I Sold My AI Startup Before Revenue: Here’s What Investors Missed — And Founders Shouldn’t

2026-06-17 · Source: Artificial intelligence - Crunchbase News · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Intermediate, quick

Summary

Alexander Kardos-Nyheim, founder of Safe Sign Technologies, sold his AI research company to Thomson Reuters pre-revenue after approximately 20 months, marking Thomson Reuters' first such acquisition in its 170-year history. Safe Sign Technologies, developed with researchers from Cambridge, DeepMind, Harvard, and MIT, created highly capable legal reasoning models using novel, capital-efficient algorithms, achieving top performance with a fraction of the spending by larger labs. Despite this scientific achievement, early investors, particularly in the U.K., prioritized product and traction, leading Kardos-Nyheim to secure most funding in the United States. He now advocates for investing in foundational AI, observing that in Q1 2026, foundational AI startups raised around \$178 billion, with 97% going to OpenAI, Anthropic, and xAI. Kardos-Nyheim argues that long-term value and durability reside in addressing core scientific challenges like cost, speed, reliability, interpretability, and safety at the foundational model layer, rather than building application-layer solutions on external models.

Key takeaway

For deep-tech founders weighing where to build, prioritize solving fundamental AI challenges at the model and system level. Your efforts in training efficiency, model architecture, and inference cost will create durable value, unlike application-layer products built on external models. Avoid chasing early revenue; instead, build something indispensable that the entire AI stack will eventually rely on, attracting capital when its importance becomes undeniable.

Key insights

Foundational AI, addressing core scientific challenges, offers more durable value than application-layer solutions built on external models.

Principles

• Foundational AI innovation, not application wrappers, drives long-term value.
• Building on external models creates dependency and competitive risk.
• Deep-tech capital markets often lag in recognizing foundational breakthroughs.

Method

Founders should assess if their technical team rivals top labs, if their problem is at the model/system level, and if their solution will become indispensable over five years.

In practice

• Invest in solving core AI challenges like training efficiency and inference cost.
• Prioritize scientific talent and deep-tech research over immediate revenue generation.
• Develop proprietary model architectures to avoid upstream dependencies.

Topics

• Foundational AI
• Deep-tech Startups
• AI Investment Strategy
• Model Architecture
• Inference Cost
• AI Capital Efficiency

Best for: Entrepreneur, Investor, Director of AI/ML

Related on AIssential

• Europe is at risk of sleepwalking into AI armageddon — European AI startups face "armageddon" due to rising US platform costs and capital market volatility.
• AI Payoff in Focus During Tech Earnings Bonanza — Strategic AI investments are yielding substantial returns for cloud-centric tech giants, while others face scrutiny over their AI spending justification.
• AI investment hits $800 billion, India emerges as key applications hub - Business Standard — AI value creation is shifting from foundational models to application-layer deployment and rapid enterprise monetization.
• AI startup revenue hits $80 billion, but Anthropic and OpenAI take almost all of it — AI startup revenue is highly concentrated, with Anthropic and OpenAI capturing 89% of the $80 billion market.
• AI Models as a Commodity and Why Data Foundations Decide Who Wins - with Guillermo B. Vazquez of SAP — AI models are a commodity; robust data and process foundations are paramount for successful enterprise AI adoption.
• Deepseek reportedly prioritizes AGI research over quick profits despite billions in funding — Deepseek prioritizes AGI and open-source development over short-term profits, even with significant state-backed funding.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial intelligence - Crunchbase News.