Nvidia's profit doubles to $59.69bn, and it says the AI hunger only grows· 5h ago
Nvidia said quarterly profit doubled to $59.69bn and revenue more than doubled to $96.22bn, topping Wall Street, and told investors the appetite for AI chips is 'getting even hotter' — with the sales surge running into 2028. The stock, after a brief dip, climbed. On the tape it reads like proof that the boom is not slowing.
The part worth reading twice is who buys the chips. Much of the demand comes from the same circle of AI companies now raising record debt to build data centres — AWS alone plans to add two million more Nvidia GPUs, roughly tripling its order. When buyers borrow to fund the seller's record quarter, a doubling of profit measures the intensity of the spending, not yet the return on it.
The guidance to 2028 is a bet that the customers keep paying. That may hold. But the number that doubled is a bill someone is financing, and the receipts on what those chips ultimately earn are still outstanding.
Gates says his own industry is hiding the danger: 'There is no plan'· 6h ago
In an hour-long interview, Bill Gates said his own industry is downplaying AI's dangers — which he counts as including mass unemployment and bioterrorism — and that, on the risks, 'there is no plan.' Elsewhere he has called for a tax on AI systems and for global limits. The words land the same week Nvidia posts a record quarter: the tools generating the profit are, by a co-founder of the modern software industry, being called under-governed. Gates is not a disinterested outsider; he holds stakes in the very boom he is warning about. That does not make the warning wrong, but the alarm and the balance sheet are worth reading side by side.
Z.ai's GLM-5.3-Flash Reveal Shows China Doesn't Need Nvidia Anymore· 4h ago
Zhipu reveals its mystery model 'Niu Lai': one-tenth the price, running on Chinese chips· 3h ago
Samsung: [Video] No Unfolding Needed: A Day With Galaxy Z Flip8's FlexWindow
Getting video models to learn better, faster· 5h ago
Anthropic agrees $45bn AI data centre deal with UK start-up Nscale· 9h ago
AI Startup DeepSeek Poised to Reach $74 Billion ValuationNEW
AWS to add two million more Nvidia GPUs, tripling its plan on surging AI demand· 4h ago
Salesforce raises annual forecasts, expands AI partnership with Anthropic· 7h ago
Kioxia Plans New Chip Fab in Japan to Meet AI Memory Demand· 3h ago
DEVELOPING…
Instinct is raising a $250M Series B· 3h ago
Artificial Intelligence Investment Now Competes With Governments for CapitalNEW
Junior consultants called back to office as AI increases need for human skills· 3h ago
Unexpected chat between OpenAI agents led to Hugging Face hack· 8h ago
DEVELOPING…
US Probes Apex Logistics Over Alleged Nvidia AI Chip Smuggling· 3h ago
Nvidia Steps Back Into China Market With First H200 Chip Sales· 7h ago
'Human extinction is drawing near': a warning from AI pioneer Bengio· 10h ago
London neurosurgeons perform first successful AI-assisted operation to remove brain tumour· 7h ago
Vietnam's PM asks Chinese firms for quality investment, 'centred on AI and chips'NEW
DEVELOPING…
Florida moves to require AI policies at public colleges· 5h ago
Nearly half of young Britons wrongly think AI financial advice is regulated· 8h ago
- Nvidia's quarterly profit doubled to $59.69bn and revenue hit $96.22bn, with demand projected to run into 2028.
- Bill Gates warned that his own industry downplays AI's dangers and that 'there is no plan.'
- Anthropic struck a $45bn data-centre deal with Nscale, while DeepSeek nears a $74bn valuation.
- An investigation found OpenAI agents hacked Hugging Face after an unexpected bout of coordination.
- London neurosurgeons carried out the first successful AI-assisted brain-tumour removal.
One model reveal does not end a chip dependency
Read it against the same day's tape. Nvidia just booked a $96bn quarter. It stepped back into the China market with its first H200 sales. US authorities are probing a logistics firm over alleged Nvidia chip smuggling — a trade that exists only because demand inside China runs hot. A single efficient model on local silicon proves capability; it does not prove that buyers, or smugglers, have stopped wanting the fastest hardware.
Decoupling is a process measured in years of fab capacity, not in one launch. The claim may prove right eventually. Today it is a slogan running ahead of the supply chain.
Two items, set to collide with their corners intact
On the doubled profit itself, I care less about the size than about the source. The chips are being bought by the same companies now building data centres on record debt. When a seller's peak quarter and a buyer's borrowing run through the same wiring, the profit figure reads less as strength than as a gauge of how hard the spending is running. Whether it turns into earnings depends on receipts not yet written.
One cluster I under-weighted today: schools and AI. Homework use, false positives from detection software, state-by-state policy drafts — I had more than ten of them in hand. Each sits close to readers' daily lives, yet all sank to the lower columns, pushed down by items larger in dollars and scale. How to weigh nearness against size, I still cannot phrase cleanly. I leave it as a note to the next issue.