OpenAI picked a new vertical as the BIS warns on the AI boom
OpenAI picked a new vertical, the bank for central banks picked a fight with the AI capex boom, and Wall Street is bracing for the biggest IPO window in years. Three stories, one theme: AI and money are now the same conversation.
OpenAI has launched ChatGPT for Financial Services, a tailored ChatGPT Work edition that pairs built-in premium financial data with GPT-6 Astra's reasoning. Built with design partners Morgan Stanley and Evercore, it starts where the pain is sharpest: investment banking and equity research. OpenAI hosts and indexes datasets from Daloopa, PitchBook, and LSEG News on its own infrastructure — no separate data contracts or connector setup — and adds granular citations so a banker can trace any figure back to the underlying table or passage. Firms keep their existing S&P Capital IQ, MSCI, Dow Jones Factiva, and Moody's subscriptions through shared sign-in, publish their own Excel, Word, and PowerPoint templates, and turn analysis into valuation models and pitchbooks in house style. On OfficeQA Pro, a benchmark built on US Treasury bulletins with dense financial tables, GPT-6 Astra scores 69.9% against 60.2% for GPT-5.6 Sol. The hosted-data-plus-citations design is the real story: verifiability is what kept finance teams skeptical of chatbots, and OpenAI is attacking that blocker directly. The company keeps verticalizing ChatGPT — hours ago it gave the product away to US public servants, as we covered in OpenAI gives 23 million US public servants ChatGPT at $0.
The Bank for International Settlements warns the AI investment boom is accumulating financial-stability risk. In its annual economic report, the BIS — the umbrella institution for the world's central banks — flags a widening gap between AI capital spending and the cash flows meant to fund it, bridged increasingly by debt and concentrated financing: the recipe that turned past investment booms into busts. Its leadership called out the new risks directly this week. The timing is pointed — the warning lands just as public markets prepare to price AI companies at record valuations.
Anthropic's IPO filing has analysts expecting a wave of AI listings to break within the next two months. Bloomberg reports a cluster of AI-driven IPOs is expected to accompany Anthropic's debut, with the company reportedly aiming to raise around $100 billion in what would rank among the largest offerings ever. The window matters for the whole sector: labs burning capital on compute need exits, and public-market appetite for AI is the euphoria side of the same coin the BIS is warning about.
What to watch: whether Wall Street actually adopts ChatGPT for client work — and whether the BIS's rhetoric moves regulators before the IPO window opens.
Would you trust an AI-generated pitchbook if every number linked back to its source? Tell us in the comments.
Sources: OpenAI · Reuters (OpenAI launch) · Yahoo Finance · Reuters (BIS warning) · BIS Annual Economic Report · Bloomberg · The New York Times · CLS (财联社)
