Scooped at a Discount: What Airtable Says About SaaS
There are still a lot of 2021 marks hanging around that nobody would pay today. Airtable is one of the first big examples of where that gap is showing up.
Airtable sold in August at a $1.29bn EV, or 2.7x ARR, about 81% below its 2021 peak. So yeah, SaaS is cooked, AI is eating software, and you should get out while you can. But that probably overstates it. Software isn’t all moving the same way, and Airtable doesn’t tell you what happens to the rest of SaaS.
The multiples tell the story
Tomasz Tunguz’s latest post on public software shows the leaders in each category trading well above the median. CrowdStrike is at 3.9x the security median, Cloudflare at 3.4x infrastructure, and Shopify at 8.1x commerce.
Source: Tomasz Tunguz, “A Winner in Every Category”
For these companies, the premium seems to reflect an AI advantage in the business, rather than just stronger growth. Every AI agent an enterprise deploys creates another identity and attack surface that needs to be secured, which plays directly into CrowdStrike’s platform. As AI agents generate more web and API traffic, Cloudflare has more machine activity flowing through the infrastructure it routes and secures.
None of these are even the fastest growers in their category. CrowdStrike is growing 23% and trades at 34.4x forward revenue, while Rubrik, another security name, is growing 46% at 10.4x. Salesforce is growing just 11% and still trades at 4.0x forward revenue versus a 2.3x category median. So clearly there is more going on here than growth.
It’s not just the AI-native companies either. Thoma Bravo’s data shows public software businesses getting better even as the market pays less for them. FCF margins and growth are up, while EV/FCF multiples are down.
Source: Thoma Bravo
Public software has started to bounce too. IGV, the iShares software ETF, dropped to around $74 in April and was back above $100 by early August, up roughly 38% from the trough. It’s still below its 2025 high, but that doesn’t exactly look like a market giving up on software.
Source: Yahoo Finance, iShares Expanded Tech-Software Sector ETF (IGV), six-month performance.
Snowflake is a good example. The stock got hammered earlier this year because investors were sceptical that its AI push would translate into actual growth. Then its May results came in stronger than expected, with AI starting to show up as a tailwind. The stock jumped ~40% in a day and is now ~80% above the February lows. So the market clearly doesn’t think AI is bad news for all of software.
Databricks tells a similar story in private markets. Its valuation has risen from $62bn at the end of 2024 to $188bn this summer, while revenue run-rate passed $5.4bn in February with growth over 65%. So investors are still willing to pay up for software they think matters, but Airtable shows what happens when they don’t.
When everything was a winner
At near-zero rates, the bar for growth investing got very low. With the discount rate close to zero, investors were willing to pay huge multiples for growth and spend less time worrying about how durable it would be.
Airtable’s December 2021 round valued it at roughly 75x ARR. Databricks had raised at around 63x earlier that year. That’s a pretty wild comparison when you think about what the two businesses actually do. Databricks is buried much deeper in a customer’s data and AI stack, and Airtable is much easier to swap out. But back then, the market was basically willing to look past that difference.
Public software went through the same thing. Multiples climbed sharply into 2021, then rates went up and investors became much less willing to pay almost any price for growth.
Source: Bessemer Venture Partners, BVP Nasdaq Emerging Cloud Index. 2013–2026.
Higher rates changed that, and AI makes the question harder again. It’s not just about how quickly the business can grow over the next few years. You also have to think about what that growth is actually worth if AI starts putting pressure on margins, pricing or how long the growth can last. Under ZIRP, investors could afford to care a lot less about the difference between a point solution and something customers are locked into.
Airtable isn’t the only one. Vimeo, Eventbrite and Brex all ended up selling well below their peak valuations. Three of the four went to Bending Spoons, which has been rolling up established tech businesses where there’s still plenty to fix.
The useful distinction
AI is clearly disrupting software, just not evenly. Some categories are far more exposed than others. Thoma Bravo frames the distinction well.
Source: Thoma Bravo
AI is clearly going to hurt some bits of software more than others. A product buried inside a critical workflow, sitting on proprietary data or protected by regulation is a very different position from a general-purpose productivity tool. And product survival isn’t really the bar anyway. Seat pricing can weaken as agents do more work, while the AI functionality itself adds inference and compute costs. The software can still be useful while becoming a worse business.
