Ramtin Naimi on the State of Venture Investing - [Invest Like the Best, EP.499]

Why today's AI venture boom may be healthier than 2021

AI venture moves as fast as 2021, but Naimi thinks it rests on firmer ground because companies reach category leadership in two years and outcomes now run to trillions. That expectation explains messy revenue, 100x multiples and mega funds burning through capital early. With crowding into the largest names now consensus, he sees lasting returns in great founders pursuing non-obvious ideas.

Takeaways

• Faster paths to scale justify today's prices. The pace resembles 2021, when $8m SaaS companies got billion-dollar valuations and reaching $100m ARR took about seven years. Now products find fit faster, some grow 10 to 20 times a year, and investors expect category leaders within 18 to 24 months.
• Hot-company revenue is rarely clean SaaS. Many have low or negative margins, customer concentration, one annualized month or usage-based revenue. Data labeling runs 15 to 20 percent margins and model-dependent apps pass margin to providers, though some portfolio companies are moving from negative 15 to positive 35 percent.
• Labor replacement is the thesis behind 100x multiples. Some vertical SaaS companies trade at 100 times revenue against roughly 13 times for public SaaS. Naimi says that only works if AI captures labor budgets such as law firm paralegal spending, which is plausible while companies still grow five to ten times at scale.
• Mega funds are deploying three times faster. Funds plan for 18 to 24 month deployment, but mega platforms raised two to three times more and underestimated what holding ownership costs. One fund given $3bn nine months ago now seeks $5bn, and some term sheets commit money that arrives months later.
• Concentrated LPs are short on liquidity. Some LPs hold 30 to 50 percent of assets in SpaceX, Anthropic, OpenAI and Databricks, leaving little room elsewhere. Tranched rounds give a few firms preferential first-tranche pricing while others pay far more, and term sheets for one company can differ fourfold.
• Non-obvious founders are where alpha remains. Obvious ideas draw 15 term sheets and steep prices, while two or three keep pricing sane. Conduit, a brain-computer interface company seeded nine months earlier, hit 80 percent accuracy on 30,000 hours of data and drew 13 term sheets. Backing the largest companies is now consensus.
• Compute financing is outgrowing pure equity. Perhaps 80 or 90 NeoLabs have raised $1bn to $5bn each, one spending $1bn on a single training run. Naimi doubts equity is the best fuel for compute, cites GC's customer value fund, and expects AI debt costs to rise.

How Abstract has changed

• Moved beyond seed to lead Series B rounds in companies such as Socket and Doctronic, plus growth investing in Crusoe.