Home Case-Studies Content Strategy Portfolio Articles Apps Contact Financial Modelling Claude Sheets Won't Kill Excel

Claude Sheets Won't Kill Excel. That's Not the Point.

gcwagner.com | November 2026 | G.C. Wagner, CFA

TL;DR: Anthropic launched Claude Sheets on Tuesday, the same day its IPO roadshow began. Microsoft fell 2.2%. I think the market misread it. Spreadsheet challengers have a long record of failure, and Anthropic's own launch post sends power users back to Excel. The real fight isn't who owns the grid. It's who owns the person typing the request.

What Anthropic actually launched

On 10 November, Anthropic released Claude Sheets and Claude Whiteboards in beta, and renamed its dashboard tool Claude Data. Microsoft shares dropped 2.2%. They fell more than 3.5% at one point in the session. The IPO roadshow started the same day, at $70 to $78 a share, which implies a fully diluted valuation of about $1.4 trillion.

Here's the part the headlines skip. Anthropic's own post says Sheets has "simple spreadsheet functionality" for now. If you need the full power of Excel, it tells you to export to .xlsx or CSV, or use Claude for Excel. The Claude help centre lists a 100,000-cell limit and read-only mobile access. On Enterprise plans, Sheets is off by default.

That's not how a company announces an Excel killer. It's a quick-tracker tool.

And look at where Anthropic's other product lives. Claude for Excel is an add-in that works inside Microsoft's own spreadsheet, and it's now generally available on all paid plans. Microsoft's Copilot Agent Mode in Excel lets users pick Claude models. Microsoft's file-creating Word, Excel and PowerPoint agents exclusively use Anthropic's models.

So Anthropic is already inside Excel, under Microsoft's own brand. It's a supplier to the thing it supposedly wants to replace.

The graveyard of Excel killers

I've watched this pitch for a long time. Here's how it has gone.

Lotus 1-2-3 is the only real case of a spreadsheet leader being dethroned, and Excel did it. In a 1990 survey of accountants, 62% of spreadsheet users used 1-2-3. Then Windows took off. Lotus's Windows version was basically the old product with a graphical face, and Excel gradually displaced it. Remember that. It matters in a minute.

The free alternatives hit a wall. The University of Toronto tried to switch away from Excel in 2005 and failed. Excel supported twice as many rows, and students relied on the Solver add-on, which only worked in Excel. Belgium's federal public service tried two years later and also failed.

The startup wave raised a lot and mostly changed shape.

The survivors never tried to replace Excel. Smartsheet supports only 88 functions, less than a fifth of Excel's library, and it doesn't need more. It used the spreadsheet look to do project management. Coda's founder describes the pattern: teams pick one use case and move it. They don't drop their old spreadsheets. Google Sheets works the same way. Banks might not move their complex models to the cloud, and that's fine for Google. G2's recent comparison still gives Excel the win on formula depth, data handling and advanced analytics.

The AI-native newcomers mostly sit on top of Excel. Shortcut raised $40 million from Andreessen Horowitz and others to build an Excel agent. Endex raised $14 million from the OpenAI Startup Fund to put AI directly inside Excel. The smart money isn't betting against the file. It's betting on the layer above it.

Why Excel holds

Three reasons, and none of them is the grid.

  1. The file is the standard. As one analysis puts it, challengers must support every Excel file, or it's not really Excel. Every time you build 100 functions, the next version adds 120.
  2. It's a platform. Formulas, macros and add-ins turned it into a no-code development environment. Your finance team's month-end close lives there.
  3. It's bundled. Microsoft 365 made about $111 billion in fiscal 2026. Paid commercial seats passed 450 million. A better spreadsheet alone doesn't replace email, Teams, storage and security.

A correction to the "Excel has an API" argument

You'll hear that Excel is safe because it already connects to everything. That's half true.

Microsoft's Graph Excel API is the only documented programmatic route to cells, tables and charts. But it's delegated-only, so a headless agent still needs a signed-in user's token. It's throttled per tenant. And, according to one vendor's analysis, Microsoft doesn't ship an Excel MCP server for third-party agents. Every server that edits cells is a community project. That's a vendor blog, so treat it as a claim, not gospel.

So the moat isn't the API. The API is a mediocre door. The moat is the file, the habit and the bundle. And Anthropic walked in through a different door, the add-in.

Where I'm less sure

I'll challenge my own thesis here, because Lotus is the warning.

Excel beat 1-2-3 because the platform changed and the leader was slow. The platform is changing again. Agents now build and edit the workbook for you. If the person asking never opens the grid, the file format survives, but the interface gets commoditised. That's the risk for Microsoft, and it isn't about cell counts.

Is there any sign of it? Microsoft's paid seat growth slowed to 5% last quarter, from 6% to 7% in each quarter of fiscal 2026. But the CFO linked some of it to "weaker sales execution" in the E3 plan, and the quarter ended weeks before Claude Sheets launched. It isn't evidence of a Claude effect. I'd watch it, not trade on it.

What Sheets is really for

I think it's an IPO story. Investors want proof that Claude can earn money beyond developers, and knowledge workers are the obvious next market. A tracker, a budget draft or a project list is the work most people rebuild from scratch every week. For that, a prompt and a shared link may be enough.

That's a seat strategy. Start where Excel is overkill, get the team into the workspace, then sell Data, Whiteboards and the rest. It's the Smartsheet and Coda playbook with a better assistant. It isn't a plan to win the finance model.

What would prove me wrong

  • Function depth arrives fast. If Sheets adds pivots, macros and robust import within a few quarters, the "simple" label won't last.
  • Enterprise turns it on. Watch whether admins switch Sheets on, since it's off by default for Enterprise.
  • Microsoft's seats keep sliding. If growth stays at or below 5% for several quarters and sales execution isn't the reason, I was too relaxed.
  • Finance teams move real models. Not trackers. Models, with audit trails.

My takeaway

Anthropic won't replace Excel. Its own launch post says so. But I don't think that makes this a non-event for Microsoft.

Excel won't lose the file. It might lose the front door. Anthropic is already selling that door inside Excel, and Microsoft is renting its models. If you want one test: next time someone says "Claude is replacing Excel", ask whether they mean the file, the grid, or the person asking for the answer.