Nvidia's $96B quarter puts it on the edge of $100B

The AI build-out has a new eye-watering milestone: Nvidia just turned in a record quarter and told Wall Street the next one could be even bigger — the clearest signal yet that the compute arms race shows no sign of slowing.

Nvidia posted a record $96.2 billion in revenue for its second fiscal quarter, up 106% year over year, setting the company up to become the second trillion-dollar quarterly club member on AI demand alone. Data center revenue more than doubled to $89 billion, and net income jumped 126% to $59.7 billion — the most profitable quarter any listed company has recorded. The headline number is striking: $96 billion in a single three-month stretch, driven almost entirely by the datacenter AI chips that have made Nvidia the financial fulcrum of the industry. Guidance is even more telling — Nvidia projects $108 billion next quarter, which would put it over the $100 billion line for the first time. Crucially, that forecast assumes essentially zero datacenter compute revenue from China, meaning the run-rate it's guiding toward is being hit without the market's historically largest buyer counted in.

What to make of it. The numbers confirm that the AI capex supercycle is not softening — it's still accelerating. The wall of money from hyperscalers, plus a record-shattering 106% growth rate, is what makes Nvidia the purest barometer of whether the AI boom has legs. There's a cautionary wrinkle worth watching: investors were unimpressed by the beat — shares slipped slightly in after-hours trading because the headline guidance, while above consensus, fell short of the most bullish Wall Street estimates. The stock story may be decelerating even as the revenue story accelerates, the classic signal that the market is pricing in the future rather than the present. Nvidia also warned of component shortages and rising AI chip prices heading into 2027, which the market will read in both inflationary and demand-confirming directions.

AWS and Nvidia are expanding their multi-year GPU deal by another 2 million chips, aimed squarely at agentic and physical AI workloads. At AWS re:Invent, the pair said they'll deliver 2 million additional Nvidia GPUs across the Blackwell Ultra, Rubin, and Rubin Ultra generations in 2027 and 2028, on top of the 1 million GPUs announced in March. The expanded agreement bundles next-generation infrastructure specifically tuned for agentic and physical AI — the two computing frontiers AWS and Nvidia are betting will dominate the coming deployment wave. It's a big, concrete commitment to a specific timeline, and it turns Nvidia's just-published guidance into more than spin: when your single biggest cloud customer signs up for two million future chips, the demand story has a name attached.

Which of these two compute signals matters more to you — the record number, or the forecast that doesn't need China to hit $108 billion?

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and product

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Read more

Houmo moves compute inside memory — and A20 enters production

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Brain organoids classified Braille at 83% — in a trio, not alone

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute