OpenAI and Google spent September 2026 racing to plant a flag on the same piece of ground: Wall Street’s back office. OpenAI introduced ChatGPT for Financial Services, a tailored ChatGPT Work experience built around GPT-6 Astra’s reasoning and a set of built-in financial data feeds. Google answered with Gemini Enterprise for Financial Services, an agentic platform aimed at capital markets and corporate banking teams. Both launched within days of each other, and both are chasing the same prize: the hours analysts spend building models, chasing data, and drafting client materials.
The timing is not a coincidence. Financial services has become the proving ground for whether large language models can move past drafting emails and into work that carries real financial consequences. This piece breaks down what each company actually shipped, who is already using it, and what the ChatGPT for Financial Services vs Gemini Enterprise fight signals about where enterprise AI is headed next.
What OpenAI Just Launched: ChatGPT for Financial Services
OpenAI’s announcement, titled “Introducing ChatGPT for Financial Services,” describes the product as a tailored ChatGPT Work experience that combines built-in financial data with GPT-6 Astra’s reasoning to help teams develop research, financial models, and customized client materials. OpenAI states plainly that ChatGPT for Financial Services is available to eligible financial institutions, which signals a controlled rollout rather than a self-serve signup most consumer ChatGPT users are used to.
According to CNBC, the product was built with design partners Morgan Stanley and Evercore, two firms whose day-to-day work runs on exactly the kind of research and modeling tasks OpenAI is targeting. CNBC also reports that the tool draws native data from LSEG, Daloopa, Crunchbase, and PitchBook, four sources that cover public market data, financial statement extraction, and private company intelligence respectively. That combination matters because it moves ChatGPT from a general-purpose assistant into something closer to a purpose-built research terminal, at least for the institutions granted access.
The OpenAI announcement frames the release around GPT-6 Astra specifically, the reasoning model OpenAI rolled out earlier this year as its flagship agentic system. Pairing a frontier reasoning model with licensed financial data is the clearest signal yet that OpenAI wants ChatGPT for Financial Services measured against the workflows of junior analysts, not against a chatbot.
Inside Gemini Enterprise for Financial Services
Google’s answer sits inside its broader Gemini Enterprise platform, which the company describes generally as an intranet search, AI assistant, and agentic platform. Google’s documentation lists prebuilt connectors for Confluence, Jira, Microsoft SharePoint, and ServiceNow, plus a single, multimodal search interface with permissions-aware access to enterprise information. That permissions layer is the detail compliance teams at banks will scrutinize first, since financial data access rules do not bend for a chat interface.
Google’s blog post, titled “Introducing Gemini Enterprise,” says the platform brings the best of Google AI to every employee through an intuitive chat interface that acts as a single front door for AI in the workplace. On the finance-specific side, Google Cloud’s announcement puts it directly: “Today, we are delivering on this vision with Gemini Enterprise for Financial Services, bringing Google’s agentic AI directly into the workflows of capital markets and corporate banking.”
Google’s own product page adds a commercial detail that OpenAI has not matched publicly: a Business edition free for 30 days, open to individuals, small businesses, and teams up to 300 seats, alongside a Standard or Plus edition also free for 30 days with higher quota and unlimited seats at the enterprise tier. Google’s finance-industry landing page sums up the pitch in one line: the Gemini Enterprise app lets a firm run AI agents in one secure platform. That framing pushes the conversation from a single assistant toward a fleet of task-specific agents operating under one governance layer.
GPT-6 Astra: The Reasoning Engine Behind OpenAI’s Pitch
GPT-6 Astra is doing more work in this launch than a typical model upgrade. OpenAI’s own materials describe the product as combining built-in financial data with GPT-6 Astra’s reasoning, which suggests the model itself, not just the data plumbing around it, is central to the pitch. Astra has already drawn attention this year for its scale, and shattered.io previously reported on its 100,000-GPU training run, a scale that OpenAI has leaned on to justify claims about the model’s reasoning depth.
Financial modeling is a demanding test case for a reasoning model. A comparable company analysis, a discounted cash flow build, or a credit memo requires chaining dozens of small judgment calls together without losing track of the assumptions made three steps earlier. It also requires the model to show its work in a form an analyst, or a regulator, can audit. Fortune reports that Nick Turley, OpenAI’s vice president and head of ChatGPT, has named finance as one of a small set of verticals the company has chosen to build specific products for, alongside cybersecurity and software engineering. That is a narrower bet than OpenAI’s earlier horizontal push with ChatGPT Enterprise, and it is a tell about where the company thinks the highest-value, most defensible use cases sit.
