AI Radar

Chips & infrastructure

AI chips, data centers and cloud computing power. Newest first.

Importance 2 of 5NVIDIA Blog

Nvidia AI Day Seoul highlights Korean firms using its platforms for robotics

More than 2,000 attendees gathered at COEX for Nvidia's AI Day Seoul, focused on physical AI and agents. Nvidia said Doosan Robotics, Hyundai Motor Group, LG Electronics, SK Group and startups are using its platforms to bring general-purpose robotics into real-world use.

Why it matters: It shows Korean industrial groups building robotics on Nvidia's stack, which matters for manufacturers weighing physical AI.

Importance 3 of 5The Verge AI

Rumors suggest Jensen Huang may be seeking a return as Nvidia CEO

The Verge reports rumors that former Nvidia CEO Jensen Huang might be trying to find a path back to the top job after his resignation. The situation around Nvidia's leadership continues to change quickly.

Why it matters: Leadership uncertainty at Nvidia affects chip supply relationships, startup funding and strategy for companies dependent on its hardware.

Importance 4 of 5The Verge AI

Congress moves to tighten AI chip export rules after Nvidia spy scandal

Following the scandal involving former Nvidia CEO Jensen Huang and a suspected Chinese spy, lawmakers from both parties back measures to further restrict AI chip exports. They have also demanded that Huang and other Nvidia leaders testify before Congress.

Why it matters: Tighter export rules could affect global AI chip supply and compliance for companies buying or reselling accelerators.

Only want what affects you?
Get a weekly digest filtered to your company. Free.
Filter my radar
Importance 2 of 5r/LocalLLaMA

User benchmarks 1Cat-vLLM expert caching against Strata on Tesla V100 rig

A user compared inference speeds of 1Cat-vLLM, a Tesla V100 focused fork of vLLM that recently added expert caching, with Strata. The test used a Qwen3.8-Flash-Next IQ3_XXS quant on four 16 GB V100 cards with an Epyc CPU and DDR4 memory.

Why it matters: Expert caching on older GPUs could make large mixture-of-experts models cheaper to run locally on used hardware.

Importance 2 of 5r/LocalLLaMA

Open-source tool tunes cudaMemcpy to speed MoE CPU offload prefill by about 35%

A developer released a tool that tunes CUDA memory copies used when offloading mixture-of-experts layers to the CPU. On an RTX 3090 with Qwen3.6 35B-A3B in llama.cpp, prefill rose from 345 to 466 tokens per second.

Why it matters: Cheap software tuning can noticeably improve local inference on consumer GPUs for models too large for VRAM.

Importance 2 of 5r/LocalLLaMA

Basalt 0.4.0 inference engine adds Qwen 3.8 27B, reaching up to 776 tok/s on an RTX 5090

The Basalt local inference engine version 0.4.0 adds its first dense model, Qwen 3.8 27B, with up to 776 tokens per second on structured output on an RTX 5090. Recent updates also added multi-model and router support, DeepSeek V4.1 Flash, video support and files in chats.

Why it matters: Faster local inference makes running capable open models on a single workstation more practical.

Importance 4 of 5The Decoder

WSJ: Firm owned by Jensen Huang's ex-partner brokered nearly $12B in Nvidia chip orders for China

Megaspeed, owned by Alice Huang, arranged nearly $12 billion in Nvidia chip orders for Chinese companies in 2024 and 2025, the WSJ reports. Short seller Culper Research estimates Nvidia's real China revenue share is above 20 percent, versus the reported 7 percent.

Why it matters: The report raises export control and disclosure questions that could bring tighter rules on Nvidia chip supply.

Importance 4 of 5The Decoder

Meituan releases open weights for LongCat-2.5, a 1.6T-parameter agent model trained on Chinese chips

Meituan released the weights for LongCat-2.5, a 1.6-trillion-parameter model built for AI agents and trained entirely on Chinese hardware. It scores 88.4 on Terminal Bench, just behind Claude Opus 5. The sample deployment runs on 194 Huawei Ascend chips with no mention of Nvidia GPUs.

Why it matters: A near-frontier open agent model trained without Nvidia hardware shows Chinese labs can work around export controls, widening open-weight options for enterprises.

Importance 3 of 5r/LocalLLaMA

US Commerce moves to finalize rule blocking Chinese AI firms from renting Nvidia GPUs abroad

Reports say the US Commerce Department is moving to finalize a rule that would stop Chinese AI firms from remotely accessing Nvidia GPUs through overseas cloud providers. The effort builds on the Remote Access Security Act (H.R.2683), which passed the House earlier.

Why it matters: Limits on remote GPU access could reduce compute for Chinese labs that release popular open models. They could also add compliance duties for cloud providers serving foreign customers.

Importance 2 of 5Google Developers Blog

Smaller apps and faster GPU inference with LiteRT in Google Play services

LiteRT in Google Play services now offers GPU inference powered by ML Drift, delivering about 1.4x faster performance alongside the new CompiledModel API. Switching from bundled binaries reduces app size by approximately 8 MB while enabling seamless, automatic runtime updates across billions of devices. Developers can…

Importance 2 of 5Google Developers Blog

Integrating TorchTPU into TorchTitan and Optimizing Models for TPU

TorchTPU integrates a native torch.device("tpu") backend directly into PyTorch, allowing developers to train large-scale models on Google Cloud TPUs without rewriting code in JAX. The TorchTitan TPU fork enhances this with production-grade scaling primitives like SimpleFSDP, Graph Trainer, and custom Pallas kernels…

Importance 2 of 5r/LocalLLaMA

Update: tandem.cpp — DeepSeek-V4.1 on a 64 GB PC with a 16 GB card, and AMD (ROCm) support

Quick update on tandem.cpp, my patch set for llama.cpp that runs MoE models too big for your VRAM faster. It keeps the hot experts on the GPU and lets the CPU help with the rest. Repo: https://github.com/ewhacc/tandem.cpp DeepSeek-V4.1-Flash IQ2_XS (ISTA GSQ-RCO, 139 GiB of experts) on normal PCs Decode tok/s, one…

Importance 2 of 5The Decoder

Anthropic CEO Amodei admits "there's a huge uncertainty in what's going to happen next year" days before a record IPO

Days before what could be the largest IPO in history, Anthropic CEO Dario Amodei called buying compute 'the deepest business dilemma' his company faces. Anthropic's commitments now total nearly $901 billion, and it's still far from clear whether future revenue will justify those costs. The planned IPO is largely meant…