[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-post-\u002Ffr\u002Fnews\u002Fgtc-2026-bilan-recap-annonces-jensen-huang-nvidia":3,"news-related-fr-\u002Ffr\u002Fnews\u002Fgtc-2026-bilan-recap-annonces-jensen-huang-nvidia":378,"hreflang-gtc-2026-recap-nvidia":398},{"doc":4,"redirect":8},{"id":5,"title":6,"author":7,"author_avatar":8,"body":9,"category":354,"date":355,"description":356,"extension":357,"image":358,"meta":359,"navigation":360,"path":361,"seo":362,"seoDescription":363,"seoTitle":364,"stem":365,"tags":366,"translationKey":376,"updatedAt":8,"__hash__":377},"news_fr\u002Ffr\u002Fnews\u002Fgtc-2026-bilan-recap-annonces-jensen-huang-nvidia.md","GTC 2026 : le bilan complet — Vera Rubin, Physical AI, modèles open-weight et 1 trillion de raisons d'y croire","Paul Lefizelier",null,{"type":10,"value":11,"toc":345},"minimark",[12,16,21,29,41,48,51,55,67,70,79,85,91,95,108,115,127,135,139,145,148,155,163,167,170,176,183,186,189,192,199,253,340,342],[13,14,15],"p",{},"La Nvidia GTC 2026 s'est terminée aujourd'hui, 19 mars — quatre jours d'annonces qui dessinent l'IA des trois prochaines années. Jensen Huang n'a pas juste présenté des GPU. Il a annoncé que Nvidia devient un lab IA, un constructeur de robots et le financier de l'open source. Voici les cinq annonces qui comptent vraiment.",[17,18,20],"h2",{"id":19},"vera-rubin-blackwell-est-déjà-dépassé","Vera Rubin : Blackwell est déjà dépassé",[13,22,23,24,28],{},"Dans l'industrie automobile, un nouveau modèle sort tous les trois ou quatre ans. Nvidia sort une nouvelle génération de GPU tous les douze à dix-huit mois. ",[25,26,27],"strong",{},"Vera Rubin NVL72"," est la preuve de cette cadence folle.",[13,30,31,32,40],{},"Annoncé lors de la keynote de Jensen Huang à la ",[33,34,39],"a",{"href":35,"rel":36,"target":38},"https:\u002F\u002Fwww.nvidia.com\u002Fgtc\u002F",[37],"nofollow","_blank","GTC 2026"," (GPU Technology Conference, la conférence annuelle de Nvidia à San Jose), Vera Rubin succède à Blackwell — sorti fin 2025. Disponibilité prévue : second semestre 2026.",[13,42,43,44,47],{},"Les chiffres : ",[25,45,46],{},"3,3 fois plus puissant que Blackwell sur l'inférence"," (la phase où le modèle IA génère des réponses, par opposition au training où il apprend). L'architecture embarque 72 GPU par rack, interconnectés via NVLink, la technologie de liaison ultra-rapide de Nvidia.",[13,49,50],{},"Le prix ? Plusieurs millions de dollars par système. Ce n'est pas un produit grand public. C'est l'infrastructure des AI factories — les datacenters entièrement dédiés à l'IA.",[17,52,54],{"id":53},"physical-ai-les-robots-arrivent-entraînés-en-simulation","Physical AI : les robots arrivent, entraînés en simulation",[13,56,57,58,61,62,66],{},"Jensen Huang l'a dit clairement : le ",[25,59,60],{},"Physical AI"," est ",[63,64,65],"em",{},"\"la prochaine grande vague après le LLM\""," (les grands modèles de langage comme ChatGPT ou Claude).",[13,68,69],{},"Le Physical AI, c'est l'IA qui contrôle des objets physiques — robots, véhicules autonomes, bras industriels. Et Nvidia veut fournir toute la plateforme d'entraînement.",[13,71,72,75,76],{},[25,73,74],{},"Isaac Sim"," est la pièce centrale. C'est un simulateur de robots : on entraîne un robot dans un environnement virtuel avant de le laisser toucher le monde réel. La métaphore de Jensen Huang : ",[63,77,78],{},"\"tester une voiture en crash test virtuel avant de construire la vraie voiture.