Kwanan nan, an sami tarin haɗe-haɗe da saye-saye a masana'antar semiconductor ta duniya, inda manyan kamfanoni kamar Qualcomm, AMD, Infineon, da NXP duk suka ɗauki mataki don hanzarta haɗakar fasaha da faɗaɗa kasuwa.
Waɗannan matakan ba wai kawai suna nuna dabarun kamfanonin na neman ƙawance mai ƙarfi da fa'idodi masu dacewa a cikin gasa mai zafi ta kasuwa ba, har ma suna nuna cewa yanayin masana'antar semiconductor na iya haifar da sabbin sauye-sauye.
Ta hanyar nazarin haɗe-haɗen da aka yi tsakanin na'urorin semiconductor na ƙasashen duniya kwanan nan, na taƙaita kalmomi huɗu masu mahimmanci: AI, MCU+, motoci, da EDA.
MCU+AI: yanayin da ba makawa
STMicroelectronics ta sami Deeplite, tana mai da hankali kan fasahar AI
A watan Afrilun wannan shekarar, STMicroelectronics (ST) ta sayi kamfanin samar da fasahar AI na Kanada Deeplite, wanda ya jawo hankalin masana'antu. Kamar yadda muka sani, babban ƙalubalen da ke fuskantar samfuran ilmantarwa masu zurfi a fannin amfani da kayayyaki na kasuwanci shine girman aikinsu, buƙatun na'urori masu sarrafawa, da kuma ƙarfin amfani da wutar lantarki. Deeplite ta magance wannan matsalar ta hanyar samar da injin software mai sarrafa kansa don inganta samfuran DNN (deep neural network), wanda ke ba AI damar yin amfani da kwamfuta mai ƙarfi akan kowace na'ura.
An kafa Deeplite a shekarar 2017, kuma an san ta da mafita ta DeepSeek ta gefen AI, tana mai da hankali kan ingantawa, ƙididdigewa da matse samfuran AI. Sabuwar na'urar ingantawa mai sarrafa AI mai suna Neutrino na iya matse manyan samfuran koyo masu zurfi zuwa kashi ɗaya bisa goma na girmansu na asali yayin da take kiyaye daidaito sama da kashi 98%. Ta hanyar manyan fasahohi guda uku - yanke nauyi (cire sigogi masu yawa), ƙididdigewa (rage buƙatun daidaiton lissafi) da kuma rarrabawa (ƙara yawan nauyin sifili), manyan samfuran AI na iya gudu da sauri, ƙanana da kuma amfani da makamashi mai kyau akan na'urorin gefe. Aikace-aikacen da a da ke buƙatar ƙarfin sarrafa girgije yanzu suna iya aiki cikin sauƙi akan na'urori na gefe kamar kyamarorin wayar hannu da na'urori masu auna firikwensin masana'antu.
Deeplite ta jawo hankali sosai a farkon zamaninta kuma Gartner, Forbes, Inside AI, da ARM AI sun nada ta a matsayin babbar mai kirkire-kirkire ta AI. Wannan saye yana da alaƙa da canjin dabarun STMicroelectronics zuwa Edge AI, wanda ya haɗa kayan aiki da software ta hanyar "hulɗa biyu". Fasahar inganta samfurin Deeplite ta haɗu sosai da jerin MCUs na STM32 na STMicroelectronics da NPUs na musamman don tallafawa gina mafita na AI daga ƙarshe zuwa ƙarshe. Misali, a cikin yanayin masana'antu masu wayo, kyamarorin da aka sanye da guntuwar STMicroelectronics na iya gano lahani kai tsaye ba tare da loda bayanai zuwa gajimare ba, kuma saurin amsawa yana ƙaruwa da sau 40.
A gefe guda kuma, Deeplite tana da ƙungiyar injiniyoyin algorithm na AI na duniya, wanda ta hanyarsa ST za ta haɗa kayan aikin haɓaka AI sama da 200 don ƙirƙirar yanayin ci gaba mai haɗin kai na "dandalin kayan aikin ɗakin karatu na samfuri". A takaice, samun Deeplite ba wai kawai ya kammala aikin wasanin gwada ilimi na ST a matakin software na AI ba, har ma yana nuna canjin masana'antar semiconductor daga "yin kwakwalwa" zuwa "yin kwakwalwa".
NXP ta sayi kamfanin NPU Kinara don sake tsara Smart Edge
A watan Fabrairun wannan shekarar, NXP ta sanar da sayen kamfanin kera guntu na Amurka mai suna Kinara akan tsabar kuɗi na dala miliyan 307. An kafa Kinara a shekarar 2013 kuma an fara sanya mata suna Core Viz, daga baya aka sake mata suna Deep Vision, sannan aka sake mata suna Kinara a shekarar 2022. Kinara's dissect NPU (gami da Ara-1 da Ara-2) tana jagorantar masana'antar a fannin aiki da ingantaccen makamashi, wanda hakan ya sa ta zama mafita mafi dacewa ga aikace-aikacen AI masu tasowa waɗanda hangen nesa, murya, motsin hannu da sauran aiwatar da AI daban-daban ke jagoranta, kuma tsarin shirye-shiryenta yana tabbatar da cewa za ta iya daidaitawa da sabbin hanyoyin AI.
