AMD and OpenAI just announced a massive deal to deploy 6 gigawatts of AMD GPUs for next-gen AI infrastructure. The rollout kicks off in the second half of 2026 with an initial 1-gigawatt deployment of Instinct MI450 chips. AMD is also granting OpenAI warrants for up to 160 million shares tied to delivery milestones.
OpenAI just recruited Samsung and SK into its massive Stargate infrastructure push to lock down critical memory supplies and build data centers in Korea. The partnership targets 900,000 DRAM wafer starts per month to power frontier models, alongside new regional data center builds. Why it matters: High-bandwidth memory is the ultimate AI bottleneck, and locking in the world's two largest memory giants gives OpenAI an unmatched hardware pipeline.
Musk’s xAI just filed a federal lawsuit against OpenAI, alleging the rival induced poached employees to steal trade secrets around its Grok model and Colossus data center playbook. OpenAI fired back, calling the complaint the latest chapter in Musk’s ongoing harassment.
Qualcomm just officially announced the Snapdragon 8 Elite Gen 5, its next-generation flagship processor focused heavily on personalized, on-device AI. The chip features a third-gen Oryon CPU alongside a Hexagon NPU that runs 37% faster and cuts power draw, paving the way for upcoming Android flagships.
OpenAI and SAP just teamed up to launch 'OpenAI for Germany,' a dedicated sovereign AI platform for the country's public sector planned for 2026. Powered by SAP subsidiary Delos Cloud on Microsoft Azure, the setup will scale to 4,000 GPUs to help government workers automate workflows while strictly complying with local data sovereignty laws.
Microsoft just made Windows ML generally available, letting developers deploy local ONNX models directly across AMD, Intel, Nvidia, and Qualcomm chips. Windows now automatically handles execution providers in the background, shaving hundreds of megabytes off app bundle sizes. It brings real-time, privacy-first local AI to Windows 11 24H2 PCs.
Sam Altman just laid out his vision for "Abundant Intelligence," arguing that access to AI will soon become a basic human right. To keep up with soaring demand, he envisions a system capable of adding one gigawatt of AI infrastructure every single week, with partners and financing plans set to be unveiled later this year.
NVIDIA and OpenAI just announced a landmark partnership to deploy at least 10 gigawatts of AI systems powered by millions of GPUs. Backing the rollout, NVIDIA plans to progressively invest up to $100 billion as capacity comes online. The first gigawatt will deploy in late 2026 on the upcoming Vera Rubin platform to power OpenAI's push toward superintelligence.
OpenAI is rolling out a series of compute-intensive offerings over the next few weeks, according to CEO Sam Altman. Due to heavy compute costs, some features will initially be limited to Pro subscribers, while other products will carry additional fees. Altman noted that while driving down intelligence costs remains the ultimate goal, the team wants to see what's possible when throwing massive compute at ambitious ideas today.
OpenAI just laid out plans to spend an extra $100 billion over five years on backup server rentals, on top of $350 billion already projected through 2030. That pushes its annual computing budget to an eye-watering $85 billion as it races to secure scarce infrastructure. While the company projects burning $115 billion through 2029, it is betting explosive demand will push revenue to $200 billion by 2030.
NVIDIA and Intel just announced a landmark partnership to co-develop custom data center and PC processors using NVLink. Intel will manufacture custom x86 CPUs for NVIDIA's AI platforms and build PC chips integrating RTX GPU chiplets, backed by a $5 billion equity investment from NVIDIA. The deal tightly marries the dominant x86 ecosystem with NVIDIA's AI compute engine.
OpenAI just reportedly signed a staggering $300 billion computing deal with Oracle spanning roughly five years, kicking off in 2027. The contract fuels the massive 4.5-gigawatt Project Stargate initiative, completely dwarfing OpenAI's expected $12.7B revenue this year. The mega-deal helped Oracle's future contract revenue surge past $317 billion, sending its shares soaring.
Thinking Machines AI just debunked the prevailing theory behind why LLM inference yields inconsistent results even with greedy sampling. While developers usually blame GPU floating-point concurrency and race conditions, modern forward passes are actually deterministic. The real culprit is a lack of batch invariance, meaning an individual request's output subtly shifts depending on which concurrent queries share the batch.
NVIDIA just submitted its first MLPerf Inference v5.1 results for the Blackwell Ultra architecture via the GB300 NVL72 rack-scale system. It swept performance records across newly added benchmarks like DeepSeek-R1 and Llama 3.1 405B, delivering roughly 5x higher per-GPU throughput than previous Hopper-based systems.
Apple just unveiled the iPhone Air, its thinnest phone yet at 5.6mm, powered by the flagship A19 Pro chip with custom Neural Accelerators delivering triple the peak GPU compute for on-device generative AI. Paired with iOS 26, it powers upgraded Apple Intelligence features like Live Translation and on-screen visual search. Pre-orders begin September 12.
Apple just announced the iPhone 17, packing an A19 chip designed with built-in neural accelerators in every GPU core to run on-device generative AI models. The phone pairs with iOS 26 to power upgraded Apple Intelligence capabilities like Live Translation and screen-wide visual search, starting at $799.
NVIDIA just introduced Rubin CPX, a new chip system built for massive AI inference workloads like video generation and software coding. By packing video decoding, encoding, and inference onto a single platform, it cuts multi-chip bottlenecks for tasks demanding 1M+ token contexts.
Mark Zuckerberg just promised Donald Trump at least $600 billion in U.S. AI spending through 2028, but a hot mic later caught him apologizing and admitting he "wasn't sure what number you wanted to go with." Zuckerberg quickly clarified on Threads that he wasn't sure which timeframe Trump meant, so he shared the lower estimate. It is a telling glimpse into tech's political maneuvering.
OpenAI is reportedly teaming up with Broadcom to mass-produce its own custom AI chips next year for internal workloads. Broadcom just disclosed a mystery $10 billion chip order that analysts widely pin on OpenAI. Much like Google and Amazon, the ChatGPT maker is moving in-house to slash skyrocketing infrastructure costs and loosen Nvidia's grip.
Nvidia just reported record Q2 revenue of $46.7 billion, up 56% year-over-year, alongside $26.4 billion in net income. Growth was spearheaded by data center demand, with its Blackwell chips alone driving $27 billion in sales. However, regulatory uncertainty left the company with zero sales of its China-focused H20 chip to Chinese customers.
