OpenAI solves its internal crisis, snaps up Windsurf
Windsurf’s early angel investors are probably now prime candidates for private poker invites.
Welcome to Cautious Optimism, a newsletter on tech, business, and power.
Oddly enough, the news is coming up Rhode Island? After Providence-based Utilidata raised a $60 million, former governor and later Secretary of Commerce Gina Raimundo confirmed that she is considering running for president? Shoutout Gina, she was better than our current premier, but isn’t it a bit early?
On topic, there’s a huge run of news. Let’s go! — Alex
📈 Trending Up: Completely shocking developments … Databricks getting busy with the checkbook … political corruption … future trade deals? … one-man economic rule … the Baltimore Banner … spam … Wonder … EU drone prowess …
AI model leaderboard: Gemini 2.5 Pro, o3, 4o, Grok 3 Preview, GPT 4.5 Preview
📉 Trending Down: Oil prices … hope … anti-interdependence … the global birth rate … American soft power? … burying the hatchet … driving your own boat … Palantir shares …
📚 Quote of the Day: “Perhaps a sign of the times: I have more VCs asking me for intros to LPs than founders asking for intros to VCs.” — Ed Suh, Alpine Ventures
I don’t know where to file this, but it’s incredible to watch so many of the current generation of startups grow at stupendous pace. Lovable crossed $40 million ARR recently, continuing its torrid expansion, and proving that you can build something remarkable in Europe.
But there are other names, too. And not just the Cursors and Lovables of the world. A startup called Default is smacking the market in the mouth, too. The company’s offering — lead routing, scheduling, enrichment, and CRM sync — is performing incredibly well. Default’s CEO wrote on Twitter yesterday that his company has “grown ARR 2X since the start of the year and want to grow 4X AGAIN this year.”
That’s exactly the sort of nutso growth you want to see from startups. It’s just rough to contemplate how capital-starved leftover nine, and ten-figure startups from the early 2020s are going to wind up when the next cycle of startups aren’t saddled with expensive cap tables and are flashing salivation-level growth.
DoorDash goes shopping
Yesterday after the bell, consumer meal delivery service DoorDash announced its Q1 2025 results. With revenue of $3.03 billion (+21%), net income of $193 million, and adjusted EBITDA of $590 million it’s certainly a healthy company.
But one that missed expectations. DoorDash’s adjusted earnings per share was perfectly peachy, but its trailing revenues was about $60 million short. Investors hated that, selling its shares in after-hours trading.
But DoorDash has a plan to get things back up and growing at a market-approved pace for its $87 billion valuation it set before reporting its Q1 results.
That plan? Buy stuff:
DoorDash is buying Deliveroo for £2.9 billion, worth about $3.88 billion at current exchange rates. Why buy a slower growing asset? International market share. Deliveroo’s paltry 8% revenue growth in the first quarter of the year won’t ignite DoorDash’s overall growth rate, but it will instantly gift it market share in the UK, Ireland, parts of Europe, and some Asian markets.
In contrast, pre-purchase DoorDash was only present in the USA, Canada, Australia, and New Zealand.
In simple terms, DoorDash is spending about 4.5% of its market cap buying entry into a huge slice of the global market in a single swoop. This is why the equity premium we’re dancing with losing here in the States matters — if you get extra credit, you can use that spare change to do deals. (Deliveroo was once a venture-backed startup, but as it’s also now a public concern we’re not interrogating its historical cap table today.)
The Deliveroo buy is also nice in that it should prove adjusted EBITDA positive from day one, with the company previously anticipating the loose profit metric to turn in a result of “£170-190 million” this year. The deal is nto predicated on an EBITDA multiples basis, but it’s still nice to see.
DoorDash is also buying SevenRooms, a restaurant software company based in New York that had raised around $70 million while private, including a large $50 million check back in 2020. Why would DoorDash buy SevenRooms? That’s simple: Deeper integration into restaurant operations. SevenRooms helps customers with everything from table management to online ordering — DoorDash is about to delve even more deeply into how many food spots operate day to day.
And high-margin software revenue is always welcome, yeah?
Notably Olo isn’t doing that well. Olo offers a competing — if not perfectly overlapping — feature set as SevenRooms. It went public a few years ago, but has struggled post-listing. From a $25 per-share IPO price, it’s now worth about $7.30 per share. However, that fact didn’t stop DoorDash for kicking up $1.2 billion in cash to snap up the technology startup.
Will the two buys solve DoorDash’s near-term growth woes? No. But they could, separately, help DoorDash hunt for growth across a large cut of the world’s geography, and burrow more deeply into chefs’ minds. Worth it, I think, and good to see at least some venture capital recycle back to their LPs in the case of SevenRooms.
OpenAI solves its internal crisis, snaps up Windsurf
Speaking of which, Windsurf is selling to OpenAI as expected. For $3 billion, as expected. The company was last valued at $1.25 billion in August of 2024, meaning that General Catalyst, KP and Greenoaks just got paid a lot for very little — venture can be quite lucrative! If you scroll far enough back, Windsurf’s early angel investors are probably now prime candidates for private poker invites.
News that the AI model giant is buying the viral AI-infused developer tool comes after two other headlines of note:
Maker of AI ‘vibe coding’ app Cursor hits $9bn valuation (our note from yesterday here)
OpenAI reverses course, says its nonprofit will remain in control of its business operations (though Microsoft is worried)
How are the stories connected? Simple:
OpenAI wanted to buy Cursor, failed, and then decided to buy Windsurf per reporting.
Cursor turning down OpenAI’s cash makes sense, as it had a huge funding round on deck and could stay indie sans any capital concerns. Smart.
The Windsurf deal comes right after OpenAI announced that it had scrapped contentious plans to shed its non-profit umbrella. Now that it has found a new path forward that should feature fewer lawsuits, it can close deals.
Hence, Windsurf’s new home.
I remain by bullish on OpenAI buying Windsurf. By snapping up the AI-predicated developer service OpenAI is buying itself a hot slice of developer mindshare, a cut of the value generated at the application layer above the model layer, and a huge amount of data.
Recall that Windsurf (under its former nomenclature Codeium) used enough Anthropic API calls to earn a public profile from the latter company. Now, OpenAI will get all those API queries. More data means more gooder AI models. Which OpenAI wants, desperately, as whichever AI company earns the most developer-focused applications will be worth — no matter what — a lot.
And as with the SevenRooms deals, OpenAI buying a hot startup means returning either cash (welcome) or OpenAI shares (perhaps even more welcome) to LPs. Venture capitalists, thank SoftBank’s work to reprice OpenAI up by a factor of more than three for this one.
Onward!