Airtable 2.7 倍 ARR 出售：AI 转型未重振增长

Bending Spoons is acquiring Airtable in an all-cash deal at a $1.285B enterprise value. With Airtable’s net cash, that’s roughly $2.25B of equity value. Expected to close by year end, subject to regulatory approval.

What they’re buying: approximately $480M ARR as of June 2026, growing over 20% year over year. Roughly 90% gross margins. Cash-flow positive since late 2024, throwing off over $100M annually. More than 500,000 organizations. 80% of the Fortune 100.

That’s 2.7x ARR for one of the best-known B2B companies of the last decade.

How Airtable Got Here

Founded in 2012 by Howie Liu, Andrew Ofstad and Emmett Nicholas, Airtable spent a decade building a category-defining product: a relational database wearing a spreadsheet interface, sold bottoms-up, expanded into enterprise. The “Lego of software.”

In December 2021 it raised $735M at an $11.7B post-money, at $187.28 per share. ARR was around $156M. A 75x round.

Then the cycle turned. Two rounds of layoffs totaling 491 people. The “Airtable is dead” meme cycle in 2023. Growth decelerated. The company did not fall apart: it cut to cash-flow positive by late 2024, held roughly half the capital it raised on the balance sheet, kept enterprise retention strong, and kept shipping.

Then it did the AI pivot, and did it for real:

• June 2025: Liu declared a “refounding” as an AI-native company, launching Omni, a conversational app builder, and Field Agents as the execution layer
• October 2025: acquired DeepSky and hired David Azose, formerly head of engineering for ChatGPT business products at OpenAI, as CTO
• January 2026: launched Superagent, the first standalone product in 13 years, followed by Hyperagent
• Throughout: AI bundled into every plan including free, HyperDB scaled to 100 million rows with direct connections to Snowflake, Databricks and Salesforce

This was not a company phoning in an AI press release. It was a full re-architecture led by a founder-CEO who went back to writing code.

The result was 20% growth.

Not 60%. Not re-acceleration. The refound stabilized a good business and made it defensible. It did not change the trajectory. And at 20% growth on $480M, you get priced like a cash-flow asset, not a compounder.

Why It Sold, and Why Now

Airtable had no forcing function. No debt, no burn, roughly $700M in the bank. It could have stayed private indefinitely.

What it did not have was a path back to $11.7B. The IPO window rewards scale with profitability, or real AI-driven re-acceleration. Airtable had the first and not the second. Filing an S-1 at 20% growth means going public well under the last private round, in public, forever.

Meanwhile the secondary market had been repricing it for three years:

Everyone treated the 66% haircut in January as capitulation. It was the halfway point. The clearing price landed 44% below that seven months later, and about 25% below where secondary desks were marking the stock three weeks ago.

Private marks are not prices. This is the price.

The competitive read is harder. The “build an app without writing code” position Airtable owned for a decade is being attacked from below by tools that do it better, faster, and without seat-based pricing. I build production apps in Replit every day with no engineering background. That was Airtable’s entire promise, and the promise got commoditized.

One Detail (That May Not Be a Big Deal): Hyperagent Isn’t in the Deal

Evan Armstrong at The Leverage pulled the transaction’s SEC filing and found a pre-signing reorganization. In the filing’s language, “assets and liabilities relating to the ‘Hyperagent’ business line were transferred” out of the company being sold, into a separate entity called Hyperagent Inc.

Hyperagent is the platform Airtable announced in February 2026 for building and running autonomous AI workers across business tools. Liu had been personally evangelizing it for months, including a Founding 500 program putting $10M of inference credits into the hands of 500 agent-first founders.

Don’t overread it. Bending Spoons cited the same ~$480M ARR after the carve-out was already done, so Hyperagent was contributing roughly nothing to revenue and the 2.7x holds either way. The extra value to shareholders is real but small against $2.25B, and could be zero.

What it does tell you is where Liu is going next, and that the agent bet was structured to survive the sale. Who owns Hyperagent Inc. isn’t disclosed. The reorg was done by the seller and its affiliates, which most likely means it went pro rata to existing shareholders rather than to founders alone. Whether Superagent went with it is also unknown, since only Hyperagent is named.

Everyone is a M&A expert today. But if your hot take on Bending Spoons acquiring Airtable doesn’t factor in that it spun out the AI unit, Hyperagent, it’s missing key pieces of the puzzle.

Could be where execs and top engineers go, and changes the math on their exit value👇 https://t.co/ikwvKmrFP5

— Alex Konrad (@alexrkonrad) August 4, 2026

Who Bought It, and What Happens Next

Bending Spoons is the Milan-based serial acquirer that IPO’d on Nasdaq on July 1, 2026 at $29 per share, roughly $18.4B, and closed its first day up nearly 40%. Airtable is its first deal since listing.

The portfolio: Evernote, WeTransfer, Meetup, StreamYard, Issuu, Brightcove, Vimeo, AOL, Eventbrite. Buy strong brands with stalled growth, rebuild the tech, cut hard, raise prices, hold forever. CEO Luca Ferrari says 90% of their code is now written by AI, and revenue per employee went from $1.12M in 2023 to $2.57M in 2025.

The pattern is documented:

• Evernote went from $100 per year to $249
• WeTransfer lost 75% of staff within weeks of close
• Vimeo had mass layoffs roughly two months after close
• Per SEC filings, $78.6M of reorganization expense in 2025 covering 1,830 people acquired from AOL, Eventbrite and Vimeo, with “only a few hundred” expected to remain by end of 2026
• Another $75.8M of reorganization expense in Q1 2026 alone

For Airtable’s 935 employees and its customers: expect significant headcount reduction within a quarter or two of close, price increases, a tighter free tier, and metered AI credits. That’s the playbook, not a prediction.

The model has a cost. As of March 31, 2026 Bending Spoons carried roughly $4.4B of debt against about $1.06B of shareholders’ equity, with $93.2M of interest expense in Q1 alone and full-year 2025 GAAP net income of essentially zero on $1.31B of revenue. The compounding story is real. So is the leverage.

Who Makes What, and Why It Explains the Deal

Airtable raised $1.4B across seven rounds. The history, with the implied return at $2.25B on a straight pro-rata basis, before liquidation preferences:

Post-money figures from the Series C on are reported. Seed, A and B are estimates.

Now layer in the preference stack, assuming standard 1x non-participating preferred, which is not publicly confirmed:

• Series F takes its money back. $735M in, $735M out. XN, Silver Lake, Salesforce Ventures, T. Rowe Price, Franklin Templeton, J.P. Morgan Growth Equity, MSD Capital. The preference is worth roughly $590M versus converting. Five years, zero return, negative after inflation.
• Series E and D take their money back. Greenoaks at $270M, Thrive at $185M. Both roughly 1x.
• Series C sits at the conversion line. Somewhere between 1x and 1.7x. CRV, Coatue, Benchmark.
• Seed, A and B convert and take the residual. On my estimates, Series A returns 15-25x and Seed 40-80x. CRV led the A and co-led the B, and is the one firm on this cap table that unambiguously won.

Roughly $1.29B goes off the top to preferred holders owning about 25% of the shares. That leaves $900M to $1B for all common, spread across roughly 75% of the shares. Call it $18 to $25 per share, against $187.28 in the last round.