggawa na iya sa ƙungiya ta mai da hankali, amma kuma yana haifar da haɗari. Guntun da aka gina [Chip] cikin ƙuntataccen lokaci da kuɗi dole a gwada [Testing] shi da kulawa ta musamman saboda ƙila babu damar sake yunƙuri. Darasin ba shi ne za a iya tsallake gwaji idan mai kafa ya yi imani sosai ba. Shi ne ƙungiyoyi masu iyaka suna buƙatar hanyoyin da za su nuna kuskure kafin farashin gyara ya zama mai hallakarwa.

Shin Huang ya ceci NVIDIA daga fatara sau da yawa

Akwai shaida mai gamsarwa ta kusan durƙushewa akai-akai, musamman tsakiyar shekarun 1990. Mark Stevens, daraktan dogon lokaci, ya bayyana cewa kamfanin yana da kimanin kwanaki talatin daga gazawa a wasu lokuta a wannan zangon. Wannan yana goyon bayan furucin tsira daga rushewa sau da yawa. Ba ya tabbatar cewa NVIDIA ta shiga shari'ar fatara ta hukuma akai-akai. [8]

Ya kamata a kuma rarraba gazawar dabarun baya daidai. CUDA ya haɗa da sadaukarwar kuɗi mai tsada da amfaninta ya ɗauki lokaci ya bayyana. Yunƙurin wayar hannu ya haɗa da ja da baya daga kasuwa mai wahala. Faɗuwar farashin hannun jari, aikin da ya yi tsada da rashin iya biyan kuɗi nan take yanayi ne dabam. Kiran dukansu fatara yana sa labarin ya fi ban mamaki amma ya rage amfaninsa.

Maganar da ta fi ƙarfi a iya karewa ita ce Huang ya taimaka wa NVIDIA ta tsira daga lokuta masu haɗari da yawa, tare da gaggawar ƙarewar kuɗi mafi bayyane a shekarun farko. Shawarwarinsa sun kasance masu muhimmanci, amma haka ma tallafin Sega, aikin abokan aiki, haƙurin masu zuba jari da nasarar kasuwancin kayayyakin da aka sake ƙira.

8 Daga kayayyakin zane [Graphics products] zuwa dandalin ƙididdiga [Computing platform]

Shiga kasuwar hannun jari da ayyana rukuni

Bayar da hannun jari ga jama'a a karon farko [Initial public offering] na NVIDIA a 1999 ya nuna sabon matakin ci gabanta. Kamfanin ya kuma gabatar da GeForce 256 a wannan shekarar, yana tallata shi a matsayin na'urar sarrafa zane ta farko [Graphics processing unit]. Sharadin tarihi yana da muhimmanci: masu hanzarta zane [Graphics accelerators] sun wanzu kafin shi. Da'awar NVIDIA ta shafi wani takamaiman ra'ayin GPU [GPU] da aka haɗa, ba ƙirƙirar duk kayan aikin zane na kwamfuta [Computer graphics hardware] ba. [7]

Sunan kamfanin a jama'a ya ci gaba da alaƙa sosai da wasanni [Gaming]. Wannan ba ƙaramin gabatarwa ga kasuwanci mafi mutunci ba ne. Wasanni sun samar da abokan ciniki masu buƙatu masu yawa, zagayowar kayayyaki [Product cycles] masu sauri da kasuwa mai girman da zai iya tallafa wa gagarumin jarin injiniyanci. Sun kuma ba masu bincike na'urorin sarrafawa a layi ɗaya [Parallel processors] masu ƙarfi da za su iya gwadawa.

Dandali [Platform] yana tasowa idan wasu za su iya gina aikin da zai daɗe a kan samfur. Katin zane [Graphics card] mai sauri zai iya sa a saya sau ɗaya. Muhallin rubuta shiri [Programming environment] mai ƙarfi, ɗakunan shiri [Libraries] masu amfani da tarin ilimin masu haɓakawa [Developer knowledge] na iya shafar sayayya da yawa a gaba. Wannan dangantakar kasuwanci ce dabam domin sauya mai samarwa na iya buƙatar sauya manhaja [Software] da sake horar da mutane.

Motsawa zuwa dandali [Platform] yana sauya ma'anar inganci ma. Saurin kayan aiki [Hardware] har yanzu yana da amfani, haka kuma takardun bayani [Documentation], dacewa da tsofaffin sigogi [Backward compatibility], gyara kurakurai [Debugging], sanya aiki [Deployment] da tallafi [Support]. Amfanin gwajin ma'auni [Benchmark advantage] na iya ɓacewa idan masu amfani ba za su iya samunsa tabbatacce a manhajojinsu [Applications] ba.

Gasa ba ta ɓace ba

Masana'antar zane [Graphics industry] ta ci gaba da canzawa ta gasa da haɗewar kamfanoni. NVIDIA ta amince ta sayi wasu kadarorin zane [Graphics assets] daga 3dfx a Disamba 2000, kuma an kammala cinikin a Afrilu 2001. Kada a bayyana sayen kadarorin da aka zaɓa a matsayin sayen kowane ɓangare na kasuwancin abokin hamayya kawai. Bambancin yana da muhimmanci a doka da tarihi. [35]

Zanen wasanni [Gaming graphics] ma ya ci gaba da bunƙasa bayan zurfin koyo [Deep learning] ya shahara. A Agusta 2018 NVIDIA ta gabatar da kayayyakin GeForce RTX bisa tsarin gine-ginen Turing [Architecture], suna haɗa inuwa ta al'ada [Shading] da ƙwarewa ta musamman ta bin haske [Ray tracing] da AI [AI]. Wannan ƙarin yunƙuri ne na sa dabarun gani da suka yi tsada a baya su yi amfani a kayayyaki masu amsa hulɗa [Interactive products]. [36]

Bin haske [Ray tracing] yana yin kwaikwayon hanyoyin haske don samar da abubuwa kamar nunawa da inuwa. Manhajojin lokacin gaske [Real-time applications] suna fuskantar tsauraran iyakokin lokaci: hoton da yake ɗaukar mintoci a samar [Render] na iya amfani a aikin da ba na kai tsaye ba [Offline production], amma ba zai iya tallafa wa wasa mai saurin amsa kai tsaye ba. Don haka ƙalubalen ba kawai lissafa hoto mai gamsarwa ba ne, sai dai yin hakan cikin iyakar aiki [Performance budget] mai amfani.

Waɗannan ci gaba suna rikitar da tunanin cewa NVIDIA ta bar wasanni [Gaming] kawai ta zama kamfanin AI [AI company]. Zane [Graphics] ya ci gaba da kasancewa filin fasaha mai aiki, kuma dangantakar zane, kwaikwayo [Simulation] da AI [AI] ta ci gaba da samar da dama. An fi fahimtar tarihin kamfanin a matsayin faɗaɗar ƙwarewa fiye da cikakken maye gurbin ainihi ɗaya da wani.

Yadda GPU [GPU] ya bambanta da CPU [CPU]

An ƙera na'urar sarrafawa ta tsakiya [Central processing unit] don ayyuka iri-iri, ciki har da sarƙaƙƙiyar tafiyar umarni [Control flow] da ayyukan da dole su bi jeri. Na'urar sarrafa zane [Graphics processing unit] tana ware yawancin ƙirarta don yin ayyuka masu alaƙa da yawa a layi ɗaya [Parallel]. Kwatancin ba cewa ɗaya mai hankali ne ɗayan kuma mai sauƙi ba, ko GPU yana da sauri koyaushe. Ƙarfinsu ya dogara da nauyin aiki [Workload].

A yi la'akari da aiwatar da canji [Transformation] iri ɗaya a kan miliyoyin ɗigon hoto [Pixels]. Yawancin aikin za a iya rarraba shi zuwa ayyuka masu kama. Kayan aiki masu layi ɗaya [Parallel hardware] za su iya sarrafa da yawa tare. A gefe guda, aikin da kowace shawara ta dogara da sakamakon wadda ta gabata na iya samun ƙarancin amfani daga GPU. Motsa bayanai [Data] tsakanin na'urorin sarrafawa [Processors] kuma na iya cin lokaci da makamashi.

AI na zamani [AI] ya dogara sosai da ayyukan lissafi [Numerical operations] da za a iya shirya su a layi ɗaya [Parallel work], ciki har da manyan ninkewar matrix [Matrix multiplications]. Wannan yana sa GPU [GPUs] su yi amfani, amma ba ya sa kowane aikin AI ya zama iri ɗaya ko kowane girke-girken GPU [GPU deployment] ya yi tattalin arziki. Ƙarfin ƙwaƙwalwa [Memory capacity], sadarwa [Communication], daidaiton lissafi [Numerical precision] da aiwatar da manhaja [Software implementation] duka suna shafar sakamakon.

Damar Huang ta dabaru ita ce haɗa waɗannan ƙwarewa da matsalolin da suka wuce zana hotuna [Graphics]. Hakan ya buƙaci fiye da sanar da cewa guntun zane [Graphics chip] yana da ƙarfi. Ya buƙaci sa kayan aiki [Hardware] su zama abin isa ga mutanen da aikinsu bai da alaƙa sosai da samar da hoton wasa [Rendering].

9 CUDA da gina tsarin haɗin gwiwa [Ecosystem] cikin haƙuri

Sa ƙididdiga a layi ɗaya [Parallel computing] ta zama mai amfani

CUDA dandalin ƙididdiga a layi ɗaya [Parallel-computing platform] da samfurin rubuta shiri [Programming model] na NVIDIA ne. Bunƙasarsa ta sa masu rubuta shiri su iya amfani da GPU na NVIDIA masu goyon baya [GPUs] don aikin lissafi na gama-gari [General-purpose numerical work] maimakon bayyana kowace matsala ta hanyar zane [Graphics interface]. NVIDIA ta gabatar da tsarin gine-ginen [Architecture] a 2006, kayan aikin manhaja na jama'a [Software toolkit] kuma suka biyo baya a 2007. [7, 11]

Bambancin tsarin gine-gine [Architecture] da tarin kayan aiki [Toolkit] yana bayyana dalilin da ya sa kwanakin duka suke cikin tarihi. Ranar guda na iya ɓoye jerin yadda ƙarfin kayan aiki [Hardware capability], kayan rubuta shiri [Programming tools] da karɓar masu haɓakawa [Developer adoption] suka samu. Muhimmancin kasuwanci ya fito daga haɗuwarsu.

