The Fragile Future of AI: Beyond Venture Capital Subsidies

AI is booming. In H1 2025 , AI startups raised over $83B , nearly 58% of all venture funding ( ainvest.com ). In Q2 2025 , AI alone soaked up ~$50B of $101.5B , while non-AI funding fell to a 7-year low ( bestbrokers.com ). Those two facts pretty much define the moment: capital has picked a winner, at least for now. I’ve helped banks, insurers, manufacturers, and media companies ship AI into production. The excitement is earned — some deployments truly change outcomes. But I’m also uneasy: a huge slice of what looks like traction is subsidized by venture money , not sustained by profits. That makes the system fragile when sentiment turns or rates rise. This piece is half technical, half theoretical, with a dose of lived experience. A few years ago I was “unavoidable.” Today, parts of my work — data prep, baseline modeling, doc writing — can be done by AI. That irony is the point: what happens to an ecosystem (and the people in it) when the VC gravy train slows? A Venture Capital–Fueled AI Boom The scale and concentration are historic. By mid-2025 , there were 500+ AI unicorns with an average $5.4B valuation — many young and pre-revenue ( elephas.app ). FOMO pushed prices beyond normal metrics ( elephas.app ). PitchBook’s read: nearly half of VC dollars in recent quarters went to AI, up from ~20% only a few years ago ( bestbrokers.com ). Non-AI startup investment sank to $41.8B in Q2 2025 , the lowest in 7+ years ( bestbrokers.com ). A handful of mega-checks tell the story. OpenAI raised $40B in early 2025 (SoftBank-led) — a private-tech record ( ainvest.com ). Meta bought ~49% of Scale AI via a $14.3B stake ( bestbrokers.com ). In Q2 2025 , $40B of the $91B global VC total went into AI — and over one-third of that to Scale AI alone ( ainvest.com ). This is hyper-concentration . Speculative heat is obvious. Thinking Machines Lab raised $2B seed at a $12B valuation with no product ( elephas.app ). Safe Superintelligence raised $2B at $32B with no revenue ( elephas.app ). Sam Altman called much of this “not rational,” predicting “someone is going to lose a phenomenal amount of money.” ( elephas.app ). Markets echo it. Tech is ~34% of the S&P 500 , exceeding th…