SpaceX closes $40 billion in debt financing to buy GPUs for AI, CEO Elon Musk says. Musk says the financing is mostly callable debt, with the chips arriving in the first half of 2027. A regulatory filing says SpaceX signed a binding commitment with an Apollo-managed entity for the GPU financing.
Cerebras says it spreads Kimi K3, an open-source 3-trillion-parameter model, across 59 of its WSE-3T wafers, each holding 44 GB of on-chip memory. One dense layer sits on each of 24 wafers and two linear-attention layers on the other 35. Cerebras claims 1,000 to 2,000 or more tokens per second.
SemiAnalysis InferenceX benchmarks show OpenAI's Jalapeño inference chip beating Nvidia's Blackwell on performance per watt across models including GPT-OSS and DeepSeek R1. Ho, OpenAI's hardware vice president, says each of the chip's 64 cores addresses its own HBM bank, and the design reached tapeout in nine months.
A LessWrong post by josefely argues the US and China can only durably slow AI through a treaty governing compute, because a unilateral pause would let China erase a roughly 10-point Epoch Capabilities Index lead. Verification would depend on satellite monitoring, on-site inspections and tamper-resistant chips that meter and license compute, the post says.
Ant Group and the vLLM community open-source Humming, a kernel library for quantized inference that supports NVIDIA GPUs from Turing onward and T-Head's Zhenwu 810E. Kernels compile just in time from shared CUDA components. The groups claim switching DeepSeek-V4.1-Flash MoE execution to Humming's W4A8 path on eight H20 GPUs raises throughput 14% to 43%.
DeepSeek-AI published DeepInfer, the inference framework it says powers its production services, under Apache 2.0. It serves DeepSeek-V4.1-Flash on Nvidia SM100 GPUs and Huawei Ascend 950 NPUs. The release omits a KV cache system and prefill/decode disaggregation, which the repository says a complete service requires.
Google released TorchTPU, an open-source version of PyTorch that runs AI apps on its Tensor Processing Units with a configuration change and no rewriting. Amazon released similar software, TorchNeuron, for its Trainium chips a day earlier. Both companies want to loosen Nvidia's hold on AI development, where PyTorch is used by most AI apps.
Meta says the first phase of Prometheus, its gigawatt-scale supercluster in New Albany, Ohio, is complete and trained its Muse foundational models, including Muse Wonder. The company claims more than 85% goodput on pre-training runs and plans to scale to roughly 1 million GPUs.
Tenstorrent raises $1.5 billion at a $6.2 billion valuation, more than double the $2.6 billion it fetched in late 2024. Trace Capital and Innovation X led the round. CEO Jim Keller told Bloomberg the chip startup stayed independent after its Galaxy servers ran DeepSeek models.
Qualcomm chief executive Cristiano Amon says AI companies are asking for phones that can run 100-billion-parameter models by 2028, running all the time. He would not name them, and cites ByteDance's Doubao phone by Nubia. Amon blames memory prices up five to six times for a 20% drop in the phone market this year.