|
Thursday, August 27, 2026
|
|
|
|
| |
|
Today’s top story
INDUSTRY
Nvidia nears $12.9 billion deal to acquire Hugging Face
TechCrunch
Nvidia is closing in on an acquisition of Hugging Face at roughly $12.9 billion, according to reporting from The Information and Business Insider. No agreement has been signed and the talks could still collapse. The price would nearly triple the open-source model hub's $4.5 billion 2023 valuation against roughly $150 million in annual revenue, and follows a $500 million Nvidia investment offer at about $7 billion that Hugging Face turned down earlier this year.
|
|
Why it matters
Owning the default distribution point for open-weight models would give Nvidia a software and ecosystem moat as OpenAI, Google, Amazon, and Anthropic build their own inference silicon. It also puts a neutral, widely used piece of open-source infrastructure under the control of the industry's dominant chip vendor.
|
|
|
More headlines
|
|
INDUSTRY
Anthropic signs $45 billion, six-year compute deal with Nscale
Anthropic will rent $45 billion of capacity over six years from British infrastructure firm Nscale, running Nvidia Vera Rubin systems out of a West Virginia data center starting in late 2027. It extends a buying streak that includes a $10 billion Volta deal this month, $5 billion with AMD in July, and expanded Amazon, Google, and Broadcom arrangements.
TechCrunch
|
|
INDUSTRY
Amazon triples Nvidia GPU order to 2 million chips
Announced on Nvidia's quarterly earnings call, Amazon will add 2 million Blackwell Ultra, Rubin, and Rubin Ultra GPUs plus Vera CPUs across 2027 and 2028, with no financial terms disclosed. Nvidia CFO Colette Kress said demand has exceeded expectations set five months ago, when Amazon committed to deploying over 1 million GPUs.
TechCrunch
|
|
RELEASE
Z.ai revealed as the lab behind Ox Alpha model
Chinese lab Z.ai, maker of the GLM series, confirmed it built Ox Alpha, a reasoning model for coding and sustained agentic work that launched anonymously on OpenRouter and climbed to the top of several leaderboards. Z.ai said it will release the weights, keeping a frontier-tier coding model in the open-weights tier.
TechCrunch
|
|
RELEASE
OpenAI's Jalapeño inference chip beats Blackwell on benchmark
On SemiAnalysis' InferenceX benchmark, OpenAI's Broadcom-co-designed Jalapeño chip delivered more tokens per user and more throughput per kilowatt than current state-of-the-art inference processors, including Nvidia Blackwell systems. Small-volume deployment is expected by the end of 2026 with a broader rollout in 2027.
TechCrunch
|
|
FUNDING
AI assistant startup Instinct raises $250 million Series B
Index Ventures and Benchmark co-led a $250 million round valuing the year-old company at $2.5 billion and bringing total funding to $350 million. Instinct connects to users' apps and devices and is reached by text and call; it remains in private beta and has drawn scrutiny over app permissions and privacy terms.
TechCrunch
|
|
This week in benchmarks
|
|
vs. Wednesday, August 19, 2026
|
|
Intelligence
Kimi K3 (max) fell from #7 to #8
|
|
Intelligence
GLM-5.3 (max) fell from #8 to #9
|
|
Intelligence
GPT-5.6 Sol (xhigh) fell from #9 to #10
|
|
Coding
Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) climbed from #9 to #8
|
|
Coding
Claude Opus 5 (Adaptive Reasoning, High Effort) fell from #8 to #9
|
|
Intelligence
Grok 4.6 (xhigh) entered the top 10 at #7
|
|
Read past briefs and the live benchmarks at
ai-newsletter-eta-lemon.vercel.app.
Forward this to a friend who’d appreciate the brief.
|
