Yesterday, we watched $1T of value disappear from public markets. Today, we try to quantify the impact on private markets.
According to PitchBook, in 2024 ~40% of private capital invested in AI was in the foundational model layer. “Pure plays” (i.e. little/no alternative revenue streams) are OpenAI, Anthropic, xAI , Cohere, Mistral AI.
These 5 companies raised $66B in the past 3yrs, vs. ~$200B across all private AI.
If DeepSeek AI is the real deal and can match model performance with a fraction of the capex, Model 1.0 businesses need to adapt rapidly to compete. Some may not succeed.
In that case, where is the value at risk?
Founder and employee paper net worth will come down, materially. One silver lining: most Model 1.0 companies have done a secondary liquidity event in the past year, allowing shareholders to de-risk. Hopefully, some gains were locked in.
What about the VCs?
Is this an existential event for name-brand VC funds?
Probably not. At least looking at risk as a % of capital base.
"Venture” investing is as much a game for VCs as it is for hedge funds, traditional asset managers, and corporates. Private AI investment as a % of assets gets small when looking at some of the more active sovereigns and corporates.
No doubt traditional VC funds that have been leading AI rounds will take a hit. But we looked at name-brand VC exposure to the most well-funded private AI companies and didn’t find overwhelming concentration risk.
Our analysis:
- Start with $66B invested in private Model 1.0 companies over the past 3 years.
- Reduce by 40%, which is the collective private investment from Microsoft ($13B), Amazon ($8B), Google ($3B), and Nvidia ($1B) in Model 1.0 companies over the same period.
- Sovereigns and non-traditional VCs (hedge funds, asset managers) likely gobbled up another $20B. (my estimate, based on working with these groups)
- That leaves approximately $20B for traditional VCs.
- We estimate the 5 most active traditional VCs (Sequoia Capital, Andreessen Horowitz, Lightspeed, Thrive Capital, Index Ventures) took about $10B of this.
These investors only led rounds in 1 Model 1.0 company each. Let's keep it simple and estimate ~$2B invested across the 5.
So the question is across these 5 investors, who has the capital base and reputation to absorb a $2B hit? I’d bet on most of these groups living.
Are there smaller funds that invested heavily in Model 1.0 companies and are at serious risk? Yes. But it seems traditional VCs getting outgunned on some of these competitive fundraises over the past few years may have been a saving grace.
At Caplight, we’ll be tracking the potential fallout in secondary market pricing of Model 1.0 companies, and sharing updates.
Ping us if you’d like to follow the action in real time with a subscription. :)
**Past performance is not indicative of future returns.