Airtable sits to the left here. It is a broad, general-purpose tool, not deeply embedded in a regulated or mission-critical workflow, and the customer owns the underlying data. If you get it wrong, nothing blows up and no one’s going down the jobcentre. Replacing it is relatively manageable too.
Salesforce is the opposite. It sits right in the middle of critical workflows, data and integrations, so ripping it out is painful. The US Department of Veterans Affairs is a good example, recently signing a $1.6bn, three-year deal covering Salesforce’s CRM, AI, data and collaboration products across more than 150 medical centres. Nobody is vibe coding this.
For Salesforce, I think the change is in what people are actually paying for. That is seats now, but as agents start doing more of the work, that could shift toward usage, actions, or just access to the data and workflows underneath. This doesn’t cut Salesforce out as agents still need the customer record, the permissions, the workflows underneath them to actually work.
Intercom shows another route, rebuilding around AI, cannibalising legacy revenue and changing the pricing model with it. But Airtable’s AI push has been more retrofit than reinvention, with Hyperagent sitting outside the business Bending Spoons bought. Intercom was recently acquired by Salesforce for £3.6bn or ~9x ARR. Same backdrop, but a much deeper response from Intercom and a much better outcome.
Incumbents with distribution and embedded workflows have something real to build AI on. So, if it’s done well, AI can strengthen the moat rather than erode it.
The Airtable problem
Airtable is sticky in places. Some teams run it as a lightweight CRM, and once you’ve built a load of views, automations and workflows around it, ripping it out is a faff. But there’s still a bit of a skill to using Airtable well, people literally sold courses on it. AI chips away at that when you can just describe the workflow you want and have an agent build or run it for you. Knowing your way around the interface here matters a lot less.
AI also doesn’t need to rebuild Airtable feature for feature. If an agent can create or update the underlying data directly, some of the value in the no-code layer just disappears. And if it’s suddenly cheap to vibe code a decent internal tool, buying another general-purpose point solution sounds less compelling.
At 2.7x ARR, the market is pricing Airtable as a business with relatively little insulation. The decision to carve Hyperagent out before the sale probably suggests the same. The team also faces an opportunity-cost problem. Why spend several more years grinding to get the core product from 20% to 30% growth when the same people could be building something with more upside?
It’s not a great outcome, but $1.29bn only looks terrible because everyone remembers the $11.7bn peak. If you forget that for a second and it’s not as bad. The business was still growing 20%+ a year, and Bending Spoons isn’t exactly buying an old banger. There’s still a decent software business in there, just with quite a bit to fix.
And Airtable seems to have given them plenty to work with. It started as a PLG business and later built out a much heavier sales motion. RepVue has only around 30% of the sales team hitting quota, and that number has barely moved for three years. For Bending Spoons, the obvious move is to trim a sales organisation that isn’t really working, lean harder on PLG and improve the economics.
Source: Ryan Walsh / RepVue, RepVue
Cost is only part of the problem. RepVue reviews from current Airtable employees point to a business still working out where it fits in an AI-native market, with sellers leaving for AI-native competitors, top performers leaving and customers becoming harder to close. That is harder to fix than an oversized sales team.
Source: Ryan Walsh / RepVue, RepVue
Airtable was never getting back to its 2021 valuation, and it also happens to sit in a part of software that looks exposed to AI. That still doesn’t tell me SaaS is dead. IGV’s recovery, Snowflake’s rebound and Databricks’ recent rounds all point the other way, as do ServiceNow and CrowdStrike. The market is still paying up for embedded or differentiated products, particularly where AI helps rather than hurts.
Airtable tells us more about the 2021 private cohort. Plenty of those companies are still decent businesses, but they are carrying valuations from a market that disappeared years ago. AI will make that reckoning harsher for some, particularly in categories like no-code, where tools like Supabase make it easier to work directly with the underlying database layer rather than through a general-purpose interface.
Airtable has now done what a few others may eventually have to do. Take the hit, admit the 2021 mark is dead and move on. The talent can go and build something new. Airtable won’t be the last to do this. The question is who is next and whether Bending Spoons is already on the phone.
Sources:
A Winner in Every Category – Tomasz Tunguz