Morgan Stanley and Evercore: Why Design Partners Matter
Naming Morgan Stanley and Evercore as design partners is not just a marketing flourish. Both firms have spent the past two years building internal AI programs of their own, which means OpenAI was shipping a product into an environment where the customer already had opinions about what “good” looks like. Design partners of this caliber also carry weight with other banks watching from the sidelines: if a firm the size of Morgan Stanley trusts a tool with research and modeling work, smaller shops read that as a signal the compliance and security review has already been stress-tested once.
CNBC’s reporting frames the collaboration as building the product together, not simply piloting a finished tool. That distinction matters for how fast the product can adapt to real trading-desk and deal-team workflows versus a generic enterprise chatbot retrofitted with a finance skin. It also raises a fair question for every other bank watching this launch: does early access to shape the tool create a durable edge for the design partners, or does the feature set converge quickly once OpenAI opens access more broadly to eligible institutions?
Data Access: LSEG, Daloopa, Crunchbase, and PitchBook vs Google’s Connectors
The two companies are approaching the data problem from opposite directions, and that gap is the single most important technical difference in this comparison. OpenAI’s approach, per CNBC, licenses named financial data providers directly into the product: LSEG for market and reference data, Daloopa for structured financial statement extraction, and Crunchbase and PitchBook for private company and deal intelligence. That is a bet on depth in a narrow domain.
Google’s public documentation instead emphasizes breadth of enterprise systems: connectors for Confluence, Jira, Microsoft SharePoint, and ServiceNow, wrapped in permissions-aware search across whatever a firm already has stored internally. That is a bet on being the layer that sits on top of a bank’s existing document sprawl rather than replacing any single data vendor. Google’s financial-services messaging leans on the word “agentic,” implying the platform is meant to trigger workflows across those connected systems, not just answer questions about them.
Put simply: OpenAI is selling depth on market and company data, Google is selling reach across a firm’s internal systems. A bank chasing faster comp sets and DCF models might lean toward OpenAI’s approach. A bank trying to cut the time analysts spend hunting through SharePoint and ServiceNow tickets for the right internal document might lean the other way.
Feature Comparison: ChatGPT for Financial Services vs Gemini Enterprise
The table below lines up what each company has publicly confirmed as of this week. Where a company has not disclosed a detail, the cell says so rather than guessing.
| Category | ChatGPT for Financial Services (OpenAI) | Gemini Enterprise for Financial Services (Google) |
|---|---|---|
| Underlying model | GPT-6 Astra | Google’s Gemini family (specific model not disclosed for the finance product) |
| Base platform | ChatGPT Work | Gemini Enterprise agentic platform |
| Named data partners | LSEG, Daloopa, Crunchbase, PitchBook | Not disclosed for finance-specific data; general connectors listed below |
| Enterprise system connectors | Not detailed publicly | Confluence, Jira, Microsoft SharePoint, ServiceNow |
| Named design/launch partners | Morgan Stanley, Evercore | Not named publicly for the financial-services edition |
| Access model | Available to eligible financial institutions | Business edition free 30 days, up to 300 seats; Standard/Plus free 30 days, unlimited seats |
| Core positioning | Research, financial modeling, client materials | Intranet search, AI assistant, and agentic workflows across capital markets and corporate banking |
Pricing and Access: Who Can Get In, and What It Costs
Neither company has published a public price list for its financial-services product specifically, which is typical for enterprise software sold through direct sales motions rather than self-serve checkout. What both companies have confirmed is how a firm gets a foot in the door.
| Access detail | OpenAI (ChatGPT for Financial Services) | Google (Gemini Enterprise) |
|---|---|---|
| Eligibility | Eligible financial institutions (application-based) | Open sign-up for the general Gemini Enterprise trial tiers |
| Free trial | Not publicly stated | 30 days, both Business edition and Standard/Plus edition |
| Seat cap on trial | Not disclosed | Up to 300 seats (Business edition, individuals through teams) |
| Enterprise tier | Custom, through design-partner and eligible-institution track | Unlimited seats on Standard or Plus edition with higher quota |
| Public pricing page | Not published for the finance-specific product | Published trial tiers; enterprise pricing on request |
The Analyst Replacement Question: What Nick Turley Actually Said
Every enterprise AI launch in finance now runs into the same question: is this a tool for analysts, or a replacement for them? OpenAI’s own framing leans toward augmentation, but the language it chose is pointed. 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.”