\"",[13,80,81,84],{},[25,82,83],{},"Omniverse",", la plateforme de simulation physique de Nvidia, a été mise à jour pour supporter ces cas d'usage. Les partenariats annoncés à la GTC 2026 sont révélateurs : Boston Dynamics, Figure et Tesla Optimus. Les trois leaders de la robotique humanoïde s'appuient sur l'infra Nvidia.",[13,86,87,88],{},"Citation de Jensen Huang : ",[63,89,90],{},"\"Every robot, every autonomous vehicle, every factory will be trained in simulation before touching the real world.\"",[17,92,94],{"id":93},"_26-milliards-dans-lopen-weight-nvidia-devient-concurrent-dopenai","26 milliards dans l'open-weight : Nvidia devient concurrent d'OpenAI",[13,96,97,98,103,104,107],{},"C'est l'annonce la plus structurante à long terme. Un filing SEC (déclaration auprès du régulateur boursier américain) a confirmé le chiffre : Nvidia investit ",[33,99,102],{"href":100,"rel":101,"target":38},"https:\u002F\u002Fwww.implicator.ai\u002Fnvidia-commits-26-billion-to-open-weight-ai-ships-nemotron-3-super\u002F",[37],"26 milliards de dollars sur cinq ans"," dans le développement de ses propres modèles IA ",[25,105,106],{},"open-weight"," (des modèles dont les poids sont publiés — tout le monde peut les télécharger et les utiliser).",[13,109,110,111,114],{},"Premier résultat concret : ",[25,112,113],{},"Nemotron 3 Super",". Le modèle embarque 120 milliards de paramètres au total, dont 12 milliards actifs simultanément. Il est déjà déployé chez Perplexity, Siemens, Palantir et Cadence.",[13,116,117,118,121,122,126],{},"Le signal est massif. Nvidia ne vend plus seulement des GPU. Il construit toute la stack IA — du silicium aux modèles. Si Nvidia fabrique les GPU ",[63,119,120],{},"et"," les modèles, la question devient : qui a encore besoin d'OpenAI ou d'Anthropic pour les modèles de base ? La ",[33,123,125],{"href":124},"\u002Ffr\u002Fnews\u002Fmeta-avocado-retard-licencier-gemini-google","course aux modèles"," prend un nouveau tournant.",[13,128,129,130,134],{},"C'est aussi un signal fort pour l'investissement dans le secteur. Comme le montre ",[33,131,133],{"href":132},"\u002Ffr\u002Fnews\u002Fexplosion-funding-ai-openai-anthropic-record-2026","l'explosion du financement IA en 2026",", les capitaux se déplacent vers l'infrastructure.",[17,136,138],{"id":137},"le-deal-groq-nvidia-linférence-devient-le-nouveau-marché","Le deal Groq × Nvidia : l'inférence devient le nouveau marché",[13,140,141,144],{},[25,142,143],{},"20 milliards de dollars."," C'est le montant du deal entre Nvidia et Groq, annoncé à la GTC 2026. Groq 3, le nouveau chip d'inférence ultra-rapide, a été dévoilé sur scène.",[13,146,147],{},"Pourquoi autant d'argent sur l'inférence ? Parce que le marché bascule. L'entraînement d'un modèle IA se fait une fois. L'inférence — chaque fois qu'un utilisateur pose une question — se fait des milliards de fois par jour.",[13,149,150,151,154],{},"Jensen Huang l'a résumé : ",[63,152,153],{},"\"L'inférence sera aussi grande que le training dans les deux prochaines années.