🏙️ I-OpenAI izokwenza iLondon ibe yindawo yayo enkulu yocwaningo ngaphandle kwe-US ↗
I-OpenAI ithi iLondon ikhushulelwa esikhundleni sayo esikhulu kunazo zonke socwaningo ngaphandle kwe-US - uphawu oluqondile lokuthi ifuna izimpande ezijulile echibini lamakhono lase-UK (futhi yebo, khona kanye emnyango we-DeepMind).
Imininingwane isalokhu iyimfihlo - azikho izibalo ezinkulu zokubalwa kwabasebenzi noma ukusetshenziswa kwemali - kodwa ithoni ifundeka ngokuthi "ukwanda okukhulu," hhayi "ihhovisi lesathelayithi elihle."
🧠 I-Meta isayina isivumelwano sezigidigidi zamaRandi sokuqasha ama-chip e-Google AI, kubika i-The Information ↗
Kubikwa ukuthi iMeta iqasha ama-TPU e-Google ngesivumelwano seminyaka eminingi, esibiza izigidigidi zamaRandi - okuyinto ewusizo futhi ehlekisayo kancane, uma ubheka indlela lezi zinkampani ezivame ukuziphatha ngayo njengezitha ezifungelwe.
Kuyingxenye yalowo mzamo omkhulu wokwehluka ekuthembeleni kweNvidia - hhayi ukulahla iNvidia, kodwa nje... ukuvikela, ngokuzwakalayo.
🤖 I-Google ithatha izintambo zokulawula iphrojekthi ye-"Android of robotics" ukuze ifune i-AI ebonakalayo ↗
I-Google idonsela i-Intrinsic emuva enkampanini eyinhloko, ngokuyisisekelo ithi: i-robotics ayiseyona into eseceleni, indaba yayo eyinhloko. I-Intrinsic yabekwa njengeplatifomu "ye-Android yama-robotics" - isendlalelo sesofthiwe sokwenza ukwakha izinhlelo zokusebenza zamarobhothi kungabi yinto embi eyenziwe ngokwezifiso.
Manje kuzohlala eduze ne-DeepMind ne-Gemini, okufanele kusheshise izinto - noma okungenani kwenze ishadi le-org lingaguquki kangako (noma kubonakala kanjalo).
🛑 I-Anthropic yenqaba imigomo emisha yePentagon, imile ngokuqinile ezikhalini ezibulalayo ezizimele kanye nokuqapha okukhulu ↗
I-Anthropic iphikise obala imigomo emisha yenkontileka yePentagon, idweba imigqa eqinile mayelana nokuqapha okukhulu kanye nezikhali ezibulalayo ezizimele ngokuphelele. Ingenye yalezo zindlela "esisekela ukuzivikela, kodwa hhayi kanjalo" - ezisekelwe emithethweni, futhi futhi ukugembula okukhulu kwebhizinisi.
Ukungezwani lapha akucacile: uhulumeni ufuna ukufinyelela okubanzi, i-Anthropic ifuna izivikelo, futhi zombili izinhlangothi zibonakala zikulungele ukudlala. Kufana nokubuka abashayeli ababili bephikisana ngokuthi ubani ozosungula isiteringi.
🧩 I-Broadcom ilindele ukuthengisa ama-chip ayi-1 million e-3D stacked ngo-2027 ↗
I-Broadcom ibheja kakhulu ngama-chip ahlanganiswe nge-3D - ukufaka ama-stacking kuyasebenza ukuze kuthuthukiswe ukunyakaza kwedatha futhi kuncishiswe ukudonswa kwamandla, okuyinto ebaluleke kakhulu ekubalweni kwe-AI. Le nkampani ithi i-demand curve iphakeme ngokwanele ukuthi ilindele okungenani isigidi salezi zingxenye ezithengisiwe ngaphakathi kwefasitela layo lokubikezela.
Okuthakazelisa kancane ukuthi lokhu kubeka kanjani i-Broadcom njengokungaphezu "kokudlala i-networking kanye ne-silicon ngokwezifiso - ingena emzileni "wokwenza ingqalasizinda ye-AI engathi sína".
Imibuzo Evame Ukubuzwa
Kusho ukuthini ukuthi i-OpenAI yenza iLondon ibe yindawo yayo enkulu yokucwaninga ngaphandle kwe-US?
