ChatGPT for financial services brings built-in Wall Street data and GPT-6 Astra reasoning to investment banking teams

OpenAI announced on 10 September 2026 that it is launching ChatGPT for Financial Services, a version of ChatGPT Work built specifically for investment banking and equity research teams. It is not a generic finance-themed chatbot. It is a specialised workspace that bundles institutional-grade financial data directly into the product, powered by GPT-6 Astra, OpenAI’s newest reasoning model.

The product was shaped through design partnerships with Morgan Stanley and Evercore, two firms whose teams helped OpenAI identify where the real friction sits: getting reliable access to financial data and producing high-quality deliverables quickly.

What is actually included

The headline feature is built-in data from four providers: PitchBook, Daloopa, LSEG News, and Crunchbase. These cover earnings transcripts, normalised financial statements, company fundamentals, private market data, and news. OpenAI indexes and hosts this data on its own infrastructure, which means teams can use it immediately without negotiating separate data contracts or configuring connectors.

That hosting arrangement also enables granular citations. When a banker generates an analysis, they can trace individual figures back to specific tables and passages in the source material. For something like an adjusted EBITDA reconciliation, that means being able to inspect which costs were excluded and why, rather than accepting a number at face value.

Beyond the four deeply integrated providers, the product connects to more than 50 model context protocol connectors, including Datasite, Box, Preqin, and Intapp. OpenAI is also building sign-in and entitlement integrations with S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva, and Moody’s, which would let users access data they already subscribe to through their ChatGPT login rather than switching between platforms.

The workflows it targets

OpenAI has been specific about the use cases: value analysis, LBO modelling, buyer screening, earnings analysis, and pitchbook preparation. The idea is that the research, calculations, and final deliverables can increasingly happen inside the same controlled workspace rather than across a patchwork of terminals, spreadsheets, and document editors.

Firm administrators can publish Excel, Word, and PowerPoint templates through a dedicated admin page, so outputs arrive in the organisation’s own format and style rather than generic AI-generated layouts. That matters in a business where the appearance of a pitchbook carries its own signal.

GPT-6 Astra has been tuned specifically for retrieval across financial data tools, financial reasoning, and document generation. As Nick Turley, OpenAI’s vice president of product, told CNBC: “We’re effectively teaching ChatGPT to research like an analyst and back up its conclusions like an analyst as well.”

Security and governance

The product builds on the security foundations of ChatGPT Enterprise. It includes SAML single sign-on, SCIM provisioning, role-based access controls, encryption at rest and in transit, configurable data retention settings, and exportable compliance logs that feed into existing audit workflows. Firms can create multiple workspaces to enforce information barriers between teams. OpenAI states that firm business data is not used to train its models by default.

What this means for you

If you work in investment banking or equity research, the practical shift here is speed and traceability. Tasks that have historically consumed hours of an analyst’s time, pulling comparables from PitchBook, normalising financials from Daloopa, screening buyers, assembling a first draft of a pitchbook, become something that begins inside a single interface with the source data already present and citable.

For senior bankers and heads of research, the more immediate question is governance. The compliance logging, role-based controls, and information barrier tooling are clearly aimed at making this viable within regulated environments, not just technically impressive in a demo. Whether those controls satisfy your firm’s compliance requirements will depend on your jurisdiction and internal standards, but OpenAI has clearly engineered for that conversation.

For data providers, the split between the deeply integrated group (PitchBook, Daloopa, LSEG News, Crunchbase) and the sign-in integration group (S&P Global, MSCI, Moody’s) is worth watching. The former opted to let OpenAI host and index their data, gaining prominent placement in the product experience. The latter chose a more defensive model that preserves their own distribution. Over time, that distinction may matter a great deal to which providers become the default starting point for financial research.

The broader picture

The rollout places OpenAI squarely in territory that junior bankers, analysts and associates, have occupied for decades. Wall Street has long debated how AI changes the entry-level role, and this product makes that question more concrete. If AI handles the research and initial modelling, how do junior staff develop the judgement that comes from doing that work themselves? That is a genuine question the industry has not yet answered.

Financial institutions including Morgan Stanley, BNY, Fidelity International, MUFG, and Commonwealth Bank are already using or evaluating OpenAI’s tools. The integrations OpenAI builds with these partners will, the company says, inform both post-training improvements and its expansion into other financial services categories beyond investment banking.

Pricing is not public. Eligible financial institutions can contact OpenAI sales or their account team directly.