\""," Groq 3 se positionne comme concurrent d'Intel sur les processeurs de calcul dédiés à l'IA. Le pari : rendre l'inférence aussi rapide que la pensée humaine.",[13,156,157,158,162],{},"Nvidia a aussi investi dans ",[33,159,161],{"href":160},"\u002Ffr\u002Fnews\u002Fnvidia-investit-thinking-machines-lab-mira-murati-2026","Thinking Machines Lab, le nouveau lab de Mira Murati"," — un autre signal que l'écosystème IA se restructure autour de Nvidia.",[17,164,166],{"id":165},"_1-trillion-de-dollars-et-des-watts-gaspillés","1 trillion de dollars — et des watts gaspillés",[13,168,169],{},"La citation qui résume toute la GTC 2026, c'est celle-ci :",[13,171,172,175],{},[63,173,174],{},"\"There is so much power squandered in these AI factories. Every unused watt is revenue lost.\""," — Jensen Huang",[13,177,178,179,182],{},"La projection de Nvidia : le marché total des AI factories atteindra ",[25,180,181],{},"1 trillion de dollars",". Les datacenters IA sont la nouvelle infrastructure critique mondiale — plus que les routes, les câbles ou les tours de télécommunication.",[13,184,185],{},"Mais Jensen Huang pointe un problème : une part massive de cette puissance est gaspillée. Des GPU bridés pour éviter des pics électriques. Des cycles de calcul inutilisés. Des watts qui ne produisent rien.",[13,187,188],{},"C'est exactement ce que des startups comme Niv-AI tentent de résoudre — optimiser chaque watt, chaque cycle, chaque GPU dans les datacenters. L'opportunité à 1 trillion de dollars ne se capture pas en ajoutant des GPU. Elle se capture en éliminant le gaspillage.",[190,191],"hr",{},[13,193,194],{},[63,195,196],{},[25,197,198],{},"En résumé :",[200,201,202,213,225,238,245],"ul",{},[203,204,205,206,209,210,212],"li",{},"La ",[25,207,208],{},"Nvidia GTC 2026"," (16-19 mars, San Jose) a présenté ",[25,211,27],{},", le successeur de Blackwell, disponible au second semestre 2026 — 3,3x plus puissant sur l'inférence.",[203,214,215,216,218,219,221,222,224],{},"Jensen Huang a déclaré le ",[25,217,60],{}," \"prochaine grande vague\" et présenté des simulations de robots via ",[25,220,74],{}," et ",[25,223,83],{},", avec Boston Dynamics, Figure et Tesla Optimus.",[203,226,227,228,231,232,234,235,237],{},"Nvidia investit ",[25,229,230],{},"26 milliards sur 5 ans"," dans les modèles ",[25,233,106],{}," — ",[25,236,113],{}," est déjà déployé chez Perplexity, Siemens, Palantir et Cadence.",[203,239,240,241,244],{},"Le deal ",[25,242,243],{},"Groq × Nvidia à 20 milliards"," positionne l'inférence comme aussi stratégique que l'entraînement — Groq 3 a été dévoilé sur scène.",[203,246,247,248,234,250],{},"Jensen Huang estime l'opportunité totale des AI factories à ",[25,249,181],{},[63,251,252],{},"\"every unused watt is revenue lost.\"",[254,255,256,272],"table",{},[257,258,259],"thead",{},[260,261,262,266,269],"tr",{},[263,264,265],"th",{},"Annonce",[263,267,268],{},"Détail",[263,270,271],{},"Impact",[273,274,275,288,301,314,327],"tbody",{},[260,276,277,282,285],{},[278,279,280],"td",{},[25,281,27],{},[278,283,284],{},"3,3x Blackwell, H2 2026",[278,286,287],{},"Nouveau standard GPU IA",[260,289,290,295,298],{},[278,291,292],{},[25,293,294],{},"Physical AI + Omniverse",[278,296,297],{},"Robots entraînés en simulation",[278,299,300],{},"Révolution robotique",[260,302,303,308,311],{},[278,304,305],{},[25,306,307],{},"26 Md$ open-weight",[278,309,310],{},"Nemotron 3 Super déjà live",[278,312,313],{},"Nvidia devient lab