The Repricing, Beyond Airtable

I wrote that Airtable selling at 2.7x ARR is the whole lesson. A few readers pushed back, correctly: one company is an anecdote. A repricing thesis needs comparables, or it is just a single deal with a story wrapped around it. So here are the comparables, and reading them together shows something the Airtable deal alone did not: there are actually two repricings happening, and they are easy to confuse.

The short version

Airtable at roughly 2.7x ARR is one instance of a pattern, not the pattern itself. Chegg lost about 99% of its 2021 value as ChatGPT and Google AI Overviews ate its answer library. Prosus wrote Stack Overflow down from a $1.8 billion purchase toward roughly $564 million as AI coding tools collapsed its question volume by more than three quarters. Squarespace went private at about $7.2 billion and Smartsheet at $8.4 billion, both at cash-flow multiples the public market no longer paid. Put together, these split cleanly into two repricings. The first is AI destroying the option on moats denominated in accumulated human effort or content: Airtable, Chegg, Stack Overflow. The second is zero-interest-rate growth assets repricing to cash-flow multiples through private-equity buyouts: Squarespace, Smartsheet. The severe one is AI, and it is severe because the thing that changed is the value of the moat, not the price of patience. To know which is coming for you, ask what your moat is denominated in.

The anchor, restated

Airtable is being acquired by Bending Spoons at a $1.285 billion enterprise value against roughly $480 million in ARR growing over 20%, down from a 2021 raise at about $11.7 billion. Roughly 2.7x revenue. The business did not die. The 2021 valuation thesis did, and the mechanism was specific: Airtable's moat was the effort customers spent modeling their workflows into bases and automations, and effort-denominated switching costs deflate at exactly the rate that agents make effort cheap.

Hold onto that mechanism, because it is the thread that connects the two clearest comparables.

The effort-and-content moats: Chegg and Stack Overflow

Chegg is the cleanest AI repricing in the market. At its 2021 peak it was worth about $14.5 billion. It has since lost roughly 99% of that. Revenue fell about 30% year over year in the first quarter of 2025 and about 49% by the fourth, with the subscriber base down about a third. What did Chegg actually sell? A decade of accumulated, curated homework answers, a library built by human labor over years. ChatGPT reproduced the core of that value for free, and Chegg's own CEO told investors that Google's AI Overviews were as material to the decline as ChatGPT itself. The moat was a stock of past human effort, and a model that regenerates that effort on demand does not dent the moat, it dissolves it.

Stack Overflow is the same story in developer tools. Prosus bought it in 2021 for about $1.8 billion, betting on the world's largest accumulated store of programming questions and answers. By the start of fiscal 2026 Prosus had written the unit down toward roughly $564 million, a cut of nearly 60% on top of earlier impairments, after writing off well over a billion dollars across three years. The question volume tells you why: new questions fell around 78% year over year by late 2025, and the post count was down more than 90% from its 2020 peak, as developers moved to ChatGPT and Copilot. The moat was millions of human-written answers. Once a model trained on those answers could just produce the next one, the reason to visit the site, and to add to it, evaporated.

Notice the shared shape with Airtable. Three different products, three different categories, one identical mechanism: a moat denominated in accumulated human effort or content, reproduced cheaply by a model, repriced hard. This is the severe repricing. It is not a discount on a healthy business. It is the market marking down the value of a moat that stopped being one.

The rate repricings: Squarespace and Smartsheet

The other two comparables look similar on the surface and are a different animal underneath.

Squarespace went private with Permira in October 2024 at about $7.2 billion, against roughly $1.2 billion in revenue growing around 20%, so on the order of 6x. Smartsheet went private with Blackstone and Vista at $8.4 billion in a deal that closed in early 2025, against about $1.1 billion in ARR growing 17%, so roughly 7.6x. Both are real, healthy, growing businesses. Both got taken private by firms that pay for durable cash flows.

These are not AI repricings, and it would be lazy to file them as such. Six to eight times revenue is not a collapse; it is what a solid, growing software business is worth when the market stops paying a growth-story premium on top. This is the zero-interest-rate hangover: the 2021 multiples were a rate artifact, and as money got expensive, the market stopped paying for the story and started paying for the cash flow. Private equity showed up precisely because the public multiple compressed to a level where a leveraged cash-flow buyer is the most rational owner. The business is fine. The story just got expensive to believe.

