Data doesn't like to move house. Two AI companies figured that out this week, in opposite ways.
On September 10, OpenAI launched ChatGPT for Financial Services, built with Morgan Stanley and Evercore for investment banking and equity research. It runs on GPT-6 Astra, with data from Daloopa, PitchBook, LSEG News and Crunchbase indexed and hosted directly on OpenAI's own infrastructure. Every figure traces back to the source filing.
Here's the part most coverage skipped. Hosting that data yourself means a firm doesn't need to build a separate connector before analysts can use it. It's a bet that control over the data path is worth owning.
Four days later, on September 14, Anthropic launched Claude for Financial Advisors, aimed at wealth managers rather than bankers. It connects to eleven partners, including Schwab, BlackRock, Addepar and Vanguard. Schwab says more than 16,000 independent RIAs can link Claude to its Advisor Center. Client data stays on the custodian's own servers, and recommendations still require advisor sign-off before reaching a client.
Quieter detail: this launch lands weeks ahead of what Bloomberg and CNBC report is Anthropic's targeted October Nasdaq listing. Deep ties to BlackRock and Schwab double as a roadshow asset, not just a product feature.
Same industry, same week, opposite architecture. OpenAI is centralizing the data. Anthropic is staying out of its way. Competing on model quality is becoming table stakes. The real fight is over whose infrastructure the data touches on the way to the workflow.