IA",[260,315,316,321,324],{},[278,317,318],{},[25,319,320],{},"Deal Groq × Nvidia",[278,322,323],{},"20 Md$, Groq 3 dévoilé",[278,325,326],{},"Inférence = nouveau battleground",[260,328,329,334,337],{},[278,330,331],{},[25,332,333],{},"1 trillion opportunity",[278,335,336],{},"\"Every unused watt is revenue lost\"",[278,338,339],{},"AI factories = infra critique",[190,341],{},[13,343,344],{},"La GTC 2026 n'est pas une conférence sur les GPU. C'est une conférence sur l'élimination des ressources idle à toutes les échelles — des watts gaspillés dans les datacenters aux GPU bridés pour éviter des pics électriques, aux robots qui n'existent pas encore mais qui seront entraînés en simulation avant de toucher le monde réel. Jensen Huang voit un monde où rien n'est inactif. Chaque watt produit. Chaque GPU calcule. Chaque robot s'entraîne, même quand il n'existe pas encore physiquement. C'est exactement la philosophie Idlen.",{"title":346,"searchDepth":347,"depth":347,"links":348},"",2,[349,350,351,352,353],{"id":19,"depth":347,"text":20},{"id":53,"depth":347,"text":54},{"id":93,"depth":347,"text":94},{"id":137,"depth":347,"text":138},{"id":165,"depth":347,"text":166},"ai","2026-03-19T08:32:00","Récap complet de la Nvidia GTC 2026 : Vera Rubin NVL72, Physical AI, 26 Md$ dans l'open-weight, deal Groq et l'opportunité à 1 trillion de dollars selon Jensen Huang.","md","\u002Fgtc-2026-bilan-recap-annonces-jensen-huang-nvidia.jpg",{},true,"\u002Ffr\u002Fnews\u002Fgtc-2026-bilan-recap-annonces-jensen-huang-nvidia",{"title":6,"description":356},"Récap GTC 2026 : Vera Rubin NVL72, Physical AI, 26 Md$ open-weight, deal Groq 20 Md$ et l'opportunité 1 trillion $. Toutes les annonces Jensen Huang décryptées.","GTC 2026 : bilan complet des annonces Nvidia","fr\u002Fnews\u002Fgtc-2026-bilan-recap-annonces-jensen-huang-nvidia",[367,368,369,370,371,372,373,374,375,106],"gtc-2026","nvidia","jensen-huang","vera-rubin","physical-ai","nemotron","blackwell","gpu","datacenter","gtc-2026-recap-nvidia","UuzEp9yrT6N8xKuv9QCVRVoos8MtZ2-Po3P21mdXEkY",[379,386,392],{"title":380,"description":381,"path":382,"date":383,"image":384,"category":385},"Les cloud agents Cursor construisent, hébergent et publient des apps live","Les cloud agents Cursor peuvent scaffolder une app web, provisionner Postgres et le SSO Cursor, lancer des contrôles de sécurité, puis publier une URL partageable — sans compte d'hébergement séparé.","\u002Ffr\u002Fnews\u002Fcursor-live-apps-agents-cloud-deploiement-postgres-auth","2026-11-07T08:00:00","\u002Fimages\u002Fnews\u002Fcursor-live-apps-cloud-agents-deploy-postgres-auth.jpg","vibecoding",{"title":387,"description":388,"path":389,"date":390,"image":391,"category":354},"Google rend Gemini Omni 1.2 Flash généralement disponible, avec des vidéos de 15 secondes","Google DeepMind baisse le prix d'Omni Flash de 30 %, à 0,07 $ par seconde en 720p, passe de 10 à 15 secondes de clip, et ouvre le modèle sur l'API Gemini et Google Cloud.","\u002Ffr\u002Fnews\u002Fgoogle-gemini-omni-1-2-flash-video-15-secondes","2026-11-06T08:00:00","\u002Fimages\u002Fnews\u002Fgoogle-gemini-omni-1-2-flash-15-second-video.jpg",{"title":393,"description":394,"path":395,"date":396,"image":397,"category":354},"Anthropic aurait signé son premier accord de licence avec un éditeur, l'AP","Selon