The reason to put these next to Chegg and Stack Overflow is not that they are the same. It is that they are different, and telling them apart is the entire skill.

Two curves, and why confusing them is dangerous

Here is the distinction stated plainly, because your response should depend entirely on which one you are facing.

A rate repricing lowers your multiple because capital got more expensive. The underlying value the customer pays for is intact. You survive it by waiting, by generating cash, and by not needing the market to fund a story you can no longer sell at 2021 prices. Squarespace and Smartsheet are still good businesses doing fine under new owners. Patience works.

An AI repricing lowers your multiple because the thing the customer was paying for got cheaper to reproduce. The value of the moat itself fell. You do not wait this one out, because time does not restore the moat; it erodes it further. Airtable's mistake, in hindsight, was reading an AI repricing as a rate repricing and assuming the cash cushion bought time, when the thing that needed fixing was what the product was worth, not when the market would come back.

The danger is that from inside the company, early, the two look identical. Your multiple is down. Revenue is still growing. Cash is still coming in. A leadership team that tells itself "this is the rate cycle, we just have to wait for it to turn" when it is actually an AI repricing will spend its runway waiting for a recovery that structurally cannot arrive, because the recovery would require the moat to become valuable again, and the moat is what changed.

How to tell which one is coming for you

The test is a single question: what is your moat denominated in?

If it is denominated in accumulated human effort or accumulated human content that a model can now regenerate, the effort a user spent, a library you built by hand, a corpus of contributed answers, you are exposed to the AI repricing, the severe one. That is Airtable, Chegg, Stack Overflow.

If it is denominated in data gravity, regulatory approval, being the system of record that other systems read from, a difficult integration, or trust earned in production over years, you are more exposed to the rate repricing, which is survivable, and less exposed to the AI one. That is closer to the durable side of the line I keep coming back to.

Most companies are a blend, which is why the exercise is to split your value into two piles, the option and the cash flow, and price each honestly. The option is the part the market paid a premium for on the bet you would become something bigger. AI is what reprices the option. The cash flow is what a Bending Spoons or a Vista pays for. Rates are what reprice the cash flow. Know which pile is bigger, and know which repricing each pile is exposed to, before someone else does the exercise for you and puts a number on it.

Where this argument is weakest

I would not trust this piece if I did not steelman it. You can tell most of this story with interest rates alone: 2021 was a bubble, the correction started in 2022 well before agents mattered, and every one of these companies would have repriced somewhat without AI. Fair. And the private-equity deals really are rate stories, which is why I put them in their own bucket instead of forcing them into the AI narrative.

But Chegg and Stack Overflow are hard to explain on rates alone. A rate repricing does not cut a company's core usage by 78% or its revenue in half. Those are demand collapses in the specific product, caused by a specific substitute, and the substitute is AI. You can argue about how much of Airtable's markdown was rates versus AI. You cannot argue that Chegg's students left for a spreadsheet of interest-rate assumptions. They left for ChatGPT. That is the tell that at least one of these two curves is real, and it is the one you cannot wait out.

Try this week

Take your own company and do the split. Put every source of value into two columns: the option, the part investors paid a premium for on the promise of what you might become, and the cash flow, the part a disciplined buyer would pay for today. Then, for each item, mark whether it is exposed to the AI repricing (denominated in reproducible human effort or content) or the rate repricing (denominated in something a model cannot regenerate).

If your option pile is large and most of it is AI-exposed, you have found your version of the Airtable problem while you still have time to move bets to the other column. That is the entire value of reading someone else's repricing. It is a warning you get before the market writes it on your own multiple.

Sources: Airtable and Bending Spoons figures per my earlier post and Bending Spoons investor materials (August 2026). Chegg reports 99% stock decline and 45% layoffs, Forbes, October 2025; Chegg Q4 and full-year 2025 results, Chegg IR. Prosus cuts Stack Overflow valuation, Moneyweb; Stack Overflow question volume down 78%, Techzine. Permira completes $7.2B Squarespace acquisition, SiliconANGLE. Blackstone and Vista to acquire Smartsheet for $8.4B, Blackstone.