AI startups are turning their revenue into recruiting bait

This is an excerpt of Sources by Alex Heath, a newsletter about AI and the tech industry, syndicated just for noti.group subscribers once a week.

A new trend has quickly emerged for AI startups that want to stand out from the rest: brag about revenue.

Take Sierra, Bret Taylor and Clay Bavor’s AI customer support firm that was recently valued at $10 billion. On paper, you’d think Sierra could have its pick of just about anyone who wants to work in AI — both co-founders are well-known names in Silicon Valley, Taylor is also the chairman of OpenAI, and Sierra has raised more than $600 million in less than two years.

But even Sierra feels the need to put a giant number on the board to compete for talent. Taylor told me on Thursday that the company has reached $100 million in annual recurring revenue, up from about $20 million this time last year. Unlike many AI startups now flexing their ARR, Sierra books its revenue through upfront contracts. The company says its customer support agents have already been used by hundreds of millions of people, many of whom wouldn’t know they’re interacting with an AI to process a return or troubleshoot a bug. Its customers include SoFi, Wayfair, Ramp, Rocket Mortgage, and hundreds more.

Taylor spent a good chunk of our conversation explaining why he thinks Sierra’s $100 million means more than the typical AI startup ARR number. Sierra follows the same model used by public enterprise software companies like Salesforce and ServiceNow. It signs at least 12-month, often multi-year contracts, bills annually up front, and gives customers 30 days to pay after signing.

By contrast, many AI startups, especially those with more consumer-ish products or usage-based pricing, reach a public ARR figure by multiplying a good month’s revenue by 12. If growth slows or users churn, that ARR evaporates just as quickly. Taylor’s argument is that Sierra’s number looks more like what public-market investors care about: contracted revenue that’s harder to walk away from.

He wouldn’t name names, but Taylor made it clear that Sierra is trying to separate itself from AI startups that tout ARR off a leaky base of pay-as-you-go users. In those cases, an ARR figure can mask high churn or a product that’s riding a hype wave or temporarily juicing sign-ups with incentives.

“I think AI is a category where it’s relatively easy to make a demo and sort of win a popularity contest on social media,” he said. “But creating a durable revenue stream, especially from serving the Fortune 1000 and regulated industries, is incredibly challenging. I think a lot of people want to work for the leader in the category.”

That milestone puts Sierra among the fastest-growing AI companies, though it’s hard to do apples-to-apples comparisons in a world where private companies can define their own metrics.

“There is no official leaderboard, but we believe we’re fairly far ahead of the other companies in our category,” Taylor said. “We want to make sure recruits know that and potential customers know that because I think it is a signal that we’re doing something right and the product is high quality and our customers like working with us and have invested deeply in us.”

That’s the subtext here: ARR has quietly become one of the most important recruiting signals in the AI startup market. Startups used to hype funding rounds or valuation. They still do, but some are also sharing revenue numbers that, in a different era, would have stayed buried.

Loveable CEO Anton Osika recently shared at a conference that the company doubled its ARR to $200 million in four months, and Cursor this month announced that it past $1 billion in annualized revenue. For a recruit, these revenue stats are meant to signal that a startup isn’t just riding a hype cycle but has customers and real traction.

Taylor’s mental model for what’s happening now is the late ’90s. “I think the closest analog to this AI wave is the dot-com boom or bubble, depending on your level of pessimism,” he said. Back then, he explained, everyone knew e-commerce was going to be big, but there was a massive difference between working at Buy.com and Amazon.

“As a candidate, you want to work for the company that’s going to end up being the leader,” Taylor said. His pitch is that Sierra is on that path in AI customer support: a company with real contracts from big, often regulated customers.

His hiring plans reflect that ambition. Sierra has roughly 300 employees today. Taylor wouldn’t commit to a precise headcount target for next year, but he acknowledged that “doubling or more” is in scope, driven mostly by international expansion and customer-facing roles.

His real estate moves are another signal. Taylor confirmed that Sierra has signed a lease for roughly 300,000 square feet of office space in San Francisco’s China Basin neighborhood, a block from Oracle Park. The company will vacate its current building and nearly triple its footprint, marking the city’s largest office lease since OpenAI took over Old Navy’s former headquarters near the Chase Center last year.

Taylor is also already thinking about what happens when today’s crop of AI startups matures. He expects the industry to follow a familiar pattern: an early “best-of-breed” phase where specialist tools grow quickly, followed by a platform-consolidation wave. “Reductively, you either earn the right to consolidate or you get consolidated.” Sierra isn’t out shopping for acquisitions yet, he said, but it wants to be in the former camp when that moment comes.

All of this explains why a company like Sierra — backed by blue-chip investors, run by a former Salesforce co-CEO who now chairs OpenAI — is out there trumpeting its early revenue. The AI agent market for customer support is already crowded, with upstarts like Decagon and incumbents like Intercom and Salesforce vying for the same budgets. In that world, a startup announcing nine figures of ARR is a signal of strength aimed at the small pool of people who can work anywhere.