Adweek, Anthropic a conclu un accord de licence de contenus et de données avec l'Associated Press juste avant les élections de mi-mandat américaines, son premier avec un éditeur numérique. Aucune des deux entreprises ne l'a confirmé officiellement, et les conditions ne sont pas publiques.","\u002Ffr\u002Fnews\u002Fanthropic-premier-accord-licence-editeurs-associated-press","2026-11-05T12:35:00","\u002Fimages\u002Fnews\u002Fanthropic-ap-publisher-licensing-deal.jpg",{"en":399,"fr":719,"es":8},{"id":400,"title":401,"author":7,"author_avatar":8,"body":402,"category":354,"date":355,"description":710,"extension":357,"image":358,"meta":711,"navigation":360,"path":712,"seo":713,"seoDescription":714,"seoTitle":715,"stem":716,"tags":717,"translationKey":376,"updatedAt":8,"__hash__":718},"news_en\u002Fnews\u002Fgtc-2026-recap-announcements-jensen-huang-nvidia.md","GTC 2026 Recap: Vera Rubin, Physical AI, Open-Weight Models and a $1 Trillion Opportunity",{"type":10,"value":403,"toc":703},[404,407,411,417,424,431,434,438,448,451,459,464,469,473,484,490,502,510,514,520,523,530,538,542,545,549,556,559,562,564,571,620,698,700],[13,405,406],{},"Nvidia GTC 2026 wrapped up today, March 19 — four days of announcements that map the next three years of AI. Jensen Huang didn't just unveil GPUs. He announced that Nvidia is becoming an AI lab, a robot builder, and the financier of open source. Here are the five announcements that actually matter.",[17,408,410],{"id":409},"vera-rubin-blackwell-is-already-old-news","Vera Rubin: Blackwell Is Already Old News",[13,412,413,414,416],{},"In the auto industry, a new model ships every three to four years. Nvidia ships a new GPU generation every twelve to eighteen months. ",[25,415,27],{}," is proof of that pace.",[13,418,419,420,423],{},"Announced during Jensen Huang's keynote at ",[33,421,39],{"href":35,"rel":422,"target":38},[37]," (GPU Technology Conference, Nvidia's annual event in San Jose), Vera Rubin succeeds Blackwell — which shipped in late 2025. Expected availability: second half of 2026.",[13,425,426,427,430],{},"The numbers: ",[25,428,429],{},"3.3x more powerful than Blackwell on inference"," (the phase where an AI model generates responses, as opposed to training where it learns). The architecture packs 72 GPUs per rack, interconnected via NVLink, Nvidia's ultra-fast linking technology.",[13,432,433],{},"The price? Several million dollars per system. This isn't a consumer product. It's the infrastructure of AI factories — data centers entirely dedicated to AI.",[17,435,437],{"id":436},"physical-ai-robots-are-coming-trained-in-simulation","Physical AI: Robots Are Coming, Trained in Simulation",[13,439,440,441,443,444,447],{},"Jensen Huang said it plainly: ",[25,442,60],{}," is ",[63,445,446],{},"\"the next big wave after LLMs\""," (large language models like ChatGPT or Claude).",[13,449,450],{},"Physical AI is AI that controls physical objects — robots, autonomous vehicles, industrial arms. And Nvidia wants to provide the entire training platform.",[13,452,453,455,456],{},[25,454,74],{}," is the centerpiece. It's a robot simulator: you train a robot in a virtual environment before letting it touch the real world. Jensen Huang's metaphor: ",[63,457,458],{},"\"crash-testing a car virtually before building the real thing.