Skip to content
IFAI

AI compute

Industry tag545 stories

  1. IndustryReach 49

    Anthropic locks in multi-gigawatt TPU deal as revenue triples

    Anthropic just signed a major compute agreement with Google and Broadcom to secure multiple gigawatts of next-generation TPU capacity starting in 2027. The expansion comes alongside explosive customer growth: run-rate revenue has surged past $30 billion—up from roughly $9 billion at the end of 2025—while seven-figure enterprise clients doubled to over 1,000 in under two months. The vast majority of the new compute will be based in the U.S. to power frontier Claude models.

  2. IndustryReach 35

    Instacart puts Physical AI in your grocery cart

    Instacart just detailed its Physical AI push with Caper Carts, running NVIDIA Jetson chips directly on smart shopping carts. By processing basket contents, scales, and store locations in real time, the carts deliver timely nudges before checkout—driving double-digit basket size lifts for retailers.

  3. IndustryReach 43

    Bernie Sanders and AOC push to pause all new AI data centers

    Bernie Sanders and Alexandria Ocasio-Cortez just introduced a bill that would freeze construction on new AI data centers exceeding 20 megawatts peak power. The nationwide ban would last until Congress passes sweeping federal rules covering labor, environment, and civil rights. It is Washington's most aggressive swing at AI yet, though fierce tech lobbying makes passage an uphill battle.

  4. IndustryReach 53

    Meta and Arm team up on custom AI CPUs

    Meta and Arm just unveiled a partnership to co-develop multiple generations of custom data center CPUs, starting with the new Arm AGI CPU. The chip is engineered to pack more compute into tight spaces while running alongside Meta's custom MTIA silicon. Meta will also release its board and rack designs through the Open Compute Project later this year.

  5. ResearchReach 36

    Google squeezes LLM memory with TurboQuant

    Google Research just unveiled TurboQuant, a new quantization algorithm that shrinks key-value caches down to 3 bits with zero accuracy loss. By ditching the traditional memory overhead of quantization constants, it cuts KV memory by at least 6x and runs up to 8x faster on H100 GPUs. Why it matters: running massive long-context models just got substantially cheaper and faster.

  6. IndustryReach 60

    OpenAI talks fusion power with Helion, targeting 5 GW by 2030

    OpenAI is in talks to purchase electricity from fusion startup Helion Energy, aiming for 5 gigawatts by 2030 and scaling up to 50 GW by 2035. OpenAI CEO Sam Altman has stepped down as Helion's board chair to avoid conflicts of interest. The deal would secure critical clean energy for AI, though commercial fusion is still unproven.

  7. IndustryReach 43

    Tech companies are putting AI tokens in your paycheck

    Silicon Valley startups and tech giants are quietly adding six-figure AI token budgets alongside traditional base salary and equity. Nvidia CEO Jensen Huang recently suggested engineers should receive roughly half their base salary again in compute tokens. Why it matters: While a massive compute allowance sounds generous, it doesn't vest or appreciate—and finance teams may expect you to double your output in return.

  8. IndustryReach 58

    Elon Musk's $25B 'Terafab' chip mega-project

    Elon Musk just unveiled Terafab, a new Texas chip manufacturing plant operated by Tesla, SpaceX, and xAI that targets 50x current global compute output. The vertical facility plans to fabricate chips for cars, Optimus robots, and solar-powered orbital satellites. Why it matters: Building a cutting-edge fab from scratch is an enormous gamble, but orbital AI compute might be Musk's boldest bet yet.

  9. IndustryReach 39

    Bezos joins the race to put AI data centers in orbit

    Blue Origin just filed FCC plans for Project Sunrise, aiming to launch up to 51,600 sun-synchronous satellites to host orbital data centers powered by non-stop solar energy. It mirrors SpaceX's massive satellite push to bypass terrestrial grid limits, though critics like Sam Altman have already labeled space servers as pure fantasy.

  10. Generative mediaReach 38

    Runway and NVIDIA show off sub-100ms real-time video

    Runway and NVIDIA just previewed a real-time HD video model running on Vera Rubin at GTC, pushing time-to-first-frame latency under 100 milliseconds. The speed is aimed at powering Runway's General World Model, GWM-1, pointing toward truly instant and interactive AI simulation, though it is currently strictly a research preview.

