Google Sheets cell limit: it is 20 million now, and what breaks first is not cells

The Google Sheets cell limit is 20 million cells per spreadsheet, not the 10 million figure almost every page still quotes. Google made the higher cap generally available on September 10, 2026. In our view the headline number is the least interesting part: a 100MB file ceiling, Connected Sheets row caps and script timeouts stop a large sheet long before you get near 20 million.

This piece sticks to what Google publishes, with the date on each claim, and separates the limits that are real walls from the ones that only feel like walls.

What the Google Sheets cell limit is today, and when it changed

Google's Workspace Updates post dated September 10, 2026 says spreadsheets now support "up to 20 million cells," up "from up to 10 million cells," and that "this limit applies to new, existing, and imported files." The same post puts the gradual rollout at September 10, 2026 for Rapid Release domains and September 28, 2026 for Scheduled Release domains, with feature visibility taking up to 15 days in each case.

Three details in that announcement matter more than the number:

• It reaches everyone: all Google Workspace customers, Workspace Individual subscribers, and users with personal Google accounts.
• There is no switch. Google states that admins have no control over the feature and users have no setting for it, so the higher cap simply applies.
• Existing and imported files are included, so a file that refused an import in August may accept it now without anything changing on your side.

The cap was a beta first. Google's April 22, 2026 post announced the move from 10 million to 20 million cells as a beta that organizations had to join by form, with registrations processed in waves. If you tested the limit between April and September and were told you could not have it, that is why.

Google's own help page has caught up. The Drive help article on files you can store now reads "Up to 20 million cells or 100MB for spreadsheets that are created in or converted to Google Sheets" (checked October 23, 2026). Most of the first page of results for this query, including Google's own 2022 announcement, still says 10 million, so check the date on anything you read about it.

100MB is the ceiling a big sheet usually hits first

Read that help page wording again: the cap is 20 million cells or 100MB, whichever comes first. Cells are cheap when they hold a date or a two-digit number and expensive when they hold a paragraph of text, a long URL or a formula. Fitting 20 million cells inside 100MB means averaging roughly five bytes of stored data per cell, which we would not expect any sheet holding text to manage.

So the practical answer to "how many cells can a Google Sheet have" is: as many as fit in 100MB, up to 20 million. Two other documented constraints sit next to it on the same page, checked October 23, 2026:

• Converting from Excel drops content: "any cell with more than 50,000 characters will be removed in Sheets." That is silent data loss on import, not an error.
• Google lists an increase to the imported file byte size limit as coming soon in the September 2026 post, with no date. Until it ships, the 100MB figure is the one to plan against.

How many rows can Google Sheets handle? That depends on your columns

Google publishes a cell cap, not a row cap, so your maximum row count is the cell cap divided by the width of your table. The arithmetic below assumes a single tab, nothing else in the file, and the 20-million-cell ceiling:

This is where Sheets and Excel differ in kind, not degree. Microsoft documents a fixed worksheet size of "1,048,576 rows by 16,384 columns" and 32,767 characters per cell, per sheet rather than per workbook. Excel gives you a million rows no matter how wide the table is; Sheets trades rows against columns and counts the whole file. A 60-column export of 900,000 rows is inside Excel's grid and nowhere near a Google Sheet.

One thing to check on your own file: Google does not publish whether the blank grid on each tab counts toward the file total. Trimming unused rows and columns off every tab is the cheapest thing to try first, in our view, but treat it as an experiment rather than a documented fix.

Why your sheet is slow long before the cell limit

We read Google's April 2026 post as a signal about where the pain actually starts. Its performance claims are measured on spreadsheets with 1M+ cells: opening and starting work "up to 30% faster," filtering "up to 60% faster," and conditional formatting "up to 60% faster." Those are Google's own figures, with no baseline or method published, and they tell you that one million cells, five percent of the cap, is already the size at which Google considers a sheet large.

The limits that stop automated work are lower and much harder. Google documents Apps Script quotas (checked October 23, 2026) of six minutes per execution for both consumer and Workspace accounts, 30 seconds per custom function execution, 30 simultaneous executions per user, and a trigger budget of 90 minutes a day on consumer accounts versus six hours a day on Workspace. A nightly script that rebuilds a 2-million-cell sheet does not fail because of the cell limit; it fails at minute six.

Sync tools and dashboards hit a different wall. The Sheets API allows 60 read or write requests per minute per user and 300 per minute per project, with no daily cap as long as the per-minute quotas hold. Every connector writing into your file shares that budget, so an import job can slow down or start failing as you add integrations, even when the file size barely moves.

Connected Sheets has tighter caps than the spreadsheet does

The standard advice for outgrowing a spreadsheet is to leave the data in a warehouse and query it from Sheets with Connected Sheets, which uses BigQuery as its data source. That works, but it is not unlimited, and the documented caps are well under the cell limit. Google's Drive help page lists "up to 200k rows for pivot tables in Connected Sheets" and "up to 500k rows or 5m cells for extracts in Connected Sheets" (checked October 23, 2026).

In other words, an extract tops out at a quarter of what a plain spreadsheet can hold. We would use Connected Sheets when the dataset is far too big for a sheet and nobody needs more than aggregates, and avoid it when the workflow depends on having every row in the grid where people can edit it. If you go this route, the warehouse side runs on Google Cloud Platform, and the cost of the queries behind each refresh sits with that account, not with Workspace.

How to fix the cell limit error without moving tools

If an import fails with an error naming the cell limit, our first assumption is that the fix is removing empty or duplicated grid rather than deleting data. These are the steps we would take, in order, before considering a migration:

1. Delete unused rows and columns on every tab, including the tabs nobody opens. Google does not document how blank grid is counted, so measure the cell count before and after.
2. Check whether the file is near 100MB rather than near 20 million cells. If cells are long text, size is your binding limit and trimming the grid will not help much.
3. Split by time, not by topic. One file per quarter with an identical layout keeps formulas portable; one file per team does not.
4. Replace imported raw rows with a summary. If nobody reads the row-level data in the sheet, it should not be in the sheet.
5. Re-test an import that failed before September 2026. Google says the new cap applies to existing and imported files, so the same job may now fit.
6. If a script is the thing failing, time it. Under six minutes per run is a quota problem you can solve by batching; over it is a tooling problem.

When to stop using a spreadsheet for this

Our rule of thumb: a spreadsheet stops being the right container when more than one person needs to write to the same rows at once, when the row-level data is never read by a human, or when a scheduled script is the only thing keeping it current. None of those are cell-count problems, and none of them get better at 20 million.

Two caveats worth knowing before you shop. Spreadsheet-shaped databases often cap rows far lower than Sheets does: Grist lists a free plan with up to 5,000 records per document, then Pro at $10 per user per month, or $8 billed annually, and a free, self-hostable Community edition (pricing page checked October 23, 2026). And tools that promise spreadsheet-plus-code often do not publish a data ceiling at all: Quadratic lists Free, Pro at $18 per user per month billed annually and Business at $36, with no row or file-size limit stated on that page (checked October 23, 2026). Ask for the number in writing before you migrate.

As of October 2026, Upfynd lists 135 products in Spreadsheets & Documents, which is where to compare these if a sheet is no longer the right home for the data. If the answer looks more like a database with a grid on top, our write-up of what to move when a spreadsheet platform changes hands covers the questions to ask about export and lock-in first.

Start with the two numbers on your own file: the cell count and the megabytes. One of them is close to a published limit, and which one it is decides whether your next hour goes into trimming grid or into picking a new tool. Makers building in this space can list a product on Upfynd for free.