Ba dole masanin kimiyya ya zama ƙwararren zane [Graphics specialist] don cin gajiyar kayan aiki na layi ɗaya [Parallel hardware] ba. Ɗakunan shiri [Libraries] na iya bayar da ayyukan lissafi da aka sani [Mathematical operations], yayin da mai fassara shiri [Compiler] da tsarin tafiyarwa [Runtime] suke taimakawa sauya lambar shiri [Code] zuwa aikin da GPU zai iya yi. Da lokaci, tsarin aiki na sama [Higher-level frameworks] na iya dogara da ɗakunan shirin, suna sanya yaduddukan ɓoye wahala [Layers of abstraction] tsakanin mai bincike da na'urar sarrafawa [Processor].

Wannan tsarin yadudduka ɗaya ne daga dalilin da ya sa ba za a fahimci matsayin NVIDIA ta ƙirga ƙayyadaddun guntu [Chip specifications] kawai ba. Masu haɓakawa [Developers] suna gina halaye, ƙungiyoyi suna tara lambar shiri da aka gwada [Tested code], al'ummomin bincike kuma suna raba misalai. Waɗannan jari suna samar da ci gaba mai amfani, amma kuma suna iya samar da kuɗin sauyawa [Switching costs]. Mai amfani ɗaya na iya ganin tsarin haɗin gwiwa [Ecosystem] a matsayin sauƙi, wani kuma a matsayin dogaro.

Zuba jari kafin ribar ta bayyana

Huang ya bayyana CUDA a matsayin sadaukarwar dogon lokaci mai tsada da ta fara jawo shakku. Shawarar ta kawo farashi kafin kasuwar AI [AI market] ta baya ta sa ribar kasuwancin ta bayyana. Bayanin waiwayensa ya haɗa jarin da lokacin raunin amincewar kasuwar kuɗi. [9]

Kuskure ne a ɗauka cewa kowane aikin tsada ya cancanci kuɗi har abada saboda CUDA ya yi nasara daga ƙarshe. Jarin dogon lokaci mai hankali yana buƙatar shaida cewa tushen fasaharsa yana inganta kuma masu amfani za su iya yin wani abu mai daraja da shi. Haƙuri ba tare da koyo ba zai iya riƙe gazawa kamar yadda zai iya riƙe babban ci gaba.

Ƙalubalen dabaru shi ne tallafa wa yiwuwar abu ba tare da neman babbar kasuwa nan take ba. Masu binciken da suka yi amfani da GPU [GPUs] ga manufofin da ba a zata ba sun ba da bayanai. Aikinsu ya nuna cewa kayan aiki [Hardware] za su iya warware matsalolin da suka wuce abokan cinikin da suka fara ba da kuɗinsa. NVIDIA za ta iya koyo daga waɗannan amfani yayin da take ci gaba da sayarwa a kasuwannin da suke akwai.

Iyakar amfanin tsarin haɗin gwiwa [Ecosystem advantage]

Tsarin manhaja [Software ecosystem] mai ƙarfi ba ya kawar da gasa. Masu hanzartawa na daban [Alternative accelerators], ayyukan masu fassara shiri [Compiler projects] da tsarukan da ake iya ɗauka [Portable frameworks] na iya rage dogaro da takamaiman mai samarwa. Manyan abokan ciniki na iya samun girman da zai sa kayan aiki na musamman [Specialised hardware] su dace. Wasu nauyin aiki [Workloads] na iya cin gajiyar tsari mafi sauƙi ko arha maimakon dandalin GPU [GPU platform] mafi sauƙin sauyawa.

Jarabawar da take ɗorewa ita ce ko cikakken tsarin NVIDIA yana ci gaba da samar da aiki [Performance] mai amfani a farashi mai karɓuwa. Saba wa manhaja [Software] na iya jinkirta sauyawa, amma ba za ta tabbatar da amincin abokin ciniki har abada ba. Shi ya sa tsarin haɗin gwiwa [Ecosystem] dole ya ci gaba da ƙirƙirar daraja maimakon dogaro kawai da wahalar barinsa.

Ga 'yan kasuwa, CUDA yana ba da darasi mai tsauri: wani lokaci samfur mafi muhimmanci shi ne kayan aikin da suke ba abokan ciniki damar gano amfani da mai samarwa bai fara kayyade ba. Gina irin kayan aikin yana buƙatar tallafi mai ɗorewa. Yana kuma buƙatar karɓar cewa daraja na iya fara bayyana a ƙananan gwaje-gwaje maimakon babban littafin oda mai tabbas.

10 Muhimmin sauyin zurfin koyo [Deep learning]

AlexNet da sabon saƙo daga bincike

A 2012 Alex Krizhevsky, Ilya Sutskever da Geoffrey Hinton sun wallafa aikin da aka fi sani da AlexNet. Tsarin hanyoyin jijiyoyinsu mai convolution [Convolutional neural network] ya sami babban sakamako a gasar ImageNet, yana amfani da ƙididdigar GPU [GPU computation] don sa horo [Training] mai faɗi ya yiwu. Takardar tana da muhimmanci domin tana rubuta nasarar bincike maimakon fassarar kamfani ta baya. [12]

NVIDIA ba ta ƙirƙiri wannan tsarin hanyoyin jijiyoyi [Neural network], tarin bayanan ImageNet [Dataset] ko dukkan fannin koyon na'ura [Machine learning] ba. Kayan aikinta [Hardware] da tsarin shirye-shiryenta [Programming ecosystem] suna cikin ababen more rayuwar [Infrastructure] da suka ba da dama. Tarihi mai adalci yana kiyaye masu bincike da masu samar da bayanai [Data creators] tare da kamfanin da yake samar da ƙididdiga [Computation].

Muhimmancin abin da ya faru ba shi ne gwajin ma'auni [Benchmark] guda ya tabbatar da kowace manhajar AI [AI application] ta gaba ba. Ya nuna cewa haɗin bayanai [Data], ƙirar samfurin lissafi [Model design] da ƙarfin ƙididdiga da yake akwai [Computing power] zai iya kawo babban ci gaba ga aiki mai muhimmanci. Hakan ya ƙarfafa ƙarin jari da gwaji. Sakamakon alama ce mai tasiri, ba cikakkiyar hasashen makoma ba.

Ga Huang, batun dabaru shi ne yadda za a amsa alamar. Kamfani na iya ganin sakamakon bincike mai bege ba tare da shirya kansa ya yi hidima ga fannin da yake tasowa ba. Tallafa wa zurfin koyo [Deep learning] ya buƙaci zaɓin kayan aiki [Hardware], ɗakunan manhaja [Software libraries], tsaruka da hulɗa da masu bincike. Ya nufi ɗaukar nauyin aikin [Workload] a matsayin alkiblar da kamfanin zai iya gina kanta.

Horo [Training] da samar da amsa [Inference]

Horo [Training] yana daidaita ma'aunan samfurin lissafi [Model parameters] ta bayanai [Data] da tsarin ingantawa [Optimisation process]. Samar da amsa [Inference] yana amfani da samfurin da aka horar don samar da sakamako [Output]. Dukansu na iya buƙatar ƙididdiga [Computation] mai yawa, amma bukatun tafiyar da su sun bambanta. Horo na iya tafiyar da manyan ayyuka na tsawon lokaci; samar da amsa na iya buƙatar amsa wa mutane da yawa da sauri yayin sarrafa farashi da jinkiri [Latency].

Wannan bambanci yana taimakawa bayyana dalilin da ya sa ababen more rayuwar AI [AI infrastructure] ba kasuwa guda ɗaya marar bambanci ba ce. Ɗakin bincike da yake horar da babban samfurin lissafi [Model], asibitin da yake tafiyar da ƙaramin samfurin hoto [Imaging model] da manhajar wayar hannu [Mobile application] da take samar da rubutu na iya buƙatar tsare-tsare dabam. Tsari mafi kyau ya dogara da girma, sirri [Privacy], dogaro [Reliability], lokacin amsa [Response time] da yawan aikin da za a yi.

Yayin da samfuran lissafi [Models] suka ƙara girma, sadarwa [Communication] tsakanin na'urorin sarrafawa [Processors] ta ƙara muhimmanci. Tarin guntu [Chips] masu sauri ba kwamfuta mai sauri kai tsaye ba ne. Dole bayanai [Data] su motsa ta ƙwaƙwalwa [Memory] da hanyoyin sadarwa [Networks], kuma a gano gazawa a sarrafa ta. Matsalar makale [Bottleneck] na iya kasancewa a wajen sassan lissafi [Arithmetic units] da ake yawan yabawa.

Za a fahimci faɗaɗar NVIDIA zuwa tsaruka [Systems] da hanyoyin sadarwa [Networking] a kan wannan asali. Rukunin gasa ya wuce katin zane [Graphics card] guda zuwa aikin babban tsarin ƙididdiga [Computing installation]. Rawar Huang ta ƙara zama mai gina tsari da tsara haɗin gwiwa [Ecosystem] maimakon shugaban mai samar da guntu [Chip supplier] kawai.

Mayar da buƙatar bincike zuwa kayayyaki

A 2016 NVIDIA ta gabatar da DGX-1 a matsayin tsarin zurfin koyo da aka haɗa [Integrated deep-learning system]. Maimakon sai kowane ɗakin bincike ya haɗa na'ura daga sassa daban, ya ba da haɗin GPU [GPUs] da manhaja [Software] da aka tsara. Ra'ayin kasuwancin shi ne rage tazarar samun kayan aiki [Hardware] da yin bincike mai amfani. Bayanin NVIDIA cewa shi ne babban kwamfutar zurfin koyo ta farko [Deep-learning supercomputer] da'awar talla ce, ba cewa babu kwamfuta ta baya da ta horar da tsarin jijiyoyi [Neural network] ba. [28]

A shekara ta gaba, tsarin gine-ginen Volta [Architecture] ya gabatar da Tensor Cores, sassa na musamman don ayyukan da suke da muhimmanci ga zurfin koyo [Deep learning]. Wannan ƙarin mataki ne daga sake amfani da ƙarfin zane [Graphics capability] zuwa ƙirar kayan aiki [Hardware] a kan ƙididdigar AI [AI-related computation]. Ya kuma nuna dalilin da ya sa dandali [Platform] zai iya ci gaba da zama ɗaya yayin da na'urorinsa na ciki suka canza sosai. Masu haɓakawa [Developers] na iya cin gajiyar sabon kayan aiki ta sabunta ɗakunan shiri [Libraries] ba tare da sake rubuta kowane sashe na manhaja [Application] ba. [29]

Hopper, wanda aka sanar a 2022, ya ƙunshi Transformer Engine domin hanzarta nauyin aiki [Workloads] masu alaƙa da samfuran transformer [Transformer models]. Blackwell, wanda aka sanar a Maris 2024, ya faɗaɗa tsarin NVIDIA na dukan na'ura [System-level approach] zuwa AI mai samarwa [Generative AI]. Waɗannan sanarwa suna nuna sauyin alkibla da ƙarfin samfur; manyan ninkin aikin [Performance multipliers] da ake ambata sun dogara da nauyin aiki, daidaito [Precision] da tsarin da aka kwatanta, kuma kada a ɗauke su a matsayin sakamakon kowa. [30, 31]

Jerin ya fi jerin sunayen samfuran muhimmanci. NVIDIA ta daidaita kayan aiki [Hardware] da manhaja [Software] akai-akai da sauyin aikin da abokan ciniki suke son yi. Kamfanin ya buƙaci kiyaye isasshiyar dacewa [Compatibility] don kare jarin da yake akwai yayin gabatar da siffofin inganta aikin [Performance] gaba. Wannan tashin hankali yana tsakiyar jagorancin dandali [Platform leadership]: yawan sauyi na iya fasa tsarin haɗin gwiwa [Ecosystem], ƙarancinsa kuma na iya sa ya tsufa.