  11. IndustryReach 39

    Nvidia brings RTX streaming directly to Apple Vision Pro

    Nvidia just natively integrated its CloudXR 6.0 streaming suite with Apple Vision Pro. Using visionOS dynamic foveated streaming, users can stream photorealistic 4K models directly from local or cloud RTX GPUs without dumbing down heavy 3D datasets. The Swift framework is available now, letting teams review massive industrial digital twins untethered.

  12. Open sourceReach 44

    Mamba-3 takes aim at Transformers with lower latency and 4% gains

    Researchers Albert Gu and Tri Dao just released Mamba-3, an open-source state space model architecture under the Apache 2.0 license designed to outpace Transformers. Its MIMO variant hits a 57.6% average benchmark accuracy—a roughly 4% relative bump over Transformers—while matching Mamba-2's quality using half the state size. By introducing complex-valued states and cutting GPU idle time, it targets reasoning gaps and slashes inference latency.

  13. IndustryReach 41

    Nvidia unveils Groq 3 LPU for multi-agent workloads

    Nvidia just introduced the Groq 3 LPU, its first inference chip born out of a $20 billion technology deal with startup Groq. Built into Groq 3 LPX racks alongside Vera Rubin GPUs, it delivers up to 1,500 tokens per second and 35 times higher throughput per megawatt to keep multi-agent systems talking without delay.

  14. IndustryReach 46

    Nvidia launches space computing to bring AI into orbit

    Nvidia just launched Space Computing to power orbital data centers and autonomous satellites. The lineup features the Space-1 Vera Rubin Module, which delivers up to 25x more space-based inferencing compute than the H100, allowing LLMs to run directly off-planet. Partners like Planet and Axiom Space are already onboard to process data at the source.

  15. Generative mediaReach 47

    Nvidia's GPT moment for gaming graphics

    Nvidia just introduced DLSS 5 at GTC, calling it the company's biggest graphics breakthrough since real-time ray tracing in 2018. DLSS 5 introduces a real-time neural rendering model that infuses scenes with photoreal lighting and materials at up to 4K resolution. Why it matters: DLSS is moving beyond upscaling and frame generation into generative AI that turns 3D frames into cinematic visuals. It arrives this fall.

  16. Open sourceReach 55

    NVIDIA forms an open-model dream team for Nemotron 4

    NVIDIA just launched the Nemotron Coalition, teaming up with Mistral AI, Cursor, Perplexity, LangChain, and other top AI labs to build open frontier foundation models. Their first milestone is a joint base model co-developed with Mistral AI on DGX Cloud, which will be open-sourced to underpin the upcoming Nemotron 4 model family.

  17. IndustryReach 51

    Meta reportedly plans 20%+ layoffs as AI costs surge

    Meta is reportedly planning sweeping layoffs affecting 20% or more of its nearly 79,000 employees to offset soaring AI infrastructure expenses. Executives have told leaders to begin planning cuts, though the timeline and final numbers remain unfixed. Why it matters: Tech giants are facing massive AI capital expenditures, and executives increasingly expect AI-assisted workflows to justify running far leaner organizations.

  18. IndustryReach 32

    Microsoft swoops in on Texas AI data center

    Microsoft is in talks to lease a massive AI data center site in Abilene, Texas, after Oracle walked away from the deal. With Big Tech pouring hundreds of billions into compute, securing powered land has turned into a cutthroat land grab—and Microsoft isn't hesitating to pick up what rivals leave behind.

  19. AI toolsReach 54

    Elon Musk unveils Macrohard to automate software companies

    Elon Musk just introduced Macrohard, a joint Tesla and xAI project designed to autonomously control computers and emulate entire software companies. The setup pairs Grok with a Tesla agent, processing screen video and keystrokes on Tesla AI4 chips alongside Nvidia clusters. It marks Musk's most aggressive bet yet on software automation.

  20. IndustryReach 50

    Meta is rolling out four new AI chips in just two years

    Meta just detailed an aggressive roadmap for its custom MTIA silicon, planning to launch four new chip generations over the next two years. MTIA 300 is already in production for recommendation training, while the 400, 450, and 500 series will focus heavily on GenAI inference through 2027. By taking an inference-first approach and a modular design that drops into existing racks, Meta aims to ship new chips roughly every six months.