What is ChatGPT for Excel? ChatGPT for Excel is OpenAI's spreadsheet-native add-in, powered by GPT-5.5, that lives inside Microsoft Excel and Google Sheets as a sidebar and can build workbooks, update formulas, summarize tabs, and edit cells with permission. On May 5, 2026, OpenAI moved the product out of beta into general availability for Free, Go, Plus, Pro, Business, Enterprise, Edu, and K-12 plans, with a free preview for Business and Enterprise customers running through June 2, 2026. For CRE underwriters who live in multi-tab rent rolls, T12 statements, and capital stack models, this is the most consequential AI tool release of the quarter. For broader context, see our pillar guide on AI multifamily underwriting.
Key Takeaways
- ChatGPT for Excel and Google Sheets went generally available on May 5, 2026, running on GPT-5.5 from a sidebar inside the spreadsheet itself.
- The add-in can create workbooks from prompts, explain formulas, audit errors, and edit cells, but asks permission before changing data.
- Business, Enterprise, Edu, and K-12 customers get a free preview through June 2, 2026, after which usage runs against each plan's existing credits.
- Business and Enterprise data is excluded from model training by default; for Enterprise workspaces, the add-in is disabled until an admin opts in.
- Macros, Power Query, Office Scripts, advanced pivots, and very large workbooks remain unsupported, so legacy CRE templates still need human hands.
What ChatGPT for Excel Actually Does
For two years, CRE analysts had to copy rent roll data into a chat window, paste the answer back, and hope the formulas survived. The May 5 launch closes that loop. The sidebar reads the active workbook, including formulas, references, and assumptions across tabs, and can build new sheets or modify existing ones on instruction. According to OpenAI's launch announcement, the product is generally available globally and runs on the same GPT-5.5 model that became the ChatGPT default earlier this month.
Capabilities relevant to CRE underwriting:
- Workbook creation from natural language: Describe a 10 year DCF for a 200 unit multifamily with stabilized NOI, 1.25x DSCR debt sizing, and a sensitivity on exit cap rate, and the sidebar will generate the tabs, formulas, and structure.
- Cross tab summarization: Ask which sponsor model in a folder shows the lowest implied going in cap rate, and ChatGPT will read each tab and answer with cell level citations.
- Formula audit: The sidebar flags broken references, circular calculations, and inconsistent assumptions across linked tabs, which is the most common reason deals get retraded after IC.
- Permission gated edits: Before changing any cell, ChatGPT presents the proposed edit and waits for confirmation. This addresses the biggest fear underwriters had about giving AI write access to a multi year deal file.
Why This Matters for CRE Underwriting
Roughly 92% of corporate occupiers have started AI programs, but only 5% report achieving most of their AI program goals. In our work with CRE firms, the gap is rarely the model. It is the integration layer. Underwriters did not want to abandon Argus workflows, lender facing T12 templates, or partner approved IRR macros that have evolved over a decade. They wanted AI inside the file, not next to it.
Rent Roll Cleanup Becomes a 5 Minute Job
The standard rent roll arrives as a PDF or a messy Excel export with merged headers, inconsistent unit labels, and trailing notes. A prompt to normalize one row per unit with unit type, square footage, in place rent, and lease expiration in YYYY-MM-DD format now runs against the sheet itself. For a walkthrough of the same workflow in a different tool, see our guide on automating CRE rent roll cleanup.
Capital Stack Sensitivities Get Cheaper
Building a 4 by 4 sensitivity matrix on exit cap rate and rent growth used to require either a junior analyst or a macro that breaks whenever the deal structure changes. The sidebar generates the matrix from a description and updates it when assumptions change, with a permission gate that prevents the model from quietly rewriting the base case.
The Underwriting AI Question Shifts
Previously, CRE firms debated whether to replace Excel with AI. The honest answer was always no, because Argus, lender templates, and partner facing decks all live in spreadsheets. Our existing guide on AI vs Excel for CRE underwriting walks through where the replacement frame fails. The May 5 launch confirms the better frame: AI inside Excel, with the model handling repetitive structural work and the analyst owning judgment on cap rates, comps, and credit.