11 Cikas bayan tsira

Ingancin samfur da koma bayan 2008

Tsira daga shekarun farkon kafa kamfani [Start-up] bai sa NVIDIA ta tsira daga kuskure ba. A Agusta 2008 ta rahoto cirewa daga riba [Charge] na US$196 miliyan da ya shafi tsammanin kuɗin garanti, gyara, mayarwa da maye gurbin wasu kayayyakin kwamfutar tafi-da-gidanka [Notebook products] da haɗin kayan gini marar ƙarfi ya shafa. Sanarwar ta kuma bayyana asarar rubu'in shekara da yarda da kuskuren farashin gasa. [13]

Wannan wani irin rikici ne dabam da ƙarancin kuɗin farko. Ya shafi ingancin samfur, kuɗin abokan ciniki da amincin aiwatarwa. Tarihin rayuwa da yake mai da hankali ga dabaru masu kyau kawai yana rasa wannan fannin jagoranci. Abokan ciniki sun dogara da kayayyaki su yi aiki bayan an kai su, kuma shugaba mai nasara har yanzu yana da alhakin kurakuran da suke da nisa da dandalin babban jawabi.

Takardar shekara ta kamfanin ta shekarar kuɗi 2009 ta rahoto kusan US$1.26 biliyan a kuɗi, daidaitattun kuɗi [Cash equivalents] da takardun jari masu sayuwa [Marketable securities] a ƙarshen shekara. Wannan matsayin kuɗi ba ya goyon bayan cewa kowace wahalar 2008 wata fatarar nan take ce. Asarar, cirewar [Charge] da koma bayan gabaɗaya suna da tsanani ba tare da wannan lakabi ba. [14]

Ja da baya daga wayar hannu [Mobile]

NVIDIA ta kuma bi kasuwar wayoyin hannu sannan ta janye. Bayanin Huang ya gabatar da ja da bayan a matsayin shawarar da ta buɗe kulawa ga wasu dama, ciki har da aikin mutum-mutumi [Robotics] da motocin tuƙa kansu [Autonomous vehicles]. Wannan shaida ce ta yadda ya fassara ƙwarewar, ba hujjar cewa an tabbatar sauran damar za su yi nasara ba. [9]

Akwai bambanci tsakanin sauya shiri saboda yana da wahala da sauya shi saboda tattalin arzikinsa na asali ya daina jan hankali. Shugabanni suna buƙatar isasshen imani su tsallake juriya ta yau da kullum da isasshiyar sassauci su gane lokacin da nacewa ta zama ɓarna. Babu taken da zai iya bambanta waɗannan yanayi cikin tabbas tun farko.

Horon da yake da amfani shi ne fito da zato a fili. Wace buƙatar abokin ciniki ake yi wa hidima? Wane amfani na fasaha [Technical advantage] har yanzu za a iya karewa? Nawa ƙarin jari ake buƙata? Wace shaida za ta sa kamfani ya tsaya? Waɗannan tambayoyi suna mayar da ja da baya zuwa rarraba albarkatu bisa dalili maimakon labarin baya da yake sa kowace shawara ta yi kama da hikima.

Sayen kamfanin [Acquisition] da bai faru ba

A Fabrairu 2022 NVIDIA da SoftBank sun soke shirin sayen Arm [Acquisition], suna ambaton manyan matsalolin ƙa'idojin gwamnati. Sanarwar ta ce SoftBank za ta riƙe US$1.25 biliyan da NVIDIA ta biya a gaba, NVIDIA kuma za ta riƙe lasisin Arm [Licence] na shekara ashirin. Cinikin da bai cika ba babban koma bayan dabaru ne, ko da yake ba gazawar injiniyan guntu [Chip engineering] ba ne. [15]

Lamarin yana nuna iyakar ikon mai kafa kamfani. Ciniki na iya zama mai ma'ana ga masu goyon bayansa amma ya tayar da damuwa ga masu kula da doka, abokan ciniki ko masu gasa. Burin fasaha yana aiki cikin cibiyoyin doka da siyasa. Babban shugaba ba zai ɗauke su a matsayin iyakokin injiniyanci [Engineering constraints] da jajircewa kaɗai za ta kawar ba.

Amsar da ta dace ba sake rubuta sayen da ya gaza [Acquisition] a matsayin nasara marar sharaɗi ba ce. Ita ce gane farashi, fahimtar dalilin gazawar cinikin da tantance zaɓin kamfanin na gaba a kan cancantarsu. Amincin labarin jagoranci yana ƙaruwa idan zai iya bayyana shawarar da ba ta yi nasara ba ba tare da tilasta mata ƙarshen cin nasara ba.

12 Gina kamfanin ababen more rayuwa na AI [AI infrastructure]

Bayan guntu [Chip] guda

Yayin da nauyin aikin AI [AI workloads] ya faɗaɗa, kayayyakin NVIDIA suka ƙara zama sassan manyan tsaruka. GPU [GPUs] suna buƙatar ƙwaƙwalwa mai saurin wuce bayanai [High-bandwidth memory], haɗin sauri, hanyoyin sadarwa [Networking], sabar [Servers], manhajar gudanarwa [Management software] da tallafi. Sayen Mellanox, wanda aka kammala a 2020, muhimmin mataki ne na ƙarfafa ɓangaren sadarwar tsarin. [16]

Babban aikin horo [Training job] yana nuna matsalar. Idan na'urar sarrafawa [Processor] ɗaya ta jira bayanai [Data] daga wata, kayan aiki [Hardware] masu tsada za su iya zaman banza. Idan hanyar sadarwa [Network] ba ta da tabbas, aiki mai tsawon lokaci na iya rasa aiki mai amfani. Idan sanyaya [Cooling] bai isa ba, tsarin ba zai iya ci gaba da aikin [Performance] da sassansa [Components] suka yi alƙawari ba. Sakamakon abokin ciniki ya dogara da daidaitawa a dukkan abin da aka girka [Installation].

Wannan yana taimakawa bayyana dalilin jan hankalin tsarin da aka haɗa [Integrated systems]. Suna rage wa abokin ciniki nauyin haɗa kowane sashe [Component] da tabbatar da shi [Validating] dabam. Suna kuma iya ƙara dogaro da mai samarwa. Abokin ciniki yana samun mafita mafi cika amma ƙila ya yarda da zaɓuɓɓuka kaɗan na yadda za ta haɓaka.

Labarin Huang na ƙididdiga mai hanzari [Accelerated computing] ya dace da wannan sauyi. Tambayar da ta dace ta zama yawan aiki mai amfani da tsari zai iya kammalawa cikin wani lokaci, makamashi da kuɗi. Guntu [Chip] mai rahusa na iya fi tsada a aikace idan yana buƙatar na'urori da yawa ko ya jinkirta sakamako mai amfani. A gefe guda, tsarin da ya ci gaba na iya zama kuɗin da bai zama dole ba idan mai sauƙi ya isa.

Abin da bunƙasar AI [AI boom] ta tabbatar da wanda ba ta tabbatar ba

Saurin karɓar AI mai samarwa [Generative AI] ya ƙara buƙatar ababen more rayuwar ƙididdiga [Computing infrastructure]. NVIDIA ta amfana domin ta riga ta shafe shekaru tana haɓaka kayan aiki [Hardware] da manhaja [Software] da suka dace da nauyin ayyukan [Workloads]. Wannan lokacin ya yi ƙarfi a kasuwanci. Kada a ruɗe shi da zama ita kaɗai ta haifar da bunƙasar AI [AI boom].

AI na zamani [AI] ya kuma dogara da binciken jami'a, tarin bayanai na jama'a [Public datasets], manhaja mai buɗaɗɗen tushe [Open-source software], ababen more rayuwar gajimare [Cloud infrastructure], masana'antar na'urorin rabin jagora [Semiconductor manufacturing] da aikin ɗakunan bincike masu gasa. Masu haɓaka samfuran lissafi [Model developers] sun yanke shawarar gine-gine [Architecture] da horo [Training]; kamfanonin kayan aiki [Hardware companies] sun samar da kayan; masu amfani da masu zuba jari sun tsara buƙata. Gudummawar Huang ta yi yawa cikin wannan hanyar haɗin, ba mai zaman kanta da ita ba.

Sakamakon kuɗi ya kasance abin lura. Sanarwar NVIDIA ta 26 Agusta 2026 ta rahoto US$96.2 biliyan na kuɗin shiga ga rubu'i na biyu na shekarar kuɗi 2027, wanda ya ƙare 26 Yuli 2026, ciki har da US$89.0 biliyan daga Data Center [Data Center]. Waɗannan alkaluman rubu'in shekara da aka rahoto ne, ba jimillar shekarar kalanda ko hasashe ba. [17]

Sanarwar ta kuma bayyana Vera Rubin yana shiga cikakkiyar samarwa [Full production]. Ya kamata a karanta irin bayanan kamfani cikin yanayin kwanan watansu. Suna tabbatar da abin da NVIDIA ta sanar, yayin da aikin [Performance], girkawa don amfani [Deployment] da ribar tattalin arziki [Economic return] na kowane shigarwa [Installations] suna buƙatar shaidarsu. Taswirar samfur [Product roadmap] da amfanin da abokin ciniki ya gani suna da alaƙa amma ba ɗaya ba ne.

