AI Agent UpdateNvidia (NVDA) — AI News & Analysis
Every AI Agent Update story mentioning Nvidia (NVDA), with our bullish / bearish read on each.
Latest stories
Unitree's H2-X tried a side flip on stage — and lost a leg
Unitree's new security robot tried a side flip on stage and fell apart — literally. A reminder that demo-day robots and reliable robots are not the same thing.
Samsung just took the HBM crown from SK hynix — in Bernstein's estimates, at least
Bernstein's estimates put Samsung ahead of SK hynix in AI memory revenue for the first time this cycle — but these are estimates, not results.
Bank of America: data-center spending hits $1 trillion this year — and doubles by 2028
Bank of America says the buildout doubles by 2028 and nearly triples by 2030. The power grid, not demand, is the only thing that can stop it.
Unitree jumped 20% to its daily limit after a GPT-6 robot copied a drawing
A robot copied a drawing under GPT-6, and Unitree's stock hit its daily limit. Momentum, not earnings, is doing the driving — for now.
Kimi K3.1 Pro is open, cheap and fourth on the leaderboard — and deliberately worse at hacking
Kimi K3.1 Pro is cheaper, bigger-context, and fourth on the intelligence leaderboard — and its maker deliberately made it worse at hacking.
The AI funding scare just took the chip index to its worst day in four months
It wasn't demand that broke — it was the question of who pays for it.
OpenAI's $30B raise just took a $150B haircut — and it's still the biggest private valuation in history
Investors balked at $1.4T. The cut tells you where the market's real ceiling is.
Neoclouds are getting repriced like credit, not tech
CoreWeave's bonds yield 15%. Its stock just hit a 2026 low. Same company, two verdicts.
Tesla and Nvidia just became Samsung's landlords — $2.8B for chip substrates
When your customers prepay $2.8B for factory lines, that's not circular financing — that's demand.
OpenAI's models are coming to Google Cloud — yes, that Google
The fiercest rivalry in AI just became a distribution deal.
The US wants to police where AI venture dollars go — not just chips
$50,000 is the new tripwire for investing in Chinese AI.
Supermicro's new rack packs 256 Nvidia Vera CPUs for the agent era
22,500 concurrent agent sandboxes in one liquid-cooled rack.
Nvidia is prepaying its suppliers like there's no tomorrow — $279B says there isn't
Six manufacturing partners just got cash up front — substrates, packaging, testing, optics — because Jensen Huang's supply chain is now the bottleneck, not demand.
Australia's “CoreWeave” just crashed 42% — because the grid said no
NEXTDC slashed FY27 guidance by a third after a Melbourne datacenter's power upgrade got blocked. The AI buildout's binding constraint is watts, not GPUs.
DeepSeek is closing a $15B war chest — the open-weight assault is going commercial
CATL and Tencent are anchoring a round that doubled its target, one day after the US ordered DeepSeek off federal systems. The research lab is becoming an empire.
China's giants just ordered $35B of Nvidia chips — Nvidia says it can barely build them
ByteDance, Alibaba and Tencent want more than 3 million RTX PRO 5500 cards. Nvidia's entire 2027 output is about 3 million. The export-compliant loophole is now the bottleneck.
DeepSeek just open-sourced the engine behind its absurdly cheap AI
DeepInfer is the serving stack that lets DeepSeek sell tokens at $0.60 per million. Now anyone can run it. Inference margins just got a lot thinner.
Tenstorrent raises $1.5B at $6.2B to take the RISC-V fight to Nvidia
Jim Keller's Tenstorrent just raised $1.5 billion at a $6.2 billion valuation — more than double its December 2024 mark — with a cap table full of public-market crossover money. After reported Qualcomm takeover talks at $8–10B produced no deal, this raise is the independence play.
Nscale's next Anthropic deal would take one West Virginia campus to ~$90B
Nscale is in talks to sell Anthropic another ~$45 billion of AI computing capacity — a second building at its Monarch campus in West Virginia running ~460 MW of Nvidia's Vera Rubin chips. Stacked on August's ~$45B deal, a single campus would hold one of the largest AI infrastructure commitments ever made.
Default insurance on AI debt just hit a record — $9B a week of hedges
Investors are hedging the AI debt boom at a record pace: credit default swap trading on the AI complex averaged over $9 billion a week in September, Oracle's five-year spreads have more than doubled to 322bp, and CoreWeave's sit near distressed. The equity market says boom. The credit market is buying umbrellas.
Anthropic's leaked safety bill draws fire from advocates — and OpenAI
A leaked draft of Anthropic's proposed federal AI safety bill sets the intervention bar at $1 billion in damages or 50 casualties, with licensed evaluators and a Commerce-only path to court. Safety advocates say it doesn't go far enough and pre-empts the states. OpenAI says no one company should write the rules. Both arguments are really about the moat.
The $10 trillion AI buildout hits its financing reckoning
The FT says America's AI infrastructure boom is “starting to fray” — 47 stalled projects, a debt binge four times the historical average, and grid waits of up to a decade. Then, over the weekend, Oracle and OpenAI walked away from expanding the flagship Stargate site. The buildout isn't stopping; its financing model just changed.
Tenstorrent raises $1.5B at $6.2B — the biggest bet yet on life after Nvidia
While public AI hardware stocks were logging their worst week of the era, private capital wrote the year's biggest alternative-compute check: $1.5 billion for Jim Keller's Tenstorrent, at more than double its valuation of ten months ago. The post-Nvidia ladder just got very real funding.
Beijing drafts its own AI export controls — model weights are now a strategic asset
China proposes to control exports of frontier model weights, AI agents and chip design know-how — a mirror image of Washington's regime, with one elegant asymmetry: open-source models are exempt, and API access doesn't count as an export. The AI stack just got balkanized in both directions.
Samsung wins Nvidia's next-gen HBM — and TSMC builds the part that matters
Samsung landed a place in Nvidia's next-generation custom HBM program — on the condition that the base dies come off TSMC's 3nm line. Nvidia is standardizing the HBM supply chain around its favorite foundry, and Samsung's own fab gets cast as the backup.
The AI hardware rout deepens: SOX logs its longest losing streak since 2008
Chips just had their worst week of the AI era: the SOX's longest losing streak since 2008, a circuit breaker in Seoul, and double-digit losses from Nvidia to SK Hynix. Underneath the rate scare, though, Friday's flows say the capex thesis is being bought, not buried.
The chip squeeze tightens: Taiwan criminalises AI chip smuggling to China
Taiwan is putting AI servers and advanced chips on its controlled-exports list — making smuggling to China a criminal offence — while Washington blocks AMD's China-specific chip and Nvidia's China share collapses to 34%. The two-track chip world just got legal teeth.
China drafts export controls covering 23 AI technologies
Beijing's draft rules could restrict exports of models with 2T+ parameters or trained on sensitive national data. Open-source releases are exempt — a carve-out that says a lot about the strategy.
Nvidia keeps building the AI factory: $2.1B into Iren, a petaflop PC, and orbital servers
Up to $2.1B into neocloud Iren for 5GW of compute, the RTX Spark 'petaflop PC' shipping at $2,599, and co-designed orbital servers with SpaceX — 'Star Mind' hits volume production in 2027.