Federal prosecutors just charged four people with running a front company to illegally smuggle Nvidia H100 and H200 GPUs into China. The group allegedly bypassed US export controls using falsified shipping manifests and wire transfers from Chinese bank accounts. Why it matters: It offers a rare glimpse into how restricted frontier hardware still finds its way past export bans.
Nvidia and startup Starcloud just launched an 80GB H100 GPU into orbit aboard a SpaceX Falcon 9 rocket to see if data centers can truly work in space. The chip is 100 times more powerful than any computer ever flown and will process radar images and run a Google LLM in orbit. If successful, orbital computing could bypass Earth's land, power, and water bottlenecks entirely.
Anthropic just submitted formal recommendations on the Commerce Department's Diffusion Rule, urging the US to double down on AI chip export controls. The company is pushing to lower the current 1,700 H100 exemption threshold for Tier 2 nations and boost enforcement, arguing any delay will just let overseas rivals stockpile more hardware.
Huawei is reportedly developing a new Ascend 910D AI chip designed to rival Nvidia's workhorse H100, and is actively seeking domestic partners for testing. Following fresh U.S. chip export curbs, a viable 910D could plug a critical computing gap in China's AI buildout.
Elon Musk just fired up xAI's Memphis Supercluster, packing 100,000 liquid-cooled Nvidia H100 GPUs tied together on a single RDMA fabric. Musk boasts it is the world's most powerful AI training cluster and aims to deliver the strongest AI on every metric by December, though local utility contracts for its massive 100-megawatt power appetite are still not finalized.
Intel just unveiled its Gaudi 3 AI accelerator at Vision 2024, claiming it outperforms Nvidia's dominant H100 GPU. Built on TSMC's 5nm process with 128GB of HBM2e memory, Intel says Gaudi 3 runs models like LLaMA up to 1.7x faster and is 2.3x more power-efficient. Air-cooled models roll out in Q3, followed by liquid-cooled units in Q4.
Nvidia just swept all seven categories in the latest MLPerf inference benchmarks, showcasing its new H200 chip and TensorRT-LLM software optimizations. Inference now drives roughly 40% of Nvidia's data center revenue. Intel submitted results for Gaudi2 and 5th Gen Xeon, but AMD once again chose not to submit its MI300.
Mark Zuckerberg just confirmed that Meta is actively training Llama 3, aiming for state-of-the-art coding and reasoning capabilities to pursue AGI. To power the push, Meta is stockpiling over 340,000 Nvidia H100 GPUs by the end of 2024, building a computing fortress that rivals almost any competitor in the industry.
Nvidia CEO Jensen Huang just predicted that artificial general intelligence could arrive within the next five years. Speaking at the DealBook Summit, Huang defined AGI as AI that can match normal human intelligence on standard tests, revealing that Nvidia's H-100 chips already require AI to design, though models still struggle with multi-step reasoning.
The Chan Zuckerberg Initiative just announced plans to build a computing cluster powered by over 1,000 Nvidia H100 GPUs for medical research. The setup aims to train generative AI models to simulate "virtual cells," letting researchers predict cellular reactions to infections and drugs without manual lab testing. Why it matters: High-end compute is usually monopolized by commercial giants; CZI plans to make these biological models openly available to scientists.
The U.S. government just expanded its AI chip export restrictions beyond China, requiring Nvidia and AMD to secure licenses before shipping top-tier chips like the A100 and H100 to some Middle Eastern nations. Both chipmakers say the curbs won't immediately hit their revenues, but it shows Washington is tightening its grip on global compute.
Google Cloud just unveiled its fifth-generation custom TPU chips at Cloud Next, aiming squarely at high AI computing costs. The new processors deliver 2x the training performance per dollar and 2.5x for inferencing versus the previous gen, while letting massive workloads scale across physical clusters to tens of thousands of chips.
Tesla just switched on a massive supercomputer packed with 10,000 Nvidia H100 GPUs to accelerate its full self-driving development. The cluster delivers 340 FP64 PFLOPS of compute, beating Leonardo, the world's fourth most powerful supercomputer. Alongside its custom Dojo hardware, Tesla is building an unprecedented computing moat in the auto industry.
A Cointelegraph analysis warns that the generative AI hype is showing classic bubble symptoms after raising over $25B in the first half of 2023. Startups are burning funding just to stockpile scarce Nvidia GPUs rather than building product moats. With ChatGPT traffic slipping and API wrappers laying off staff, the market may face a reality check.
Nvidia just delivered what Wall Street calls 'guidance for the ages,' projecting second-quarter revenue of $11B against analyst expectations of $7.2B. The chip giant's shares exploded 27%, adding over $200B in market value in a single day. CEO Jensen Huang noted that massive demand for generative AI chips is driving a decade-long transition to accelerated computing across global data centers.