Haɗarin da yake cikin girma na ban mamaki

Saurin girma na iya ɓoye zato marasa ƙarfi. Abokan ciniki na iya saya kafin buƙatar gaba. Ababen more rayuwa [Infrastructure] na iya ɗaukar lokaci wajen haɗa wuta fiye da shiri. Sabbin samfuran lissafi [Models] na iya buƙatar kayan aiki [Hardware] dabam ko su ƙara inganci. Jarin da yake jan hankali idan ana amfani sosai [Utilisation] na iya zama mai tsada idan na'urori suna zaman banza.

Saboda haka ya kamata a yi nazarin ƙaruwar kuɗin shiga tare da tattarar abokan ciniki [Customer concentration], alkawurra [Commitments], ƙarfin samarwa [Supply capacity] da tattalin arzikin kasuwancin abokan cinikin. Mai samarwa zai iya sayarwa cikin nasara a bunƙasa yayin da wasu sassan masana'antar kewaye suke wahalar samun isasshiyar riba. Wannan yiwuwar ba ta tabbatar da kumfar tattalin arziki [Bubble] ba. Dalili ce ta gwada sarkar tattalin arzikin maimakon yabon mai girma da sauri kaɗai.

13 Tsarin jagoranci

Bayanai kai tsaye da ƙa'idoji masu tsauri

Tattaunawar Huang ta jama'a kan gudanarwa tana jaddada samun bayanai, fahimtar yanayi tare [Shared context] da shiga matsalolin fasaha [Technical problems] masu wuya. A tattaunawar 2024 a Stripe Sessions ya bayyana wasu sassan ƙungiyar NVIDIA da yadda yake amfani da amsa kai tsaye [Direct feedback]. Waɗannan maganganu shaida ce mai amfani ga falsafar da ya bayyana, amma har yanzu bayanin mutumin da yake jagorantar kamfanin ne. [18]

Ƙungiya mai ƙarancin matakai [Flatter organisation] za ta iya rage jinkirin matsala da mutanen da suke iya magance ta. Za ta kuma iya sanya babban nauyin bayanai kan shugaban kamfani. Amfanin irin tsarin ya dogara da ƙwarewar mutane, bayyanar ikon yanke shawara [Decision rights] da ƙarfin warware fifiko masu karo.

Saboda haka sauƙaƙawa ce a ba kowane kamfani shawarar kwafin yawan masu kai rahoto kai tsaye. Taswirar ƙungiya [Organisation chart] fuskar tsarin da yake aiki kawai ce. Kamfanoni biyu na iya samun taswirori masu kama da al'adu dabam sosai. Abin da yake da muhimmanci shi ne ko bayanan da suka dace suna tafiya, ko ana fahimtar shawarwari da ko mutane suna iya yin aiki ba tare da jiran sama marar dalili ba.

Manyan ƙa'idoji wani abu ne da ake maimaitawa a bayanan Huang. Amma neman aiki [Performance] mai kyau ba gudanarwa mai kyau kai tsaye ba ne. Tambayar ita ce yadda ake sanar da ƙa'idoji, ko tsammanin suna yiwuwa, ko saɓanin ra'ayi ba shi da haɗari da ko kuskure yana kaiwa koyo. Shugaba zai iya zama mai tsauri ba tare da mai da wulaƙanci hanyar inganta aiki ba.

Bayyana makoma kafin a saba da ita

Mai kafa kamfani na fasaha [Technical founder] dole ya shawo kan masu sauraro masu buƙatu dabam. Injiniya yana son sanin ko tsarin gine-gine [Architecture] yana aiki. Abokin ciniki yana son sanin ko yana warware matsala. Mai zuba jari yana tambayar ko kasuwancin zai samu riba. Gwamnati na iya kula da tsaro [Security], ƙarfin cikin gida da damar shiga [Access].

Gabatarwar Huang a jama'a sau da yawa tana haɗa ci gaban fasaha [Technical development] da babban sauyin masana'anta. Wannan bajinta tana da darajar dabaru domin abokan ciniki ba za su iya saya ko gina bisa makomar da ba su fahimta ba. Bayani mai kyau zai iya daidaita jari a tsakanin ƙungiyoyi da yawa.

Wannan bajintar ma tana haifar da alhaki. Labari mai gamsarwa na iya zarce shaida. Gwajin nunawa [Demonstration] na iya aiki cikin yanayin da aka zaɓa da kulawa. Kwatancin aiki [Performance comparison] na iya dogara da nauyin aiki [Workload] ko sigar manhaja [Software version] na musamman. Alƙawarin ƙwarewar gaba na iya zama kamar samfurin da yake samuwa yanzu. Yayin da mai ba da labari ya ƙara ƙwarewa, rarraba waɗannan rukuni ya ƙara muhimmanci.

Tattarar hukunci

Tsawon zaman mai kafa na iya kiyaye ci gaban fasaha [Technical continuity] da kare jarin da ba zai tsira da gajeren hangen gudanarwa ba. Yana kuma iya tara hukunci a hannun mutum guda. Abokan aiki na iya ƙware wajen hango abin da mai kafa yake so maimakon gwada ko daidai ne.

Kamfani mai juriya yana buƙatar cibiyar ƙarfi da hanyoyin gyara ta duka. Hukumar gudanarwa, nazarin fasaha [Technical reviews], ra'ayin abokin ciniki [Customer feedback] da tabbacin mai zaman kansa [Independent assurance] duka suna taimakawa. Amfaninsu ya fi idan suna iya ƙalubalantar shugaban da ya yi nasara ba tare da an kore su a matsayin masu hana hangen nesa ba.

Daga ƙarshe ya kamata a tantance jagorancin Huang ba kan yadda NVIDIA take aiki da kyau yayin da yake jagorantar ta da kansa kawai ba, har da yadda ƙungiyar za ta kiyaye hukunci mai amfani, alhaki da ƙirƙira bayansa. Wannan jarabawa ce mafi wuya da kwarjini, kuma ta ƙara muhimmanci yayin da kamfanin ya ƙara nauyin tattalin arziki.

14 Abin da ke motsa mutum, buri da ma'anar gazawa

Abin da bayanan jama'a za su iya tabbatarwa

Huang ya yi magana akai-akai kan rashin tabbas, aiki tuƙuru da koyo daga kuskure. Jawabinsa na kammala karatu a National Taiwan University a 2023 ya yi amfani da misalan gazawa da ja da baya na dabaru don bayyana ci gaban kamfanin. Waɗannan bayanai suna goyon bayan hoton shugaban da yake ɗaukar cikas a fili a matsayin ɓangaren ƙirƙira [Innovation]. Ba sa ba da damar shiga dalilansa na sirri ba tare da iyaka ba. [9]

Tarihin rayuwa yakan ba mutum injin motsin zuciya guda: tsoron talauci, buƙatar yabo ko son nuna masu suka sun yi kuskure. Irin bayanan suna jan hankali domin suna sa rayuwa mai sarƙaƙiya ta yi sauƙin tunawa. Idan babu shaida mai ƙarfi, zato ne.

Bayani mai horo zai iya nazarin zaɓuɓɓukan da ake gani. Huang ya zauna a kamfani guda tsawon shekaru da dama. NVIDIA ta ci gaba da zuba jari a tsarin shirye-shirye [Programming ecosystem] kafin kasuwarta ta baya ta bayyana. Kamfanin ya bar wasu ayyuka yayin da ya zurfafa wasu. Waɗannan shawarwari suna nuna fifiko ba tare da buƙatar ƙirƙirar abin da ya ji a kowace safiya ba.

Damuwa mai amfani da iyakarta

Sanannen tunanin cewa kamfani koyaushe yana kusa da gazawa na iya ƙarfafa taka tsantsan. Yana tuna wa mutane cewa abokan ciniki suna da zaɓi kuma amfanin fasaha [Technical advantages] yana tsufa. Amma gaggawar dindindin kuma na iya gajiyar da ƙungiya ko ta sa saɓanin ra'ayi cikin tunani ya yi wuya.

Bambancin yana cikin abin da gargaɗin yake haifarwa. Yana kaiwa ga ingantaccen gwaji [Testing], saurin gane matsalolin abokan ciniki da shiri mai gaskiya? Ko yana ƙarfafa gaggawar yanke shawara da ɓoye mummunan labari? Tarihin rayuwa zai iya nuna wannan tashin hankali ba tare da cewa kowane ma'aikaci ya fuskanci al'adar iri ɗaya ba.

Hakazalika gazawa tana amfani ne kawai idan ƙungiya ta koyi abin da za ta iya aiwatarwa. Maimaita gwajin [Experiment] da bai yi nasara ba ba tare da sauya zato ba ba juriya ba ce. Gwajin da aka tsara da kyau yana samar da bayanai ko da sakamakon kasuwancinsa ya ba da takaici. Marar kyau na iya cin kuɗi ya bar babbar tambayar ba ta da amsa.

Horon sauya ra'ayi

Lamarin Sega na farko yana da darasi musamman domin ya buƙaci Huang ya raba ainihi da zaɓin fasaha [Technical choice]. Mai kafa na iya zuba suna, kuɗi da shekaru na aiki a ƙira [Design]. Yarda cewa ba daidai ba ce na iya zama kamar yarda cewa kamfanin kansa ba daidai ba ne. Amma tsira na iya dogara da yin wannan bambancin daidai.

Babban darasi shi ne kiyaye sadaukarwa ga matsala yayin barin mafitar da aka gabatar ta sauya. Tarihin NVIDIA yana da misalai da yawa: gine-ginen zane [Graphics architectures] sun canza, manhaja [Software] ta zama ginshiƙi, AI [AI] ya ƙara muhimmanci kuma an bar wasu kasuwanni. Ya kamata a yi nazarin tsarin a matsayin jerin shawarwari, ba a ɗaga shi zuwa tabbacin cewa kowane sauyin alkiblar [Pivot] gaba zai yi nasara ba.

15 Lori, Spencer da Madison

Iyali a cikin labarin jama'a

Jensen da Lori Huang sun haɗu a matsayin ɗaliban injiniyanci a Oregon State kuma suka yi aure a 1985. Suna da manyan 'ya'ya biyu, Spencer da Madison. An rubuta taimakon iliminsu na haɗin gwiwa da kyau. Bayan wannan tarihin jama'a, ya kamata tarihin rayuwa ya mutunta bambancin ayyukan iyali na fili da rayuwarsu ta sirri. [1, 4]

Yana da ma'ana a gane cewa gina kamfani na shekaru da dama yana shafar gida. Ba shi da ma'ana a ƙirƙiri tattaunawa, yarjejeniya ko abubuwan zuciyar 'yan iyalin da ba su bayyana kansu a fili ba. Kiran mata sirrin da yake bayan kowace nasarar kasuwanci na iya zama yabo amma yana goge rawarta ta kanta.