That is a deliberate comparison to junior banker work: pulling comparable companies, building a first-pass model, and citing sources well enough that a senior banker can trust the output without redoing it from scratch. Whether the model clears that bar in practice, on live deals with real capital on the line, is the test every design partner is running right now behind closed doors. OpenAI has not published benchmark results specific to financial modeling accuracy for GPT-6 Astra, so the honest answer is that outside observers cannot yet grade the claim independently.
Historical Context: From Generic Copilots to Vertical Finance AI
This launch did not appear out of nowhere. Enterprise AI in finance moved in three rough phases. First came generic copilots bolted onto existing productivity suites, useful for summarizing meetings and drafting emails but not trusted with numbers that mattered. Next came retrieval-augmented tools that let a model search a firm’s own documents, which cut research time but still left the modeling and judgment work to humans. What OpenAI and Google are both shipping now is the third phase: models paired with licensed, structured financial data and enough reasoning capability that firms are willing to test them on research and modeling tasks that used to sit exclusively with analysts.
Google has been building toward this since it consolidated its enterprise AI efforts under the Gemini Enterprise brand, a move covered in earlier shattered.io reporting on Gemini’s market reach as the assistant expanded across Google’s platforms. OpenAI’s path ran through ChatGPT Enterprise and ChatGPT Work before narrowing into vertical-specific products, with finance, cybersecurity, and software engineering as the first named targets, according to Fortune’s reporting on Nick Turley’s comments.
Market Impact: What This Means for Bloomberg Terminal and Legacy Fintech
The obvious incumbents watching this launch are the terminal and data providers whose businesses depend on being the place analysts go first. Neither OpenAI nor Google has announced plans to build a standalone terminal competitor, and both are, for now, licensing data from established providers rather than replacing them outright. LSEG’s data sits inside OpenAI’s product rather than being displaced by it, which suggests the near-term impact on data vendors is a shift in how their data gets consumed, not a direct threat to their existence.
The bigger near-term impact falls on the workflow layer: the internal tools banks built themselves to stitch together research, modeling templates, and client-ready output. If ChatGPT for Financial Services or Gemini Enterprise can do that stitching reliably, firms have less reason to keep funding bespoke internal tooling teams for the same job. That is a real budget line, even if it never shows up in a press release. It also raises the stakes for other cloud and AI infrastructure providers competing for the same enterprise AI spend, a dynamic shattered.io has tracked as the broader AI infrastructure market consolidates around a handful of dominant platforms.
Competitive Landscape: Where Anthropic and Microsoft Fit In
OpenAI and Google are not the only companies with a stake in this fight, even though they are the two with named, finance-specific products this week. Anthropic has positioned Claude for regulated-industry work generally, though it has not detailed a packaged, finance-specific product on the scale of what OpenAI and Google announced this month. Microsoft, through its existing Azure OpenAI partnership and Copilot lineup, already has a foothold inside many of the same banks now piloting ChatGPT for Financial Services, since Microsoft’s cloud and productivity relationships with large financial institutions run deep and predate this specific product race.
That creates an unusual dynamic: a bank could be running Microsoft Copilot for general productivity, piloting ChatGPT for Financial Services for research and modeling, and evaluating Gemini Enterprise for its agentic connectors, all inside the same institution. Enterprise software rarely settles into single-vendor deals this early in a category’s life, and finance-specific AI looks likely to follow that same multi-vendor pattern for at least the next several quarters. Microsoft’s own cloud disclosures give a sense of scale here: the company recently broke out Azure revenue at $101.9 billion for the fiscal year, a base that funds deep, existing relationships with the same banks now piloting rival products.
Security and Compliance Concerns for Regulated Financial Firms
Neither company has published a detailed compliance or security architecture document specific to its financial-services product this week, which leaves banks’ own security and compliance teams to do the diligence work before any broad rollout. Google’s emphasis on permissions-aware access, letting the platform respect a firm’s existing document permissions, addresses one obvious risk: a model surfacing information a given employee should not see. OpenAI has not published equivalent detail publicly for ChatGPT for Financial Services, though “eligible financial institutions” language implies some form of vetting process ahead of access.
For a bank’s security team, the practical checklist is the same regardless of vendor: where does the model run, what happens to prompts and outputs after a session ends, can the firm keep data within its existing residency and retention requirements, and can an auditor reconstruct why the model produced a given number in a model or memo. Regulated industries move slowly on this kind of question for good reason, and it is a fair bet that the gap between announcement and broad production use inside most banks will be measured in quarters, not weeks.
What Experts Are Saying
OpenAI frames the release in its own words as a tailored ChatGPT Work experience “that combines built-in financial data with GPT-6 Astra’s reasoning to help teams develop research, financial models, and customized client materials,” according to the company’s official announcement.
Nick Turley, OpenAI’s vice president of product, put the ambition in blunter terms to CNBC: “We’re effectively teaching ChatGPT to research like an analyst and back up its conclusions like an analyst as well.”
Google Cloud describes its own launch as the delivery of a longer-term plan, stating in its official announcement: “Today, we are delivering on this vision with Gemini Enterprise for Financial Services, bringing Google’s agentic AI directly into the workflows of capital markets and corporate banking.”
Google Workspace’s finance-industry page keeps its pitch simple, saying the Gemini Enterprise app “lets your firm run AI agents in one secure platform,” a line that captures Google’s bet on breadth across connected systems rather than depth in any single financial dataset.
Predictions: Where Financial AI Goes From Here
- Expect at least one more major AI lab to announce a named financial-services product within the next two quarters, most plausibly Anthropic or Microsoft, given how deep both already sit inside bank IT stacks.
- OpenAI’s eligible-institution model will likely widen gradually rather than open to self-serve signup, mirroring how ChatGPT Enterprise expanded access in stages after its own launch.
- Data licensing deals, not model quality alone, will become the real competitive battleground, since LSEG, Daloopa, Crunchbase, and PitchBook access is a moat OpenAI can defend contractually in ways a raw model cannot be defended.
- Banks will run multi-vendor pilots rather than picking a single winner in year one, keeping Microsoft Copilot, ChatGPT for Financial Services, and Gemini Enterprise in parallel evaluation for most of 2027.
- Regulatory scrutiny of AI-generated financial research and client materials will sharpen once one of these tools is tied to a real trading loss or a flawed client recommendation, a moment neither company has faced publicly yet.
Frequently Asked Questions
What is ChatGPT for Financial Services?
ChatGPT for Financial Services is OpenAI’s tailored ChatGPT Work experience for banks and financial institutions, combining built-in financial data with GPT-6 Astra’s reasoning to support research, financial modeling, and client materials. It is available to eligible financial institutions rather than the general public.
What is Gemini Enterprise for Financial Services?
Gemini Enterprise for Financial Services is Google’s agentic AI platform built for capital markets and corporate banking workflows. It runs on the broader Gemini Enterprise product, which offers intranet search, an AI assistant, and connectors to systems like Confluence, Jira, Microsoft SharePoint, and ServiceNow.
Which banks are using ChatGPT for Financial Services?
Morgan Stanley and Evercore are named as design partners in OpenAI’s launch, according to CNBC. OpenAI has not published a full list of institutions with access, since the product is available to eligible financial institutions rather than the general public.
What data sources power ChatGPT for Financial Services?
CNBC reports the product has native data access from LSEG, Daloopa, Crunchbase, and PitchBook, covering market and reference data, structured financial statement extraction, and private company and deal intelligence.
Is Gemini Enterprise free to try?
Google’s Gemini Enterprise page states the Business edition is free for 30 days for individuals, small businesses, and teams up to 300 seats, and that the Standard or Plus edition is also free for 30 days with higher quota and unlimited seats.
What model powers ChatGPT for Financial Services?
OpenAI says the product combines built-in financial data with GPT-6 Astra’s reasoning, making Astra the underlying model behind the financial-services experience.
Will these tools replace financial analysts?
OpenAI’s own framing positions the tool as augmenting analyst work rather than replacing it outright, with Nick Turley describing the goal as teaching ChatGPT to research and back up conclusions the way an analyst would. Neither company has published data on job or headcount impact, so any claim about replacement remains speculative at this stage.
How does Gemini Enterprise for Financial Services differ from regular Gemini?
Gemini Enterprise for Financial Services applies Google’s broader Gemini Enterprise agentic platform, including permissions-aware search and enterprise system connectors, specifically to capital markets and corporate banking workflows, according to Google Cloud’s announcement.