What the Add In Cannot Do Yet
The release notes flag a long list of unsupported features that matter for CRE. Office Scripts and Power Query, which many shops use to pull data from Yardi, RealPage, and CoStar exports, are not yet supported. Pivot and Data Model functions, which most institutional underwriting decks rely on for portfolio rollups, are out of scope. Macro and VBA automation, the backbone of many partner built templates, will not run inside the sidebar. Very large workbooks may exceed the context window and return partial results. The practical takeaway: the add in handles new modeling and ad hoc analysis well, but does not replace your existing automation stack. The right pilot is a fresh underwriting model on a new deal, not a retrofit of a 15 tab legacy file.
Privacy and Sensitive Deal Data
CRE underwriting models routinely contain non public information: bid prices, sponsor capital sources, lender quotes, and seller pro forma covered by NDAs. OpenAI confirms that ChatGPT Business, Enterprise, Edu, and K-12 customer data is not used to improve models by default. For Enterprise workspaces, the add in is disabled out of the box; admins must enable it, set role based access controls, and choose which workspaces get write access. For deal teams, the answer to whether live LOI data can go into ChatGPT is now closer to yes, conditional on Enterprise controls being live and each workbook being opened from a Business tier account. Consumer Free and Plus plans do not carry the same default exclusion, so do not load NDA covered data into a personal account. The AI Consulting Network works with CRE owners and operators on audited AI access to underwriting files, including policy, IT controls, and the analyst playbook.
Real World CRE Applications
Three early CRE use cases:
- Multifamily acquisitions: An owner of 1,800 units generated a one page deal screen from a broker's rent roll and OM in 11 minutes, calculating in place cap rate, year one stabilized NOI, and a 5 year IRR at 65% LTV.
- Industrial leasing analysis: A regional industrial group compared lease terms across 14 prospective tenants, normalizing TI allowances, free rent, and escalators to a single effective rent per square foot.
- Self storage portfolio: A self storage operator built a four quadrant rent growth and occupancy sensitivity across 22 properties, with the model verifying each property stayed within sponsor set bounds.
CRE investors looking for hands on AI implementation support can reach out to Avi Hacker, J.D. at The AI Consulting Network. According to CBRE's 2026 U.S. Real Estate Market Outlook, gateway and AI hub markets will lead CRE recovery in 2026. Industry research projects the AI in real estate market reaching $1.3 trillion by 2030 at a 33.9% CAGR. ChatGPT inside Excel and Sheets is the cheapest way for a CRE firm to participate without committing to a new platform. For a side by side on which model fits which workflow, our coverage of GPT-5.5 Instant as the new ChatGPT default is the right companion piece.
Frequently Asked Questions
Q: How much does ChatGPT for Excel cost for a CRE firm?
A: Business, Enterprise, Edu, and K-12 plans include a free preview through June 2, 2026. After that, usage runs against each plan's existing credits. Plus and Pro share the same agentic usage limits as Codex; Free and Go have limited usage. Expect each underwriter to consume credits at a meaningful rate during heavy modeling weeks.
Q: Will ChatGPT for Excel see my deal data?
A: For Business, Enterprise, Edu, and K-12 customers, OpenAI does not use that data to train models by default. For Enterprise workspaces, the add in is disabled until an admin enables it and sets role based access controls. Consumer Free and Plus plans do not carry the same default exclusion, so do not load NDA covered data into a personal account.
Q: Can it run my existing underwriting macros?
A: Not yet. Office Scripts, Power Query, Pivot and Data Model functions, macro and VBA automation, and advanced charting are not supported in the current release. Pilot the add in on new underwriting models rather than retrofit legacy templates.
Q: How does ChatGPT for Excel compare to Claude for CRE underwriting?
A: ChatGPT for Excel wins on workflow proximity because it lives in the spreadsheet itself. Claude often wins on long form memo writing and document analysis. Most CRE shops will use both: ChatGPT for in Excel modeling and Claude for OMs, leases, and partner facing memos. The right test is to run a real deal through each this week.