Haka ma a yi taka tsantsan game da mukamin Lori na hukuma yanzu, kadarorin gidauniyar ko takamaiman kwanakin sana'a. Waɗannan suna buƙatar takardun da suke da kwanan wata. Kada a zato su daga girman dukiyar iyali ko a kwafe su daga taƙaitaccen bayani marar tushe.

Madison Huang

Tarihin jama'a na NVIDIA na yanzu ya bayyana Madison Huang a matsayin babbar daraktar samfur da tallan fasaha ga dandalin AI na zahiri [Physical AI platforms]. Ya bayyana nauyin Omniverse, Cosmos da Isaac, ya kuma rubuta tsohon aikinta na talla a ƙungiyar Louis Vuitton Moët Hennessy da MBA a Marketing and Strategy daga London Business School. [19]

Waɗannan muhimman bayanan sana'a ne. Suna nuna hanyar shiga fasaha ta talla da ilimin kasuwanci maimakon maimaita aikin ƙirar guntu [Chip design] na mahaifinta kawai. Ba a buƙatar da'awar ainihin shekaru ko cikakken jadawalin horon girki don bayyana aikinta na yanzu.

Tallata dandalin fasaha [Technical platform] ba ƙirƙirar suna kawai ba ne. Ya haɗa da bayyana ƙwarewa ga masu haɓakawa [Developers] da abokan ciniki, fahimtar matsalolin karɓuwa [Adoption barriers] da haɗa aikin injiniyanci da kasuwa. Ingancin aikin a tantance shi ta sakamakonsa, ba zato daga dangantakar iyali ba.

Spencer Huang

Tarihin fasaha na NVIDIA ya bayyana Spencer Huang a matsayin daraktan samfur na aikin mutum-mutumi [Robotics]. Ya bayyana aiki a kwaikwayo [Simulation], koyon mutum-mutumi [Robot learning], bayanai da aka ƙirƙira [Synthetic data] da ababen more rayuwar [Infrastructure] da ake amfani da su don horar da na'urori masu cin gashin kansu [Autonomous machines] da sanya su aiki. Ya kuma rubuta MBA na fasaha daga New York University da ya mai da hankali ga basirar wucin gadi [Artificial intelligence]. [20]

Hanyar sana'ar da aka bayyana a jama'a ta bambanta da farkon sana'ar mahaifinsa. Wannan yana da muhimmanci domin iyali na iya raba kamfani ba tare da raba horo ko nauyi iri ɗaya ba. Mukami shi kaɗai ba ya tabbatar da zurfin gudummawar fasahar mutum, amma kuma asalin iyaye bai ba da hujjar ƙin gudummawar ba.

Aikin iyali da gudanarwa

Idan dangin mai kafa suna aiki a kamfanin jama'a, tambayoyin gudanarwa na yau da kullum suna biyo baya. Ana bitar mukamai da albashi yadda ya kamata? Hanyoyin kai rahoto suna a fili? Sauran ma'aikata za su iya yin gasa cikin adalci? Ana bayyana ciniki da masu alaƙa yadda ya dace?

Waɗannan tambayoyi ba sa tabbatar da laifi. Suna cikin horon cibiyar da yake ƙara muhimmanci yayin girman kamfani. Haka kuma sunan iyaye ba zai maye gurbin tantance aikin ɗa da kansa ba.

Aiki kuma ba sanarwar gadon shugabanci ba ne. Kasancewar Spencer da Madison a NVIDIA ba ya tabbatar cewa an zaɓi ɗayansu ya zama babban shugaba. Zato game da daular iyali ta fasaha na iya jawo hankali, amma tarihin rayuwa mai muhimmanci ya kamata ya raba aikin yanzu da muƙamin gaba da ba a tabbatar a fili ba.

16 Mayar da alheri ta ilimi

Cibiyoyin da suka ci gaba da muhimmanci

Aikin alheri na Huangs a fili yana da alaƙa mai ƙarfi da ilimi. Oregon State ta sanar da kyautarsu ta US$50 miliyan a 2022 ga Jen-Hsun and Lori Huang Collaborative Innovation Complex. Aikin yana haɗa injiniyanci, ƙididdiga [Computation] da bincike tsakanin fannoni. [21]

Bayanin Stanford kan Huang Engineering Center ya rubuta alkawarin ma'auratan na US$30 miliyan. Bayanin Oneida ya rubuta tallafin da yake daidai da gudummawar wasu [Matching grant] na US$2 miliyan ga ginin da aka buɗe a 2019. Kyaututtukan suna haɗa matakan rayuwar Huang dabam: yarinta, digirin farko da karatun gaba. [22, 3]

Ya kamata a karanta adadin a matsayin alkawurran da aka bayyana a sanarwar cibiyoyin. Alkawari, ƙimar kadara da kuɗin da aka riga aka bayar ba abu ɗaya ba ne. Haka kuma kada a ruɗe alherin iyali da gidauniyar kamfanin NVIDIA ko jarin kasuwancinsa.

Abin da taimakon ilimi zai iya cimmawa

Ginin jami'a ba sakamakon ilimi ba ne shi kaɗai. Darajarsa ta dogara da mutanen, kayan aiki da shirye-shiryen da yake tallafawa. Gine-ginen da aka tsara da kyau na iya ba da haɗin gwiwar da yake wahala a baya, ba ɗalibai damar samun kayan aiki masu ci gaba da jawo masu binciken matsaloli masu muhimmanci.

Dangantakar mai bayar da gudummawar fasaha da jami'a ma ta cancanci bincike mai tunani. Gudummawa na iya faɗaɗa dama yayin da take shafar fifiko. Cibiyoyi suna buƙatar 'yancin ilimi, gudanarwa mai bayyana komai da fahimtar wajibcin da yake tare da kyauta.

Saboda haka mafi kyawun labarin alheri yana guje wa tsattsauran ra'ayi biyu. Ba ya ɗaukar dukiyar da aka bayar a matsayin hujjar kyawawan ɗabi'u kai tsaye, kuma ba ya zato cewa kowace kyauta tallata kai kawai ce. Yana binciken alkawarin da aka rubuta, manufar mai karɓa da sakamako idan shaida ta samu.

A wannan tarihin rayuwa, maimaituwar alaƙar ilimi tana da muhimmanci. Sana'ar Huang ta dogara da horon fasaha da cibiyoyin da suka ba shi damar bunƙasa. Tallafa musu yana ƙirƙirar haɗi bayyananne tsakanin nasarar mutum da damar wasu. Yawan abin da haɗin zai cimma daga ƙarshe tambayar shaida ce da lokaci.

17 Hujjar kwatanta Huang da Steve Jobs

Kwatanci mai amfani yana fara da ma'ana

Kiran Huang Steve Jobs mai rai na iya nufin abubuwa dabam. Zai iya nufin shugaban fasaha mai kwarjini, mai kafa kamfanin da yake sauya masana'anta ko mutumin jama'a da zai iya sa kayayyaki masu sarƙaƙiya su yi armashi. Waɗannan kwatanci ne da za a gwada. Ba zai iya nufin a zahiri yana da hali, sana'a ko rawar tarihi iri ɗaya ba.

Hujjar da ta fi ƙarfi tana fara da iya haɗa fasaha da babbar dama. Jobs ya taimaka sa ƙididdigar mutum [Personal computing] da na'urorin hannu [Mobile devices] su zama abin fahimta a matsayin ƙwarewar da mutane za su so. Huang ya taimaka sa ƙididdiga mai hanzari [Accelerated computing] ta zama abin fahimta a matsayin ababen more rayuwa [Infrastructure] da masu bincike da kasuwanci za su yi aikin da bai yiwu a aikace ba a baya.

Kwatancin kuma ya dogara da haɗawa [Integration]. Sanarwar iPhone ta Apple a 2007 ta gabatar da na'urar da ta haɗa sadarwa, kafofin labarai da ayyukan intanet ta wata hanyar mu'amala [Interface] da ake sarrafawa sosai. Dabarar NVIDIA tana haɗa na'urorin sarrafawa [Processors], hanyoyin sadarwa [Networking] da manhaja [Software] zuwa dandali [Platforms] ga masu haɓakawa [Developers] da ƙungiyoyi. A dukansu amfanin abokin ciniki ya dogara da yadda sassan suke aiki tare. [23]

Imani kafin amincewar jama'a

Duk sana'o'in suna da shawarwarin da ba su da tabbas kafin a yaba musu. Jawabin Stanford na Jobs kansa ya bayyana darajar da daga baya ya samu a gogewar da ba ta fara zama shirin sana'a a fili ba, kuma dawowarsa Apple ta biyo bayan zamansa a wajen kamfanin. Dogon jarin Huang a CUDA ya riga shaharar zurfin koyo [Deep learning] a kasuwanci gabaɗaya. [24, 9]

Kamanceceniyar ba ita ce ɗayansu ya iya ganin makoma daidai ba. Ita ce duka sun yarda su shirya babban ƙoƙari bisa ra'ayin yadda fasaha za ta ƙara amfani. Hangen nesa yana samun muhimmancin aikace-aikace ne idan ya sauya shawarwarin samfur, jari da aikin wasu.

Ba da labari a fili a matsayin kayan aiki

Gabatarwar kayayyakin Jobs ta zama ɓangaren asalin kasuwancin Apple. Gabatarwar Huang ma ta sa ya zama mai fassara dabarar NVIDIA ga jama'a. Tufafin da ake haɗawa da kowane mutum suna da sauƙin tunawa, amma tushe ne mai rauni na kwatanci. Kamanceceniyar da ta fi ƙarfi ita ce rawarsu wajen fassara zaɓin fasaha zuwa labari mai haɗuwa.

Wannan zai iya taimaka wa ƙungiya ta haɗu a kan alkibla. Ma'aikata suna fahimtar abin da samfur zai cimma. Abokan ciniki suna ganin yadda zai dace da shiryensu. Abokan haɗin gwiwa suna da dalilin zuba jari a aikin da yake cika shi. Saboda haka labari zai iya taimakawa aiwatarwa maimakon ƙawata ta kawai.

Amma ikon gamsarwa yana haifar da haɗari ma. Masu sauraro na iya fara ɗaukar ƙarfin gwiwar shugaba a matsayin shaida. Sanarwa da aka gabatar da kyau na iya ɓoye tambayoyin injiniyanci ko kasuwanci da ba a warware ba. Ya kamata yabo ya sa masu karatu su ƙara kula da tambayoyin, ba su rage ba.

Fassarar da za a iya karewa

Za a iya kare furucin Steve Jobs na zamanin AI [AI] a matsayin salon rubutu ga mai kafa mai tasiri da yake haɗa imanin dabaru, haɗawa [Integration] da sadarwar jama'a. Yana ƙara ƙarfi idan an bayyana waɗannan ɓangarori a fili.

Wannan tsarin har yanzu yana buƙatar iyaka. AI [AI] yana da masu ginawa da yawa kuma ba shi da mai shi guda. Muhimmancin Huang ba ya sa kowane mai haɓaka samfurin lissafi [Model developer], mai bincike ko mai ƙirar kayan aiki [Hardware designer] mai gasa ya zama ɗan wasa mai taimako a labarinsa. Haka kuma kwatancin ba ya tabbatar cewa kayayyaki ko shawarwarinsa za su sami matsayi iri ɗaya na dogon lokaci a al'ada kamar na Apple.

Saboda haka hujjar goyon baya tana da ƙarfi amma tana da iyaka. Za a iya kwatanta Huang da Jobs a matsayin shugaban da ya taimaka wa wasu su gane su karɓi sabuwar alkiblar fasaha. Kwatancin ba takardar daidaiton halayya ko hasashen gadon tarihi iri ɗaya ba ne.

18 Hujjar ƙin lakabin Steve Jobs mai rai

Hanyoyi dabam na shiga fasaha

Jobs da Huang sun isa jagoranci ta shirye-shirye dabam. Jobs ya bar Reed College kuma daga baya ya bayyana tasirin darussan da ya nema a wajen hanyar digiri ta al'ada. Huang ya kammala digirin farko da na gaba na injiniyan lantarki [Electrical engineering] yayin gina gogewar na'urorin rabin jagora [Semiconductor]. [24, 4]

Kada wannan bambanci ya zama gasa tsakanin ilhami da ilimi. Jobs ya yi aiki da injiniyoyi masu ƙwarewa sosai kuma ya fahimci muhimmancin aiwatarwa [Implementation]. Nasarar Huang ma ta buƙaci hukunci, sadarwa da tunanin ƙirƙira da ba a iya ragewa zuwa takardar shaida. Abin nufi shi ne babu tsarin ilimi guda na jagorancin fasaha.

Abokan ciniki dabam da ganin jama'a dabam

Kayayyakin Apple da suka fi shahara sun sanya shawarwarin kamfanin kai tsaye a hannun masu amfani. Jin daɗin amfani da na'ura ya kasance ginshiƙin darajar da ake bayarwa [Value proposition]. Yawancin kasuwancin AI [AI] na NVIDIA yana samar da ababen more rayuwa [Infrastructure] ga wasu ƙungiyoyi, waɗanda kayayyakinsu za su iya ɗaukar sunan kasuwanci dabam gabaɗaya.

Bambancin yana shafar tsarin sayarwa, fifikon injiniyanci da ma'anar nasara. Mai amfani zai iya yanke hukunci cewa hanyar mu'amala [Interface] ta fi kyau cikin mintoci. Babban kasuwanci na iya ɗaukar watanni yana tantance yawan aikin da ake iya wucewa [Throughput], dacewar manhaja [Software compatibility], haɗarin girkawa [Deployment risk] da cikakken farashi [Total cost]. Dole Huang ya gamsar da masu tantance fasaha da masu yanke shawarar ƙungiya duka.

Bambancin ba cikakke ba ne. NVIDIA tana da babban tarihin wasannin masu amfani [Consumer gaming], Apple kuma ya dogara da masu haɓakawa [Developers] da ababen more rayuwa [Infrastructure]. Duk da haka manyan cibiyoyin nauyinsu sun bambanta har kiran Huang wani Jobs kawai na iya ɓoye fiye da yadda yake bayyanawa.

Salo dabam na sarrafawa

Kayayyakin Apple da aka haɗa [Integrated products] sun sha haɗawa da tsananin sarrafa ƙwarewar mai amfani na ƙarshe [End-user experience]. Dandalin NVIDIA [Platform] kuma dole ya ɗauki babbar hanyar haɗin masu haɓakawa [Developers], masu ƙera kayan aiki, masu samar da gajimare [Cloud providers] da al'ummomin bincike. Nasararsa ta dogara da wasu suna gina abubuwa da yawa da NVIDIA ba ita take ƙirawa ba.

Ba daidai ba ne a mai da wannan kwatancin rufaffe da buɗaɗɗe [Closed versus open] kawai. NVIDIA tana da manhaja ta mallaka [Proprietary software] da hanyoyin kayan aiki [Hardware interfaces] tare da ayyuka buɗaɗɗu [Open projects]. Apple ya ba da babbar damar haɓakawar wasu [Third-party development] yayin riƙe muhimman wuraren iko. Kwatancin da ya fi amfani shi ne inda kowace kasuwanci take sanya iyakarta da dalili.

Dangantaka dabam da kamfanin da aka kafa

Jobs ya bar Apple a 1985 kuma ya dawo bayan sayen NeXT a shekarun 1990. Sana'arsa ta haɗa muhimmin aiki a wajen Apple, ciki har da Pixar. Huang ya jagoranci NVIDIA ba tare da katsewa ba tun kafawarta. [24, 25]

Ci gaba ba tare da katsewa ba na iya kiyaye tarihin ƙungiya da alkawurran dogon lokaci. Dawowa bayan rashi na iya kawo sabon nazarin ƙungiya mai zurfi. Babu hanyar da take samar da jagoranci mafi kyau kai tsaye. Suna samar da yanayi dabam da mai kafa yake koyo, samun iko da sauya dabaru.

Ba a iya sanin hali daga gabatarwar dandali

Yana da jan hankali a bayyana Jobs mai tsanani Huang mai kirki, ko Jobs mai fasaha Huang mai nazari. Irin waɗannan kishiyoyi masu tsari ba abin dogaro ba ne. Hirarrakin jama'a, bayanan ma'aikata da tarihin rayuwa suna ba da ɓangarorin gani da lokaci, yanayi da dangantakar mai lura da mutumin suka tsara.

Hanyar da take da alhaki ita ce kwatanta halayen da aka rubuta: yadda aka yanke shawara, sauya samfur, yarda da kuskure ko motsa bayanai a cikin ƙungiya. Manyan da'awa kan halin ciki suna buƙatar shaida fiye da 'yan maganganu ko jawabi mai nasara.

Akwai kuma tarkon ɗabi'a wajen ɗaukar nasarar kasuwanci a matsayin hujjar kowace ɗabi'ar matsa lamba ta dace. Samfur zai iya yin nasara duk da mu'amala mara kyau da mutane, ba saboda ita ba. Mai son kafa kasuwanci bai kamata ya kwaikwayi labari mafi ban mamaki kan sanannen shugaba ya yi tsammanin sakamakon kasuwanci iri ɗaya ba.

Hukunci

Huang yana bayyana kansa mafi kyau kusa da Jobs idan kwatancin ya nuna bambancinsu maimakon ya mai da ɗaya wani. Lakabin Steve Jobs mai rai yana da kariya a matsayin misali da aka nuna a fili, amma ya raunana a matsayin bayanin gaskiya. Yana ruɗarwa musamman idan ya nuna cewa masana'antar fasaha tana buƙatar magaji guda na Jobs.

Ƙarshen magana mafi daidai shi ne Jensen Huang shugaba ne mai hangen nesa na musamman wanda tasirinsa yake gayyatar kwatanci da Steve Jobs. Jobs ya taimaka sauya hulɗar mutane da fasaha ta mutum. Huang ya taimaka gina dandalin ƙididdiga [Computing platform] da ƙungiyoyi da yawa suke haɓaka AI [AI] ta cikinsa. Nasarorinsu suna haɗuwa a hanyar jagoranci amma suna bambanta a kayayyaki, kasuwanni da tarihin ƙungiya.

A matsayin taken babi ko maƙala, Steve Jobs Mai Rai zai iya aiki a matsayin tambaya. Amma a ƙarshen tarihin rayuwa mai muhimmanci, Huang ya cancanci a fahimce shi da sunansa.

19 Iko, dogaro da alhaki

Babban mai samarwa ba babban banki ba ne

Wani lokaci ana kwatanta NVIDIA da babban bankin AI [AI] domin samun kayayyakin ƙididdigarta [Computing products] ya zama mai muhimmanci ga dabaru. Misalin yana nuna dogaro da tasiri, amma ba bayanin zahiri ne na yadda masana'antar take aiki ba.

Babban banki yana da nauyin jama'a da ikon kuɗi da NVIDIA ba ta da shi. NVIDIA kamfani ne na kasuwanci da yake aiki cikin abokan ciniki, masu gasa, masu samarwa da gwamnatoci. Ba ta fitar da kuɗin AI [AI], ba ta tabbatar da wadatar kuɗin [Liquidity] masana'antar kuma ba ta da ƙarfin da ba shi da iyaka na daidaita ta.

Babban cikas a NVIDIA zai iya haifar da sakamako masu yawa ga abokan ciniki da kasuwa mafi faɗi. Amma ba ya nufin kowane kamfanin AI [AI] zai rushe kai tsaye. Kayan da suke akwai za su ci gaba da wanzuwa, sauran masu samarwa za su kasance, ƙungiyoyi kuma za su amsa ta hanyoyi dabam. Girma da lokacin kowace illa za su dogara da irin cikas ɗin.

Wannan bambancin yana da amfani domin misali mai ban mamaki zai iya zama tsarin tunani na ƙarya [False model] wajen yanke shawara. Kamfanin da yake tsara ababen more rayuwa [Infrastructure] ya kamata ya tantance haɗarin dogaro da wuri guda [Concentration risk], wajibcin kwangila, ɗaukar manhaja daga wuri zuwa wani [Software portability] da zaɓin gaggawa [Contingency options]. Waɗannan tambayoyi sun fi amfani da zato na cikakken tsaro ko rushewar tsarin da ba za a kauce ba.

