Oracle just proved the AI cloud trade is real — the bill for it is coming due anyway

Oracle's quarter was the strongest argument yet that AI infrastructure demand is not a bubble story on its own numbers — and the stock still owes its future to a single customer. The software giant reported fiscal Q1 results on Thursday with cloud infrastructure revenue up 121% year over year to $7.4 billion, leaving the rest of the company growing just 3%. It brought 850 megawatts of new data center capacity online in the quarter, and capital expenditures jumped to $28.5 billion from $8.5 billion a year ago as it races to stand up AI servers for enterprise customers.

The number that matters more is the backlog: $664 billion in remaining performance obligations, up sharply, with almost half of it attributable to a single cloud infrastructure contract with OpenAI. That concentration is why Oracle's stock has become a proxy for sentiment around OpenAI, which is expected to go public soon — and why the bears keep coming back to the balance sheet. Oracle ended the quarter with negative free cash flow of $5.4 billion and sits on a $125 billion debt pile, while hyperscaler rivals AWS and Google carry far stronger cash positions. The reassuring wrinkle: Oracle's non-OpenAI backlog more than doubled over the past year, so the company is not betting entirely on one startup's fortunes. Analysts quoted after the call — Rebecca Wettemann of Valoir and Brian Mulberry of Zacks Investment Management — both read the quarter as proof Oracle is converting that backlog into recognized revenue rather than just warehousing it. Our take: this is the clearest evidence yet that AI compute demand converts into actual dollars, but a $90–95 billion capex budget financed with debt against one giant counterparty is a structure that only works while AI spending accelerates. The moment it decelerates, the leverage cuts the other way.

For fiscal 2027 Oracle is targeting at least $90 billion in revenue and $8.10 in adjusted earnings per share, with capex guidance unchanged at $90–95 billion.

Enflame, the Tencent-backed AI chipmaker, lists on Shanghai's STAR Market today — one of the most anticipated Chinese chip debuts of the year. Reuters reported the IPO raised about $912 million, priced at 142.18 yuan per share, making it the last of China's so-called GPU Big Four to go public and completing a domestic listing run for the country's AI silicon champions. The timing is pointed: Enflame builds its chips without Nvidia's software stack, and the listing lands as Chinese buyers hunt for domestic alternatives amid US export controls.

We covered Enflame's IPO pricing in August — Tencent-backed Enflame sets $908M Shanghai IPO at 142.18 yuan per share — and the oversubscription frenzy a week ago, when retail demand was swamped; Enflame raises $908M, and retail got 0.025% of what it wanted. Today's first trading day is the moment the froth meets a real valuation test.

What to watch: whether Enflame opens above its IPO price — a strong pop would embolden the pipeline of Chinese AI chip listings waiting behind it.

Should AI chipmakers be valued on national champions logic or on Nvidia comparison math? Tell us in the comments.

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

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon turned its always-listening Bee line into a two-device family on Thursday, and NVIDIA opened a 2.8-trillion-token Indic corpus recipe. The connecting thread: both are bets that the next AI money is in memory — yours, or your model's. Amazon put two new always-listening devices on sale, and both now lead to a subscription. Bee Frontier, the next-generation wristband that also clips to clothing, ships now at $74.99 in Onyx, Lavender and Sage — up from the $50 predecessor — with seven-day

Perplexity open-sources its agent sandbox after 118 failed escapes

Perplexity handed its agent containment layer to the open-source world, and Washington's AI task force stopped reading briefings and started working. Perplexity open-sourced SPACE on Thursday — the sandbox it built to keep its own AI agents from getting out — and says nine frontier models failed to break out of it across 118 attempts. Each agent runs inside a small virtual machine with its own operating system, with credentials and network access managed outside the sandbox so an agent can re

Read more

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

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

Amazon's new Bee devices listen all day — for $11.99 a month

Perplexity open-sources its agent sandbox after 118 failed escapes