\"",[13,460,461,463],{},[25,462,83],{},", Nvidia's physics simulation platform, was updated to support these use cases. The partnerships announced at GTC 2026 are telling: Boston Dynamics, Figure, and Tesla Optimus. The three leaders in humanoid robotics are building on Nvidia infrastructure.",[13,465,466,467],{},"Jensen Huang's quote: ",[63,468,90],{},[17,470,472],{"id":471},"_26-billion-in-open-weight-nvidia-becomes-an-openai-competitor","$26 Billion in Open-Weight: Nvidia Becomes an OpenAI Competitor",[13,474,475,476,480,481,483],{},"This is the most structurally significant announcement long-term. An SEC filing confirmed the figure: Nvidia is investing ",[33,477,479],{"href":100,"rel":478,"target":38},[37],"$26 billion over five years"," to build its own ",[25,482,106],{}," AI models (models whose weights are published — anyone can download and use them).",[13,485,486,487,489],{},"First concrete result: ",[25,488,113],{},". The model has 120 billion total parameters, with 12 billion active simultaneously. It's already deployed at Perplexity, Siemens, Palantir, and Cadence.",[13,491,492,493,496,497,501],{},"The signal is massive. Nvidia no longer just sells GPUs. It's building the entire AI stack — from silicon to models. If Nvidia makes the GPUs ",[63,494,495],{},"and"," the models, the question becomes: who still needs OpenAI or Anthropic for foundation models? The ",[33,498,500],{"href":499},"\u002Fnews\u002Fmeta-avocado-delay-license-gemini-google","model race"," just took a new turn.",[13,503,504,505,509],{},"It's also a strong signal for sector investment. As the ",[33,506,508],{"href":507},"\u002Fnews\u002Fai-funding-explosion-openai-anthropic-record-2026","AI funding explosion in 2026"," shows, capital is shifting toward infrastructure.",[17,511,513],{"id":512},"the-groq-nvidia-deal-inference-becomes-the-new-battleground","The Groq × Nvidia Deal: Inference Becomes the New Battleground",[13,515,516,519],{},[25,517,518],{},"$20 billion."," That's the size of the deal between Nvidia and Groq, announced at GTC 2026. Groq 3, the new ultra-fast inference chip, was unveiled on stage.",[13,521,522],{},"Why so much money on inference? Because the market is shifting. Training an AI model happens once. Inference — every time a user asks a question — happens billions of times a day.",[13,524,525,526,529],{},"Jensen Huang summed it up: ",[63,527,528],{},"\"Inference will be as big as training within the next two years.\""," Groq 3 positions itself as an Intel competitor on dedicated AI compute processors. The bet: making inference as fast as human thought.",[13,531,532,533,537],{},"Nvidia also invested in ",[33,534,536],{"href":535},"\u002Fnews\u002Fnvidia-invests-thinking-machines-lab-mira-murati-2026","Thinking Machines Lab, Mira Murati's new lab"," — another signal that the AI ecosystem is restructuring around Nvidia.",[17,539,541],{"id":540},"_1-trillion-and-wasted-watts","$1 Trillion — and Wasted Watts",[13,543,544],{},"The quote that sums up all of GTC 2026:",[13,546,547,175],{},[63,548,174],{},[13,550,551,552,555],{},"Nvidia's projection: the total AI factory market will reach ",[25,553,554],{},"$1 trillion",". AI data centers are the new critical global infrastructure — bigger than roads, cables, or telecom towers.",[13,557,558],{},"But Jensen Huang