Sarkar samarwa da siyasar duniya

Guntu masu ci gaba [Advanced chips] sun dogara da sarkar samarwa [Supply chain] ta ƙwararru ta duniya. Ƙira [Design], ƙerawa [Fabrication], ƙwaƙwalwa [Memory], marufi [Packaging], kayan aiki da gina cibiyar bayanai [Data-centre] sun rarrabu tsakanin ƙungiyoyi da hukumomin ƙasashe. Nasarar NVIDIA ba ta kawar da dogaron. Takardunta na shekara suna bayyana haɗarin samarwa, abokan ciniki, ƙa'idoji da sarrafa fitar da kaya [Export controls]. [6]

Ƙuntata fitar da kaya na iya sauya kayayyakin da za a iya sayarwa a kasuwa. Iyakokin ƙerawa ko marufi [Packaging] na iya iyakance isarwa. Samun makamashi na iya jinkirta shigarwa [Installations] ko da kayan aiki [Hardware] suna shirye. Waɗannan ba al'amuran gefe ba ne: za su iya tsara damar kasuwanci da zaɓin dabarun kamfanin.

Abin da hakan yake nufi ga jagoranci shi ne hangen fasaha dole ya zauna tare da hukuncin cibiyoyi. Alƙawarin gina ƙarin ƙarfin ƙididdiga [Computing capacity] yana buƙatar sahihan tsare-tsaren masu samarwa, wuta, kuɗi da bin doka [Compliance]. Yana kuma buƙatar gane cewa wasu masu ruwa da tsaki za su iya tantance aikin ta tsaro [Security] ko muhalli maimakon aiki [Performance] kaɗai.

Jarabawar gudanarwa da ba a warware ba a Nuwamba 2026

A 8 Nuwamba 2026, hukumar NVIDIA ta sanar da bincike mai zaman kansa bayan rahoton cewa Huang ya zama abin nufin ƙoƙarin leƙen asirin wata ƙasa. Hukumar ta ce ba ta da dalili a lokacin na tunanin an lalata tsaron ƙasa ko asirin NVIDIA, ta bayyana amincewa da Huang da ƙungiyar jagorancinsa, ta kuma ce yana goyon bayan binciken kuma zai ba da haɗin kai. [26]

A 10 Nuwamba, daraktoci masu zaman kansu sun sanar da ɗaukar tsohon muƙaddashin lauya na Amurka Joon H. Kim da Wachtell, Lipton, Rosen & Katz su gudanar da binciken. A iyakar wannan bugu, waɗannan bayanai suna tabbatar da bincike da tsarin sa ido, ba kammalalliyar hujjar laifi ko wanke mutum ba. [27]

Batun da ya shafi jama'a shi ne gudanarwa da kare kamfani mai muhimmanci. Tarihin rayuwa mai hankali ba ya buƙatar maimaita zarge-zargen sirri marasa tabbaci don bayyana batun. Ya kamata a tantance binciken hukumar ta 'yancinsa, faɗinsa, shaidarsa da sakamakonsa na ƙarshe, ba ta zato game da abin da dole ya fito ba.

Lamarin yana kuma nuna ƙa'ida ta gabaɗaya: mai kafa mai hangen nesa har yanzu yana da alhaki. Nasarar fasaha, dukiyar masu hannun jari da yabon jama'a ba sa ɗaga shugaba sama da bincike. A lokaci guda, zargi ba hukunci ba ne. Adalci yana buƙatar kiyaye maganganun biyu duka.

20 Babi na gaba bai rubutu ba tukuna

AI na zahiri [Physical AI] da iyakar duniya ta gaske

Dandalin NVIDIA [Platforms] na jama'a suna ƙara shafar aikin mutum-mutumi [Robotics], kwaikwayo [Simulation] da AI na zahiri [Physical AI]. Omniverse yana tallafa wa tafiyar aiki [Workflows] mai alaƙa da kwaikwayo; Isaac yana shafar aikin mutum-mutumi; kayan samfuran lissafi [Model] da bayanai [Data] masu alaƙa suna son taimaka wa masu haɓakawa [Developers] su horar da tantance tsarukan da suke hulɗa da duniyar zahiri. Tarihin kayayyaki da sanarwar kamfanin na yanzu suna nuna wannan alkibla. [19, 20]

Sauyi daga samar da amsa zuwa sarrafa na'ura yana canza sakamakon kuskure. Sakin layi marar daidai na iya ruɗarwa; motsin mutum-mutumi marar daidai na iya lalata kayan aiki ko cutar da mutum. Kwaikwayo [Simulation] yana da daraja, amma ba zai iya tabbatar da hali mai tsaro a kowane muhallin gaske shi kaɗai ba.

Saboda haka damar kasuwancin tana tare da buƙatar gwaji [Testing], sa ido [Monitoring], tsaro [Security] da alhaki a fili. Samfurin lissafi [Model] mafi ƙarfi ba tsarin aiki [Operational system] mafi aminci kai tsaye ba ne. Masu ji [Sensors], masu motsawa [Actuators], matakan mutane da hanyoyin komawa madadi [Fallback mechanisms] duka suna da rawa a sakamakon.

Tsawon nacewar Huang kan gina dandali [Platforms] na iya dacewa a nan. Idan masu haɓakawa [Developers] suna buƙatar abin dogaro na kwaikwayo [Simulation], horo [Training] da sanya aiki [Deployment], mai samarwa da yake haɗa waɗannan matakai na iya zama mai daraja. Ko manhajojin [Applications] da suka fito suna bayar da amfanin da aka yi alƙawari har yanzu batun shaida ne, ba sakamakon burin dandalin ba.

Abin da 'yan kasuwa za su iya ɗauka daga rayuwar

Darasi na farko shi ne gina ƙwarewa kafin ɗaukar ƙarfin gwiwa a matsayin madadinta. Sana'ar Huang ta ginu kan ilimi da shekarun aikin masana'anta. Dabara da ƙarfin hali sun yi amfani, amma ilimin fasaha da hulɗar sana'a sun tallafa musu.

Na biyu shi ne ware matsala daga takamaiman mafita. Tsarin gine-ginen NVIDIA [Architecture] na farko zai iya zama marar daidai yayin da ƙididdiga mai hanzari [Accelerated computing] ta ci gaba da zama dama mai muhimmanci. Mai kafa da ba zai iya wannan bambanci ba yana haɗarin kare ƙira [Design] bayan abokan ciniki da ƙa'idoji [Standards] sun koma wani wuri.

Na uku shi ne sauƙaƙa wa wasu su ƙirƙiri daraja. Muhimmancin CUDA wani ɓangare yana cikin aikin da ya ba da dama a wajen kayayyakin NVIDIA. Dandali [Platforms] na iya girma idan sun rage wahalar gwaji [Experimentation] suka sa a sake amfani [Reusable] da aikin da ya yi nasara.

Na huɗu shi ne kiyaye sararin kuɗin koyo. Ra'ayi mai bege ba zai tsira har abada ba tare da kuɗi ba. Ceton Sega yana nuna yadda kamfani mai muhimmanci zai iya kusantar ɓacewa kafin darajarsa ta gaba ta bayyana. Haka kuma yana nuna taimakon wasu na iya zama mai yanke hukunci.

Na biyar shi ne sa jagoranci ya zama abin miƙawa. Kamfanin da yake dogaro da hukuncin mai kafa yana buƙatar hanyoyin haɓaka hukuncin wasu, kiyaye ilimi da ƙalubalantar shawarwari. Cibiyar da take ɗorewa nasara ce mafi girma da mutum wanda ba za a iya maye gurbinsa ba.

Rayuwar da ake fahimta ta shawarwari

Labarin Jensen Huang yana jan hankali domin yana da rashin tabbas na gaske. Yaro ya ƙetare ƙasashe. Ɗalibi ya zama injiniya. Injiniya ya zama mai kafa kamfani. Mai kafar ya jagoranci kamfani ta zato da suka gaza, nasarorin kasuwanci da dogon jarin dandalin rubuta shiri [Programming platform] da ya taimaka sanya NVIDIA a tsakiyar AI [AI].

Za a iya yaba wa jerin ba tare da sa ya zama wanda dole sai ya faru ba. Za a iya nazarinsa ba tare da mai da kowane cikas fatara ko kowace nasara hujjar hazakar mutum ɗaya ba. Darussansa mafi amfani suna fitowa daga hulɗar hukunci, shiri, abokan aiki, cibiyoyi da yanayi.

Kwatancin Steve Jobs yana taimaka bayyana wani ɓangaren nasarar. Shugabannin biyu sun ba sauyin fasaha murya a jama'a kuma sun tsara kamfanoni a kan manyan ra'ayoyin abin da za a iya ginawa. Amma gudummawar Huang ta musamman tana ƙara fitowa idan kwatancin bai kai ga maye gurbin ainihinsa ba.

Bai da muhimmanci saboda duniya ta sami wani Steve Jobs ba. Yana da muhimmanci domin aikin da ya taimaka tsarawa ya canza abin da wasu da yawa za su iya ƙididdigawa [Compute], bincika da ginawa. Tantance gudummawar a ƙarshe zai dogara da abubuwan da suke gaba: darajar manhajojin AI [AI applications], juriyar kamfanin, halin shugabancinsa da ikon fasaharsa ta yi wa manufofin da suka wuce bunƙasarsa hidima.

Jadawalin abubuwan da suka faru

Ƙamus na fasaha

Ƙididdiga mai hanzari [Accelerated computing]. Amfani da na'urorin sarrafawa na musamman [Specialised processors] tare da ƙididdiga ta gama-gari [General-purpose computing] don yin nauyin aiki [Workloads] masu dacewa cikin inganci.

Samfurin lissafi na AI [AI model]. Tsarin ƙididdiga [Computational system] wanda ma'aunansa [Parameters] da tsarinsa [Structure] ake amfani da su don samar da hasashe [Predictions], rarrabuwa [Classifications] ko sakamako da aka samar [Generated outputs].

Da'ira ta musamman [ASIC]. Da'ira da aka haɗa domin takamaiman amfani [Application-specific integrated circuit] da aka ƙera don wani rukunin aiki.

Na'urar sarrafawa ta tsakiya [CPU]. Na'urar sarrafawa ta tsakiya [Central processing unit]. Na'urar sarrafawa ta gama-gari [General-purpose processor] da ta dace da ayyuka masu yawa da aikin sarrafa umarni [Control tasks].

CUDA [CUDA]. Dandalin [Platform] NVIDIA da samfurin rubuta shiri [Programming model] domin ƙididdiga mai hanzari [Accelerated computing] a GPU na NVIDIA masu goyon baya [GPUs].

