Portfolio review takes on a new urgency for VCs as AI makes 2024’s great deals look like 2026’s clear mistakes

In Ferris Bueller’s Day Off , Matthew Broderick famously winks straight into the camera: “Life moves pretty fast.” Life does move fast, but AI might move even faster.

I talk to VCs all day, and it recently occurred to me: Though venture capital’s meant to be the long game, AI is moving so quickly that it’s making deals that looked good in 2024, obsolete by 2026.

“No doubt that AI is changing really fast, and it’s changing the half-life of a thesis,” said Eric Archer, cofounder of VC firm Monashees.

So, I’ve been wondering: If you’re an investor right now, what does it even mean to do a responsible portfolio review? How many companies that looked great in 2023 or 2024 are now just toast? “The two categories that have some amount of insulation are regulated license businesses and businesses with proprietary data that’s difficult to access,” said Kamran Ansari, founder and managing partner at Kapital Ventures, and venture partner at Infinity VC.

“Absent those two things, everything else feels exposed.” Portfolio review is the process in which an investor or firm sits down and assesses the health (or lack thereof) of all the companies they’ve backed.

Any portfolio review, in this environment, is necessarily nuanced, said Lily Lyman, managing partner at Underscore VC.

“Portfolio reviews right now aren’t just a question of ‘are you performing?’ or ‘are you not performing?’ Or ‘are you on track or off track?’” she told Fortune .

“It’s more like: What track are you on, and what does that mean?” Lyman says there are essentially four lanes for startups right now: the soaring AI consensus bets, efficient-growth companies, companies with paths to product‑market‑fit in tough industries, and the companies caught in the crosshairs by OpenAI and Anthropic.

“In that fourth bucket, the market’s shifted so much in terms of what’s possible with Claude or any of the models, that [the startup’s] fundamental value prop is no longer as valuable as originally thought,” said Lyman.

“So, the question becomes: Do you have something that’s valuable? If so, is it people? Is it a product? Is it distribution? What do you lean into in order to try to recoup value?” Of course, startups fail all of the time.

It’s part of the model and the power law, VC’s golden rule, dictates that there only need to be a few home-run victors.

But in the AI bubble, even the winners of a few months ago don’t always stick.

Take Perplexity.

“Perplexity is one of those funny companies where it was so molten-lava-hot,” said Ansari.