Air Street Capital

Air Street writes first checks as small as $150 K and up to $3 M, aiming for 7 – 15 % initial ownership with the flexibility to follow on to $10 M+ over the life of a winning company. Ideal prospects are “AI‑native”: teams where at least one founder has published, shipped or open‑sourced cutting‑edge ML. Sectors of interest include foundation‑model tooling, industrial robotics, novel protein engineering, defence autonomy, fintech risk and AI‑enabled drug discovery.

The fund shies away from pure consumer social, crypto‑only plays and capital‑intensive biotech that lacks a clear computational wedge. Founders are expected to present an identified wedge in data access, model performance or distribution – traction can be as early as a working demo and a technical blog readership.

Air Street is comfortable investing pre‑revenue, but it expects a credible path to gross margins above 70 % and data‑network effects. The GP strongly prefers Delaware or UK Ltd structures and discourages SAFEs with uncapped discounted notes. syndication is acceptable if the partner retains decision‑making influence; however, the firm will walk if there are more than three primary investors on the cap table at seed.

Air Street’s investments map to two epochs defined by fund vintage.

The first cohort (2019‑22) produced signature wins such as Adept – an AI agents company recently acquired by Amazon – and Recursion (NASDAQ: RXRX) in computational drug discovery. Other standouts include V7 (computer‑vision data platform), Graphcore (AI semiconductors, acquired by SoftBank), and Athenian (engineering analytics). Life‑science bets like LabGenius and Gourmey illustrate the GP’s comfort blending wet‑lab with ML pipelines.  

Epoch II (2022‑present) reflects the larger Fund II cheque book. Recent seed leads include Profluent (protein language models), Patina Systems (AI‑optimized operating systems), Fern Labs (agent‑orchestration stack) and Polar Mist (maritime autonomy). Several stealth investments in generative‑video, spatial computing and defence autonomy underline an appetite for frontier markets. The portfolio skews 60 % software/40 % compute‑heavy or biotech; roughly 30 % are US‑incorporated Delaware C‑Corps, 25 % UK Ltd, and the balance EU GmbH or SAS. Notable exits so far are Allcyte (acq. Exscientia), ClipDrop (acq. Stability AI/Jasper) and Mapillary (acq. Facebook). The blended return multiple of Fund I is not public, but LP letters cite a gross TVPI above 5× as of Q1 2025.

‍

Air Street operates as a highly leveraged solo‑GP model. Nathan Benaich (Twitter @​nathanbenaich) is Founder & General Partner and the sole investment committee. Before Air Street, he led AI investments at Point Nine Capital and personally seeded Mapillary (acq. Facebook), Tractable and Thought Machine.

He holds a PhD from Cambridge and co‑authors the “State of AI Report,” which attracts hundreds of thousands of readers and positions the fund as a thought leader.

Nathan sits on the boards of Adept, Profluent and V7 and serves as an adviser to the UK Government on AI regulation.

Supporting him are functional specialists: Paula Pastor Castaño, a Spanish‑qualified lawyer, advises on fund operations and portfolio legals after a career at Wilson Sonsini. Venture Fellow alumni include Othmane Sebbouh (ENS Paris, autonomous robotics), Nitarshan Rajkumar (co‑founder, UK AI Safety Institute), Corina Gurau (Runway ML) and Alex Chalmers (researcher and writer).

Fellows rotate for 6‑12 months and source technical diligence, giving the fund perennial exposure to PhD‑level talent. Operating partners in the wider network contribute expertise in GPU infrastructure, procurement, bio‑wet‑lab automation and US federal sales, often in exchange for advisory shares rather than salary.

Decision‑making is fast: Nathan issues term sheets within a week of a first meeting when conviction is high, then runs a lightweight confirmatory process on IP ownership and security compliance.

The absence of a multi‑partner committee is marketed to founders as a speed advantage – there is no Monday meeting bottleneck – but it also means capacity is finite, so meetings are granted selectively. Nathan prefers memo‑first communication; founders are asked to send a two‑page problem‑solution‑traction note ahead of any call, allowing him to dive directly into technical depth.

‍

Air Street prefers direct email outreach. The recommended channel is nathan@airstreet.com with a short memo and deck attachment.

A Typeform‑based “Submit a Deck” link occasionally circulates via community posts but is not currently public. Founders can also DM @airstreet on Twitter for lightweight inquiries.

Warm referrals from portfolio CEOs typically receive fastest response, yet cold emails are answered within a week if they demonstrate a strong data or research edge. Use a subject line that includes your company name and the specific AI advance – for example, “Proterra – self‑supervised radar perception beating GPT‑4o.”

Attach a concise PDF; avoid Google Drive permission friction. Nathan reads outside business hours, so send weekdays before 4 p.m. UK time to hit the top of his queue. Don’t follow up more than once in five days; instead, include an update on progress.

Air Street runs a rolling Venture Fellowship for master’s and PhD students in machine learning, robotics and computational biology. Fellows work six to twelve months on deal sourcing, technical diligence and special projects like the State of AI Report. Alumni have joined portfolio companies such as Profluent and V7.

The firm also publishes the free Guide to AI Newsletter, a monthly digest of research breakthroughs, policy moves and practical operator tips. Newsletter subscribers receive early invitations to the annual RAAIS and London.AI summits, creating a talent funnel for the portfolio.