𝐓𝐡𝐞 𝐫𝐞𝐚𝐬𝐨𝐧𝐚𝐛𝐥𝐞 𝐟𝐚𝐜𝐞 𝐨𝐟 𝐀𝐈 𝐢𝐧𝐯𝐞𝐬𝐭𝐦𝐞𝐧𝐭𝐬

SSI closed $2bn at $32bn valuation (pre or post – who cares?); and Thinking Machines Labs is rumoured to raise also $2bn “on at least $10bn”, up from the rumours last month when the company was said to raise $1bn at a $9bn valuation. This is of course on top of the new round for OpenAI ($300bn on $3.7bn of revenues, a rather rich 81x P/R for a company which is 4 years from break-even).

For those who are old enough, this brings us back to the “New Economy” of the late 1990s (you were a fool if you didn’t get it - the world was going to be different this time), when investors were rushing to finance “eyeballs” – with the outcome that we remember: by the time the markets settle down in 2002, the Nasdaq was down 80% from its 2000 peak and $5 trillion had been wiped out.

As usual, we tend to overestimate the immediate impact of innovation, and underestimate the long-term effect (AI will indeed change the world).

What is however different this time is that there is whole corner of the market taking advantage of the GenAI boom by developing useful applications around third-party “foundational models”.

⬆️ The good news is that “many of these startups are reaching up to $200m [in ARR] in less than a couple of years” (according to Dealroom). This illustrates how much pent-up demand there is among corporates to use AI to drive productivity through many tasks which are quite human-intensive, tedious, and have stubbornly resisted automation.

⬇️ The bad news is that these applications are based around third-party IP (a traditional no-no of investment). And investment in those kinds of companies is still rather expensive: Anysphere (producer of Cursor) raised $105m at a $2.5bn valuation in January (25x P/R).

No wonder investors are already exploring new ways to apply AI, for instance using it to power humanoid robots and make AI live among humans.
‍

🔗 Sources:

https://www.businessinsider.com/mira-murati-big-tech-put-thinking-machines-lab-venture-capital-2025-4

https://www.saastr.com/bloomberg-openai-to-hit-12-7-billion-this-year-but-wont-be-profitable-until-125-billion/

https://en.wikipedia.org/wiki/Dot-com_bubble

https://www.technologyreview.com/2024/04/11/1090718/household-robots-ai-data-robotics/

https://dealroom.co/uploaded/2025/03/2025_Dealroom-Deeptech-Report.pdf

‍

‍