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.
Microsoft is preparing to run parts of its Copilot AI locally on PCs to cut latency and boost privacy, according to Intel. The catch? Next-gen AI PCs will reportedly require NPUs pushing over 40 TOPS, a bar no consumer chip on the market currently clears.
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.
Stability AI just lost the core research team behind Stable Diffusion, including lead researcher Robin Rombach and two key co-authors. The high-profile departures follow a string of executive exits, mounting compute and payroll debts, and stalled fundraising efforts, putting the open-source darling under immense survival pressure.
Intel just spun out a new enterprise generative AI company called Articul8 AI with backing from data center investor DigitalBridge. Originating from an internal project with Boston Consulting Group, the platform lets clients run multimodal AI within their own security perimeters. Why it matters: It gives Intel a dedicated software wedge to drive enterprise hardware sales against Nvidia.
Stability AI is reportedly shopping itself to buyers as pressure mounts from major investors. Coatue Management recently demanded CEO Emad Mostaque step down, pointing to burn rates around $8M a month against minimal revenue. The startup has approached firms like Cohere and Jasper, though company spokespeople deny an active sale process.
Intel CEO Pat Gelsinger says AI won't stay locked in cloud data centers for long, arguing that economics, latency, and privacy will push workloads to the edge. The company is set to launch its Meteor Lake PC processors with dedicated AI silicon, leveraging its 80% x86 PC market share to convince developers to run AI locally rather than pay for costly cloud compute.
Researchers from the University of Zurich and Intel built Swift, an autonomous drone system that took down three world-champion pilots in high-speed FPV racing. Trained via reinforcement learning in simulation, Swift relied solely on onboard sensors to clock the fastest lap, beating human records by half a second. It's a breakthrough for physical AI, though human pilots still handle changing conditions like room lighting far better.
Nvidia just reported a massive $6.188B in quarterly net profit, up 843% year over year on $13.5B in revenue. The surge was driven by its data center business, which reached a record $10.32B as demand for generative AI chips exploded. Basically, Nvidia is dominating the hardware race while projecting an even bigger $16B next quarter.
Microsoft just launched its Azure ND H100 v5 virtual machines, packing eight NVIDIA H100 GPUs and Quantum-2 InfiniBand networking for massive AI workloads. The cloud giant also expanded Azure OpenAI Service to five new regions, broadening access to models like GPT-4 as enterprise adoption surges. Why it matters: Microsoft is rapidly scaling up hardware muscle to meet insatiable cloud AI demand.
Intel CEO Pat Gelsinger just told investors the company plans to build AI into every platform and product it makes. The rollout begins later this year with Meteor Lake, its first consumer chip sporting an onboard neural processor. Why it matters: As Nvidia dominates the AI cloud, Intel is betting that latency, bandwidth, and cost limits will force AI inferencing straight onto client devices and the edge.
Stability AI just rolled out Stable Diffusion 2.0, headlined by a new depth2img feature that infers 3D depth maps via the MiDaS model. By locking in geometric structure rather than just relying on text prompts and pixels, artists can now graybox a rough 3D scene and let AI explore lighting and texturing with far fewer structural distortions.
ML6 detailed how to use the open-source library Anomalib for automated visual inspection in manufacturing. Using a student-teacher setup, the model trains strictly on defect-free nominal images to pinpoint anomalies at the pixel level. Exporting to Intel's OpenVINO increased CPU inference speed by nearly 4x to 30.2 fps.