Kubonisa ukuthi i-OpenAI ihlose ukuba khona okujulile, kwesikhathi eside e-UK, hhayi indawo encane yokunyathelisa i-satellite. Ukubeka iLondon njengesizinda sayo esiyinhloko socwaningo esingeyona i-US kusho ukuqoqwa okunenjongo komsebenzi wocwaningo kanye nobuholi lapho. Lesi sihloko siphawula ukuthi inani labantu abasebenzisa imali kanye nokusetshenziswa kwemali akuzange kudalulwe, kodwa inhloso ifundeka njengokwanda okukhulu.
Ingabe ukwanda kwesikhungo sokucwaninga saseLondon i-OpenAI kumayelana nokuqasha nje, noma kukhona okukhulu?
Kusukela kulokho okwabiwe, kubhekwe njengokungaphezu kokuqashwa okuqhubekayo. Lesi simemezelo sibeka iLondon njengesizinda esikhulu socwaningo, sikhomba ekutshalweni kwezimali okubanzi emaqenjini, kumaphrojekthi, kanye nokuba khona kwezikhungo. Njengoba kungekho zinombolo ezithile ezinikeziwe, into ephephe kakhulu okufanele uyithathe ukuqondiswa: I-OpenAI ibeka phambili ithalente lase-UK kanye nokusondela ezindaweni zemvelo ezinkulu ze-AI, kufaka phakathi abancintisana nabo abaseduze.
Kungani iMeta ingaqasha ama-TPU e-Google esikhundleni sokuthembela kuma-GPU e-Nvidia kuphela?
Umbiko uwubeka njengomzamo wokuvikela kunokuba kube ukuphumula okuhlanzekile okuvela kuNvidia. Ukuqasha ama-TPU e-Google kungahlukanisa ukunikezwa kokubala futhi kunciphise ukuthembela kumthengisi oyedwa ngesikhathi sokwakhiwa kwe-AI enzima. Ezindleleni eziningi, ukuxuba ama-accelerator kuyindlela esebenzayo yokulinganisela izindleko, ukutholakala, kanye nokuhlela amandla, ikakhulukazi ngenkathi isidingo sama-chip aphezulu sisalokhu siphezulu.
Iyini i-Intrinsic, futhi kungani i-Google yadonsela iphrojekthi "ye-Android yamarobhothi" enkampanini eyinhloko?
I-Intrinsic yabekwa njengesendlalelo seplatifomu esenzelwe ukwenza isofthiwe yamarobhothi ingakhethi, kusetshenziswa indlela yesitayela "se-Android" yezinhlelo zokusebenza zamarobhothi. I-Google iyibuyisela enhlanganweni eyinhloko iphakamisa ukuthi amarobhothi ayashintsha kusuka kuphrojekthi eseceleni yokuhlola iye endaweni ebalulekile. Lesi sinyathelo siphinde siyibeke eduze neminye imizamo emikhulu yangaphakathi ye-AI, engaqinisa ukuxhumanisa futhi isheshise ukukhiqizwa.
Yini i-Anthropic eyenqabayo emigomeni yenkontileka entsha yePentagon?
Lesi sihloko sithi i-Anthropic yawachitha amagama eyayiwabheka njengokuvumela ukuqapha okukhulu kanye nezikhali ezibulalayo ezizimele ngokuphelele. Yakubeka lokhu njengokubeka izivikelo eziqinile ngenkathi isasebenza ngokuzivikela ngezinye izindlela. Ukubaluleka kulele ekungezwani phakathi kwezicelo zikahulumeni zokufinyelela okubanzi noma ukuguquguquka kanye nesifiso senkampani ye-AI sokunciphisa amanye amacala okusetshenziswa ayingozi kakhulu.
Kungani ama-chip ane-3D-stacked eyinto enkulu kwi-AI, futhi yini ebikezelwa yi-Broadcom?
Ukufakwa kwe-3D stacking kungathuthukisa ukunyakaza kwedatha kanye nokusebenza kahle kwamandla ngokufaka ama-stacking dies, okubalulekile emisebenzini ye-AI evame ukuvinjelwa yi-bandwidth kanye namandla. I-Broadcom ibheja ukuthi isidingo sizoba namandla ngokwanele kangangokuthi ilindele ukuthengisa okungenani isigidi esisodwa salezi ziyunithi ze-chip ze-3D-stacked ngo-2027. Lokhu kusho ukuthi ukuncintisana okusheshayo engqalasizinda ye-AI kungaphezu kwezindima zendabuko.