Cibiyar bayanai [Data centre]. Wurin kayan ƙididdiga [Computing], adanawa [Storage] da hanyoyin sadarwa [Networking], tare da wuta, sanyaya [Cooling] da tsarukan gudanarwa [Operational systems] da ake buƙata don tafiyar da su.

Zurfin koyo [Deep learning]. Koyon na'ura [Machine learning] ta amfani da hanyoyin jijiyoyi [Neural networks] masu matakan sarrafawa [Processing layers] da yawa.

Tagwayen dijital [Digital twin]. Wakilcin dijital na tsarin zahiri [Physical system] don nazari [Analysis], sa ido [Monitoring] ko kwaikwayo [Simulation]. Amfaninsa ya dogara da yadda yake wakiltar halin gaske mai dacewa daidai.

Na'urar sarrafa zane [GPU]. Na'urar sarrafa zane [Graphics processing unit]. Na'urar sarrafawa [Processor] da aka ƙera don aiki da yawa a layi ɗaya [Highly parallel work], ana amfani da ita a zane [Graphics] da manhajojin ƙididdigar lambobi [Numerical computing applications] da yawa.

Samar da amsa [Inference]. Tafiyar da samfurin lissafi da aka horar [Trained model] don samar da sakamako [Output] daga sabon abin shigarwa [Input].

Haɗin sassa [Interconnect]. Hanyar sadarwa [Communication link] tsakanin sassa [Components] ko tsaruka [Systems]. Ƙarfin wuce bayanansa [Bandwidth] da jinkirinsa [Latency] na iya shafar aikin ƙididdiga a layi ɗaya [Parallel computing performance] sosai.

Babban samfurin harshe [Large language model]. Samfurin lissafi [Model] da aka horar a kan rubutu da bayanai [Data] masu alaƙa da yawa domin sarrafa ko samar da harshe. Iya magana cikin santsi ba ya tabbatar da gaskiyar bayanai.

Ƙididdiga a layi ɗaya [Parallel computing]. Raba aiki zuwa ayyukan da za a iya yi lokaci guda [Concurrently].

Dandali [Platform]. Tarin fasaha da hanyoyin mu'amala [Interfaces] da wasu za su iya gina manhajoji [Applications] ko hidimomi [Services] a kansu.

Tensor Core [Tensor Core]. Sunan NVIDIA ga sassan kayan aiki na musamman [Specialised hardware units] da suke hanzarta zaɓaɓɓun ayyukan matrix [Matrix operations] masu muhimmanci ga AI [AI] da sauran nauyin aikin lissafi [Numerical workloads].

Horo [Training]. Tsarin daidaita ma'aunan samfurin lissafi [Model parameters] ta amfani da bayanai [Data] da manufar ingantawa [Optimisation objective].

Transformer [Transformer]. Tsarin gine-ginen hanyar jijiyoyi [Neural-network architecture] bisa hanyoyin kulawa [Attention mechanisms], da ake amfani da shi sosai a harshe da sauran samfuran lissafi [Models].

Karanta shaidar

Manazarta masu lamba suna nuna tushen da'awar gaskiya. Tattaunawar jagoranci, dabaru da kwatanci da take kewaye nazari ce maimakon kalaman mutanen da aka ɗauko. Tarihin cibiyoyi yana tabbatar da digiri, kyaututtuka da mukaman hukuma. Takardun doka suna tabbatar da bayanan da kamfanin ya bayar. Hirarraki suna kiyaye tunawa da ra'ayoyin masu magana. Sanarwar kayayyaki tana tabbatar da abin da aka sanar, tare da kiyaye ainihin sharuɗɗan da'awar aiki [Performance claims].

Kusan gazawa ba daidai yake da fatarar hukuma ba. Sanarwar samfur ba daidai take da kammala girka shi ga abokin ciniki [Customer deployment] ba. Aikin ɗan iyali ba naɗin magajin shugabanci ba ne. Bincike ba sakamako ba ne. Waɗannan bambance-bambance suna da muhimmanci don fahimtar wannan rayuwa daidai.

Wikipedia da bayanan iyali da sana'a da aka bayar sun taimaka gano batutuwan bincike. Jerin manazarta yana ba masu karatu tushen jama'a da aka yi amfani da su wajen labarin. Kwanan watan wannan bugu iyakar iliminsa ne, ba hasashen cewa abubuwan da za su faru daga baya za su tabbatar da sakamakonsa ba.

Tushe da ƙarin karatu

An ba manazarta lambobi cikin rubutu. Hanyoyin haɗi suna kaiwa ga mai wallafa tushen. Sai idan an bayyana ranar wallafa a ƙasa, an duba manazartar a 11 Nuwamba 2026.

1. Jensen Huang

Wikipedia. Living reference page consulted 11 November 2026; used as an orientation source, not sole authority for disputed details.

Read source

2. Only In America Jensen Huang on Vision Risk and the GPU

Hoover Institution. Interview with Condoleezza Rice; first-person recollections of family, migration, education and work.

Read source

3. Former Student Gift Paves the Way for New Facility

Oneida Baptist Institute. Institutional account of boarding, elementary-school attendance and the Huang Hall gift.

Read source

4. Jen Hsun Huang Engineering Hall of Fame 2013

Oregon State University. Degree dates, early education, table tennis and meeting Lori Mills.

Read source

5. Jensen Huang

Stanford University School of Engineering. Institutional biography confirming the 1992 electrical-engineering master’s degree.

Read source

6. NVIDIA Form 10 K for the fiscal year ended 26 January 2025

US Securities and Exchange Commission. Executive employment chronology and company risk disclosures.

Read source

7. Our History Innovations Over the Years

NVIDIA. Corporate chronology; product claims are treated as company descriptions.

Read source

8. Nvidia An Overnight Success Story 30 Years in the Making

Sequoia Capital. Crucible Moments interview with Huang, Mark Stevens and others; retrospective survival and platform accounts.

Read source

9. NVIDIA CEO Tells NTU Grads to Run Not Walk But Be Prepared to Stumble

NVIDIA Blog. 26 May 2023; company account of Huang’s commencement address and three strategic setbacks.

Read source

10. Nvidia nearly went out of business in 1996 trying to make Sega’s Dreamcast GPU

Tom’s Hardware. Reporting on Sega support and Shoichiro Irimajiri’s role; financial arrangement should not be described as a personal gift.

Read source

11. CUDA Platform for Accelerated Computing

NVIDIA Developer. Technical overview of the toolkit, compiler, libraries and programming environment.

Read source

12. ImageNet Classification with Deep Convolutional Neural Networks

Alex Krizhevsky Ilya Sutskever and Geoffrey Hinton. 2012 original research paper; the AlexNet result.

Read source

13. NVIDIA Reports Results for Second Quarter of Fiscal 2009

US Securities and Exchange Commission. 12 August 2008; US$196 million product-quality charge and quarterly results.

Read source

14. NVIDIA Fiscal 2009 Form 10 K

US Securities and Exchange Commission. Year ended 25 January 2009; cash, equivalents and marketable securities position.

Read source

15. NVIDIA and SoftBank Group Announce Termination of NVIDIA Acquisition of Arm Limited

NVIDIA and SoftBank. 7 February 2022; regulatory obstacles, prepaid amount and retained licence.

Read source

16. NVIDIA Completes Acquisition of Mellanox

NVIDIA. 27 April 2020; completion and networking rationale.

Read source

17. NVIDIA Announces Financial Results for Second Quarter Fiscal 2027

NVIDIA. 26 August 2026; results for quarter ended 26 July 2026 and dated product statements.

Read source

18. A conversation with NVIDIA Jensen Huang

Stripe Sessions. 2024 interview; leadership, information flow and strategic thinking.

Read source

19. Madison Huang Author Page

NVIDIA Blog. Current public role, prior LVMH work and London Business School MBA; consulted 11 November 2026.

Read source

20. Spencer Huang Author Page

NVIDIA Technical Blog. Current robotics role and NYU technical MBA; consulted 11 November 2026.

Read source

21. 50 Million Gift by NVIDIA Founder and Spouse Helps Launch Oregon State University Research Center

Oregon State University. 14 October 2022; institutional announcement of the Huangs’ gift.

Read source

22. Designing for the future

Stanford Engineering. Institutional history of the Huang Center and US$30 million commitment.

Read source

23. Apple Reinvents the Phone with iPhone

Apple. 9 January 2007; launch announcement and description of the integrated product.

Read source

24. You have got to find what you love Jobs says

Stanford University. 12 June 2005; prepared text of Steve Jobs’s commencement address.

Read source

25. Apple Timeline

Computer History Museum. Historical chronology of Apple, including Jobs’s departure and return.

Read source

26. Statement From NVIDIA Board of Directors

NVIDIA. 8 November 2026; independent investigation, stated confidence and the board’s position at that date.

Read source

27. Statement From the Independent Members of the NVIDIA Board

NVIDIA. 10 November 2026; appointment of Joon H. Kim and Wachtell Lipton Rosen and Katz.

Read source

28. NVIDIA Launches World First Deep Learning Supercomputer

NVIDIA. 5 April 2016; DGX-1 announcement; first claims are attributed to the company.

Read source

29. NVIDIA Launches Revolutionary Volta GPU Platform

NVIDIA. 2017; Volta and Tensor Core announcement.

Read source

30. NVIDIA Announces Hopper Architecture

NVIDIA. 22 March 2022; Hopper architecture and Transformer Engine announcement.

Read source

31. NVIDIA Blackwell Platform Arrives to Power a New Era of Computing

NVIDIA. 18 March 2024; Blackwell announcement, with performance claims dependent on stated comparisons.

Read source

32. Chris Malachowsky

Computer History Museum. Professional background at HP and Sun and co-founding of NVIDIA.

Read source

33. Curtis R Priem

Rensselaer Polytechnic Institute. Graphics engineering history and service as NVIDIA CTO from 1993 to 2003.

Read source

34. Famous Graphics Chips Nvidia RIVA 128

IEEE Computer Society. Jon Peddie historical account of NV1, graphics standards and the RIVA turnaround.

Read source

35. NVIDIA Fiscal 2014 Third Quarter Form 10 Q

US Securities and Exchange Commission. Historical disclosure of the 3dfx asset-purchase agreement and closing date.

Read source

36. NVIDIA Brings Real Time Ray Tracing to Gamers with GeForce RTX

NVIDIA. 20 August 2018; Turing and GeForce RTX announcement.

Read source