points to a problem: a massive share of that power is wasted. GPUs throttled to avoid electrical spikes. Unused compute cycles. Watts that produce nothing.",[13,560,561],{},"This is exactly what startups like Niv-AI are trying to solve — optimizing every watt, every cycle, every GPU inside data centers. The $1 trillion opportunity isn't captured by adding more GPUs. It's captured by eliminating waste.",[190,563],{},[13,565,566],{},[63,567,568],{},[25,569,570],{},"In summary:",[200,572,573,581,593,606,613],{},[203,574,575,577,578,580],{},[25,576,208],{}," (March 16-19, San Jose) unveiled ",[25,579,27],{},", Blackwell's successor, available H2 2026 — 3.3x more powerful on inference.",[203,582,583,584,586,587,589,590,592],{},"Jensen Huang declared ",[25,585,60],{}," the \"next big wave\" and showcased robot simulations via ",[25,588,74],{}," and ",[25,591,83],{},", with Boston Dynamics, Figure, and Tesla Optimus.",[203,594,595,596,599,600,602,603,605],{},"Nvidia is investing ",[25,597,598],{},"$26 billion over 5 years"," in ",[25,601,106],{}," models — ",[25,604,113],{}," is already deployed at Perplexity, Siemens, Palantir, and Cadence.",[203,607,608,609,612],{},"The ",[25,610,611],{},"Groq × Nvidia $20 billion deal"," positions inference as equally strategic as training — Groq 3 was unveiled on stage.",[203,614,615,616,234,618],{},"Jensen Huang estimates the total AI factory opportunity at ",[25,617,554],{},[63,619,252],{},[254,621,622,634],{},[257,623,624],{},[260,625,626,629,632],{},[263,627,628],{},"Announcement",[263,630,631],{},"Detail",[263,633,271],{},[273,635,636,648,660,673,686],{},[260,637,638,642,645],{},[278,639,640],{},[25,641,27],{},[278,643,644],{},"3.3x Blackwell, H2 2026",[278,646,647],{},"New AI GPU standard",[260,649,650,654,657],{},[278,651,652],{},[25,653,294],{},[278,655,656],{},"Robots trained in simulation",[278,658,659],{},"Robotics revolution",[260,661,662,667,670],{},[278,663,664],{},[25,665,666],{},"$26B open-weight",[278,668,669],{},"Nemotron 3 Super already live",[278,671,672],{},"Nvidia becomes an AI lab",[260,674,675,680,683],{},[278,676,677],{},[25,678,679],{},"Groq × Nvidia deal",[278,681,682],{},"$20B, Groq 3 unveiled",[278,684,685],{},"Inference = new battleground",[260,687,688,693,695],{},[278,689,690],{},[25,691,692],{},"$1 trillion opportunity",[278,694,336],{},[278,696,697],{},"AI factories = critical infra",[190,699],{},[13,701,702],{},"GTC 2026 isn't a GPU conference. It's a conference about eliminating idle resources at every scale — from wasted watts in data centers to throttled GPUs avoiding electrical spikes, to robots that don't exist yet but are already training in simulation before touching the real world. Jensen Huang sees a world where nothing sits idle. Every watt produces. Every GPU computes. Every robot trains, even when it doesn't physically exist yet. That's exactly the Idlen philosophy.",{"title":346,"searchDepth":347,"depth":347,"links":704},[705,706,707,708,709],{"id":409,"depth":347,"text":410},{"id":436,"depth":347,"text":437},{"id":471,"depth":347,"text":472},{"id":512,"depth":347,"text":513},{"id":540,"depth":347,"text":541},"Full recap of Nvidia GTC 2026: Vera Rubin NVL72, Physical AI, $26B open-weight push, Groq deal, and Jensen Huang's $1 trillion AI factory