NXP ta ce wannan sayen zai haɗa NPU mai zaman kanta ta Kinara tare da nata fayil ɗin software, haɗin kai da tsaro, wanda zai taimaka wajen samar da cikakken dandamalin AI mai araha daga TinyML zuwa AI mai samarwa don biyan buƙatun AI masu tasowa cikin sauri na kasuwannin masana'antu da motoci. Wannan zai taimaka wajen ƙirƙirar sabbin tsarin da ke da alaƙa da AI a fannoni na masana'antu da IoT, taimaka wa abokan ciniki su sauƙaƙa rikitarwa, hanzarta lokaci zuwa kasuwa, da haɓaka ƙwarewar fasaha a fannoni kamar motoci masu wayo, suna matsawa zuwa fannoni masu ƙima.
Edge AI: Filin Yaƙi ga Masu Kera MCU
An daɗe ana samun rashin fahimta a fannin fasahar kere-kere ta wucin gadi cewa "sikelin iko ne". Duk da cewa manyan samfura suna da kyakkyawan aiki, suna fuskantar ƙalubale a zahiri - yawan amfani da makamashinsu ya saba wa buƙatun sauƙi a gefen gefe. Masana masana'antu sun yi ta nuna iyakokin da ke tattare da manyan yanayin aikace-aikacen samfura: a gefe guda, horarwa da gudanar da manyan samfura suna buƙatar albarkatun kwamfuta masu yawa; a gefe guda kuma, muhimman fannoni don haɓaka masana'antar fasahar kere-kere su ne ainihin na'urorin kwamfuta na gefe da na'urorin ƙarewa waɗanda suka fi saurin amfani da wutar lantarki da latti.
Ba abu ne mai wahala a fahimci cewa saye-sayen da aka yi a sama ya nuna cewa babban fagen fama na MCU yana canzawa zuwa lissafin AI na gefen dama. Ana sa ran nan da shekarar 2025, kashi 75% na bayanai za a sarrafa su a gefen dama, wanda ke nuna babban yuwuwar kasuwar AI na gefen dama. Wannan yana nuna cewa buƙatar lissafin AI na gefen dama yana ƙaruwa cikin sauri, kuma MCU, a matsayin babban ɓangaren na'urorin gefen dama, za ta taka muhimmiyar rawa a wannan yanayin.
A nan gaba, MCUs ba za su ƙara iyakance ga ayyukan sarrafawa na gargajiya ba, amma za su haɗa ƙarfin tunani na AI a hankali kuma a yi amfani da su ga yanayi kamar gane hoto, sarrafa murya, da kuma kula da kayan aiki na gaba. MCUs masu ƙarfin kwamfuta na gefe za su zama muhimmin mai ɗaukar ƙarfin kwamfuta na gefe tare da ƙarancin amfani da wutar lantarki, inganci mai yawa, da amsawa nan take, suna ba da tallafi mai ƙarfi ga na'urori da tsarin wayo.
Sauran manyan masana'antun MCU suma suna saye da yin gogayya a wannan fanni, kamar sayen Reality AI na Renesas Electronics, sayen Imagimob na Sweden na Infineon, da kuma ƙaddamar da NXP na software na koyon injin eIQ da sarkar kayan aikin AI NANO.
Za a iya kammala da cewa Edge AI zai zama babban filin yaƙi ga MCUs a cikin shekaru masu zuwa.
Kayan lantarki na motoci: abin da ya fi mayar da hankali a gasar jari
Kwanan nan, haɗakar na'urorin semiconductor da saye da suka shafi aikace-aikacen motoci sun bayyana akai-akai. Baya ga ƙarfin kwamfuta, juyin halittar na'urar samar da wutar lantarki ta mota, haɗin hanyar sadarwa ta cikin mota, sauti a cikin mota da sauran fasahohi shi ma ya haifar da sake fasalin fasahar semiconductor, wanda ya sa kamfanoni masu alaƙa da shi su ƙara tsarin fasahar su ta hanyar haɗakar na'urori da saye.
Masana'antar semiconductor masana'antu ce ta yau da kullun da ke buƙatar fasaha da jari mai yawa. Idan aka yi la'akari da shekarun da suka gabata, haɗin kai da haɗewa sun zama wani yanayi na ci gaban masana'antar.
Manyan kamfanonin AI galibi suna yin sayayya don inganta tsarin fasaharsu da kuma gina cikakken fa'idar "tsarin guntu + tsarin + yanayin ƙasa". Masana'antun MCU na yau da kullun suna canzawa zuwa ga AI a hankali, suna ƙoƙarin kama kasuwar tashar mai wayo tare da ƙarancin amfani da wutar lantarki da sassauci mai yawa. A fagen kera motoci, kwamfuta a cikin abin hawa, tuƙi mai cin gashin kansa da haɗin bayanai sun zama manyan fannoni na gasa ta jari. A lokaci guda, masana'antar EDA tana canzawa daga samar da kayan aiki zuwa gina yanayin ƙasa. Manyan kamfanoni suna haɗa hanyoyin IP da ƙira, kuma suna gina rinjayen kasuwa ta hanyar tsarin "daidaitaccen kayan aiki-gine".
A cikin wannan yanayi na haɗewa da saye-saye, haɗin gwiwar fasaha, faɗaɗa kasuwa da kuma rinjayen yanayin halittu sun zama babban ma'ana. Kamfanoni suna buƙatar daidaita haɗin kai na ɗan gajeren lokaci da bincike da ci gaba na dogon lokaci a tsakanin kwararar jari. Ganin cewa shingayen fasaha da yanayin jarin masana'antar semiconductor suna da matuƙar muhimmanci, wannan sauyi ba "gajeren hanya" ba ne amma "marathon" ne wanda ke buƙatar saka hannun jari na dogon lokaci.
Lokacin Saƙo: Yuni-30-2025