Miro sold for 90% less than its peak, and it was profitable the whole time

The money in your tooling stack is moving fast, but not all in the same direction. Miro just sold for a fraction of its old price while startups you have barely heard of raise billions in a single week. Same market, opposite outcomes. Let me catch you up on what that split means for the tools your team runs on.

The deep cut

• Buyers pay for revenue now, not the story. Bending Spoons got Miro for $1.36B, 90 percent under its 2022 peak.
• Infrastructure that keeps agents alive gets funded first. Temporal doubled to $12.55B by keeping AI workflows from crashing.
• Concentration in two labs is a bet, not a plan. Deven Parekh warns funds putting 35 to 40 percent into OpenAI or Anthropic.

The whiteboard that got marked down

Miro was worth $17.5 billion in late 2021. Bending Spoons is buying it for $1.36 billion, a 90 percent haircut. And here is the strange part: Miro is fine. It has $600 million in annual recurring revenue, is profitable, and holds $435 million in net cash. It did not need to sell.

The number that dropped was the story, not the business. In 2021, buyers paid as if Miro would become a software giant. It grew into a solid, slower company instead. Bending Spoons keeps finding these deals. It bought Airtable, once valued over $11 billion, for $1.28 billion the month before.

For you, the read is simple. A tool being cheap to acquire does not mean it is dying. But it does mean the market stopped betting your whiteboard becomes a platform. Plan around the product you have today, not the roadmap the old valuation promised.

The plumbing gets the money

While collaboration tools reset, the boring layer under AI is on fire. Temporal raised $550 million at a $12.55 billion valuation, more than double its price from earlier the same year. Its revenue run rate passed $250 million, up over 200 percent year over year.

Temporal does not do the flashy AI part. It keeps multi-step agents from crashing when a step fails. CEO Samar Abbas put it plainly: "every additional step creates another place to fail." OpenAI, NVIDIA, and JPMorgan Chase run on it.

That is the pattern worth noting. Money is flowing to the tools that make agents reliable in production, not the ones that demo well. If your roadmap has "agents" on it, the question your review should ask is who catches the failures, not who writes the prompt.

Buy the users, skip the build

Productivity platforms are stitching AI into the middle of the workday, and they are buying their way in. Superhuman acquired Fathom, a notetaker with over 400,000 monthly users, rather than finish its own. CEO Shishir Mehrotra tested one internally and learned "how hard it is to create a notetaker."

Fathom's CEO Richard White was blunt about why he sold: Superhuman's 40 million users. Building distribution alone would have taken years. The meeting notes become the trigger, so AI can draft the follow-up email and schedule the next call on its own.

The lesson for your build-versus-buy calls: a deep-looking feature can hide real engineering, and reach can beat a clean codebase. If a tool your team wants gets swallowed like this, expect its focus to shift toward the parent's stack.

Everyone piling into two doors

The loudest capital is stacking behind a handful of names. Nvidia is reportedly in talks to bankroll Anthropic's IPO, a raise as large as $100 billion at a $2 trillion valuation. OpenAI and Anthropic took roughly half of all VC dollars in the first half of the year.

Deven Parekh at Insight Partners sees the danger up close. He named two funds raising their entire fund in a month, pitching that 35 to 40 percent goes into one of those labs. His warning: "you can't compound $40 billion at 50% every two months for two years without becoming the world economy."

Defense tech shows the same heat. Mach Industries doubled to $3.7 billion in three months, run by a 22-year-old who dropped out of MIT. Nscale, an AI data center startup, added former OpenAI number-two Fidji Simo to its board before an IPO. The capital is real. So is the risk that a vendor you depend on is priced on hope, not results.

Three questions for your team

• Which of our core tools were priced on a 2021 story, and what breaks if the vendor gets acquired at a discount like Miro?
• For anything agentic on our roadmap, who owns reliability when a step fails, and are we betting on plumbing like Temporal or just the model?
• How exposed are we to vendors funded almost entirely on frontier-lab hype, and what is our fallback if one of them resets like SaaS did?