opportunity.",{},"\u002Fnews\u002Fgtc-2026-recap-announcements-jensen-huang-nvidia",{"title":401,"description":710},"GTC 2026 recap: Vera Rubin NVL72, Physical AI, $26B open-weight investment, Groq deal, and the $1 trillion AI factory opportunity. All Jensen Huang announcements.","GTC 2026 Recap: All Major Nvidia Announcements","news\u002Fgtc-2026-recap-announcements-jensen-huang-nvidia",[367,368,369,370,371,372,373,374,375,106],"4Xr63rwut6LPvbrFZrxdLBSQszTGBTtvR-cz9mblLpE",{"id":5,"title":6,"author":7,"author_avatar":8,"body":720,"category":354,"date":355,"description":356,"extension":357,"image":358,"meta":937,"navigation":360,"path":361,"seo":938,"seoDescription":363,"seoTitle":364,"stem":365,"tags":939,"translationKey":376,"updatedAt":8,"__hash__":377},{"type":10,"value":721,"toc":930},[722,724,726,730,735,739,741,743,749,751,757,761,765,767,774,778,784,788,790,794,796,800,804,806,808,812,816,818,820,822,828,862,926,928],[13,723,15],{},[17,725,20],{"id":19},[13,727,23,728,28],{},[25,729,27],{},[13,731,31,732,40],{},[33,733,39],{"href":35,"rel":734,"target":38},[37],[13,736,43,737,47],{},[25,738,46],{},[13,740,50],{},[17,742,54],{"id":53},[13,744,57,745,61,747,66],{},[25,746,60],{},[63,748,65],{},[13,750,69],{},[13,752,753,75,755],{},[25,754,74],{},[63,756,78],{},[13,758,759,84],{},[25,760,83],{},[13,762,87,763],{},[63,764,90],{},[17,766,94],{"id":93},[13,768,97,769,103,772,107],{},[33,770,102],{"href":100,"rel":771,"target":38},[37],[25,773,106],{},[13,775,110,776,114],{},[25,777,113],{},[13,779,117,780,121,782,126],{},[63,781,120],{},[33,783,125],{"href":124},[13,785,129,786,134],{},[33,787,133],{"href":132},[17,789,138],{"id":137},[13,791,792,144],{},[25,793,143],{},[13,795,147],{},[13,797,150,798,154],{},[63,799,153],{},[13,801,157,802,162],{},[33,803,161],{"href":160},[17,805,166],{"id":165},[13,807,169],{},[13,809,810,175],{},[63,811,174],{},[13,813,178,814,182],{},[25,815,181],{},[13,817,185],{},[13,819,188],{},[190,821],{},[13,823,824],{},[63,825,826],{},[25,827,198],{},[200,829,830,836,844,852,856],{},[203,831,205,832,209,834,212],{},[25,833,208],{},[25,835,27],{},[203,837,215,838,218,840,221,842,224],{},[25,839,60],{},[25,841,74],{},[25,843,83],{},[203,845,227,846,231,848,234,850,237],{},[25,847,230],{},[25,849,106],{},[25,851,113],{},[203,853,240,854,244],{},[25,855,243],{},[203,857,247,858,234,860],{},[25,859,181],{},[63,861,252],{},[254,863,864,874],{},[257,865,866],{},[260,867,868,870,872],{},[263,869,265],{},[263,871,268],{},[263,873,271],{},[273,875,876,886,896,906,916],{},[260,877,878,882,884],{},[278,879,880],{},[25,881,27],{},[278,883,284],{},[278,885,287],{},[260,887,888,892,894],{},[278,889,890],{},[25,891,294],{},[278,893,297],{},[278,895,300],{},[260,897,898,902,904],{},[278,899,900],{},[25,901,307],{},[278,903,310],{},[278,905,313],{},[260,907,908,912,914],{},[278,909,910],{},[25,911,320],{},[278,913,323],{},[278,915,326],{},[260,917,918,922,924],{},[278,919,920],{},[25,921,333],{},[278,923,336],{},[278,925,339],{},[190,927],{},[13,929,344],{},{"title":346,"searchDepth":347,"depth":347,"links":931},[932,933,934,935,936],{"id":19,"depth":347,"text":20},{"id":53,"depth":347,"text":54},{"id":93,"depth":347,"text":94},{"id":137,"depth":347,"text":138},{"id":165,"depth":347,"text":166},{},{"title":6,"description":356},[367,368,369,370,371,372,373,374,375,106]]