Air Street Capital closes $232M Fund III, becoming the largest solo-GP fund in Europe

Summary

Nathan Benaich's Air Street Capital closed its third fund at $232 million, making it the largest solo-GP venture fund in Europe. The raise marks a dramatic scaling from the firm's first fund of $27 million in 2019 and $121 million for Fund II three years later, both closed before ChatGPT shifted investor appetite for AI.

Benaich has been investing in AI since 2013, when deep learning was still confined to labs and most investors dismissed it. He started Air Street as a solo decision-maker and has maintained that constraint even as the fund grew. Two colleagues handle talent and operations, plus back-office support, but Benaich makes all investment decisions, handles brand-building, founder outreach, and fundraising himself. He also authors the State of AI Report, an annual open-access document on research, industry, policy, and talent that draws contributions from 50+ researchers, PhD students, and policy specialists.

Fund sizing and strategy

With $27 million, a solo GP must choose between leading rounds and expressing conviction early, or running a large portfolio and making smaller bets across many opportunities. Benaich opted for the former. Fund III allows first checks up to $15 million and growth-stage rounds up to $25 million, deployed at high conviction across roughly 20 companies. A $27 million fund cannot be a lead investor today. A larger portfolio approach is viable but requires different skills—more network and SDR work, less original research and conviction-based positioning—which Benaich says doesn't suit his strengths.

European AI opportunity

Benaich rejects the idea that Europe needs its own Google. Past government efforts, including Quero and similar initiatives with hundreds of millions in funding, tried to build Europe-native search to better capture local culture, which he finds misguided when applied to a learning machine. The real sovereignty argument holds in defense and security, where geopolitical constraints are material. He invests in defense autonomous systems in Greece, Delhi and Alliance Industries, and notes that skeptics initially dismissed Greece as a vacation resort until Houthi drones hit UK bases in Cyprus, clarifying that southern Europe's border runs over the Mediterranean.

Breakthrough European companies like 11 Labs and Lora, originally from Sweden, don't position themselves as European competitors. They aim global from day one and expand distribution internationally. European investors need a US market foothold to apply the same quality distribution logic to European founders that prevails in Silicon Valley. Some founders start with global ambition, others grow into it as they experience early success.

Avoiding displacement by large labs

When founders navigate displacement by big labs at scale, a simple heuristic applies: avoid the problem sets that attract the smartest AI researchers. Don't build coding tools. Don't pursue AI for science. Instead, look for tacit knowledge—how people execute a task and apply taste. Benaich illustrates with his own use of Claude and Codeium. Once you realize these tools can accumulate feedback and learn across tasks rather than just handling one-off completions, you can pipe feedback back, store learnings in a skill file, and compound over time. That's the pattern founders should chase: capturing domain-specific taste and workflow knowledge that a general-purpose model won't replicate.

Benaich closes with a thought experiment. If a solo GP uses AI effectively to compound their own decision-making over time, what becomes possible with $200 to 300 million and a stack of AI tools? He jokes he'll check back in 10 years to see if it works.