An AI lab files to go public, software stocks wobble and compute gets a new option. Three AI stories that matter this week
1. Anthropic files for a Nasdaq IPO
On 26 October, Anthropic filed its public S-1 to list on the Nasdaq under the ticker ANT. The filing reports $16.3 billion of revenue in the first half of 2026, up from $1.16 billion a year earlier. It also reports a net loss of $80.7 billion, most of it a non-cash charge tied to convertible notes, and an operating profit in the second quarter. Customers paying $100,000 or more a year grew from more than 950 to more than 11,900 in twelve months. About 48% of first-half revenue came through the Amazon, Google and Microsoft cloud marketplaces.
Why it matters: This is the first time a frontier AI lab has opened its books in this much detail. For enterprise buyers, the numbers confirm how fast AI spend is growing, and how much of it already flows through cloud marketplaces and existing cloud commitments.
2. ServiceNow’s sell-off shows “AI eats software” has moved from theory to share price
ServiceNow’s Q3 revenue of $4.11 billion beat forecasts, and net new annual contract value from AI products rose 65%. Even so, slower growth in contracted revenue sent the shares down as much as 27% on 29 October, which CNBC called the company’s worst day ever, amid investor fears that AI agents will erode workflow software. On the same day, ServiceNow announced a ChatGPT plugin, a Codex integration and a “Control Tower Alliance” with OpenAI, Anthropic and NVIDIA.
Why it matters: The market is asking whether the system of record or the AI front door owns the customer. For architects, the safest position is to keep business logic, data and audit in platforms you control, and to treat AI front ends as interchangeable channels.
3. Google TPUs are coming to Oracle Cloud in 2027
Oracle and Google announced on 28 October that Google’s TPUs will be available on Oracle Cloud Infrastructure from 2027. It is the first hyperscale cloud outside Google to offer them, and it includes the new TPU 8t for training and TPU 8i for inference.
Why it matters: With AI capacity still tight, more choice of accelerators is welcome. It also rewards teams whose models and serving stacks are not hard-wired to one chip family or one cloud.
Our view
This week’s common thread is that the AI market is maturing. The labs are publishing audited numbers, incumbents are being re-priced and compute is diversifying. For enterprises, portability and clear ownership of data and process matter more than ever.
Sources
- Anthropic IPO: the company behind Claude files to list as ANT (StockTitan)
- Anthropic IPO live updates (CNBC)
- ServiceNow shares plunge 27% to worst day ever after earnings revive AI disruption fears (CNBC)
- ServiceNow’s crash and its OpenAI deal (24/7 Wall St.)
- Oracle to offer Google TPUs on OCI cloud from 2027 (IT Brief UK)