The Big Shift: How Real-World AI Can Be a Win for Investors—and Earth

AI investment dollars won’t vanish in a crash—they will be redeployed. It’s an opportunity for dollars to shift out of a handful of horizontal players and into the companies solving real-world problems.

The Big Short told the story of the 2007-08 housing collapse through the eyes of a handful of investors who saw it coming. The Big Shift will tell the story of the AI shakeout of 2030 through the founders who benefited as capital rotated out of a handful of AI giants and into the rest of the economy.

In nature, a fire clears the forest floor, letting sunlight reach the ground and returning nutrients to the soil so new growth can thrive. The coming shakeout works the same way: it clears out some of the largest AI companies and lets a generation of smaller, real-world AI companies grow in their place.

I don’t think the AI bubble is set to burst—and I say that with skin in the game, heavily invested in private and public AI and infrastructure companies. We won’t see a 2008-style crash. But we are one burnout away from a reckoning. At least one high-profile company is propping up an unprofitable, take-over-the-world business model on tranches of investor capital, and it isn’t the only one.

It only takes a single flameout to spark the rest.

AI valuations are due for a healthy reset

This fire would also force a reset in AI valuations, which frankly, would be healthy. Some of these investment dollars should instead go toward more innovative players in real-world sectors that VCs have largely abandoned since 2021. According to the OECD, AI companies absorbed 61% of all global VC investment in 2025, double their 30% share in 2022.

A lot of that traces to the venture hype cycle. The 2021 vintage from the COVID-era was highly overvalued, and OpenAI’s 2022 launch of ChatGPT only accelerated the trajectory. Today, many of the world’s largest private equity firms are trading at 52-week lows despite having trillions in assets under management. Part of this is a hangover from the SaaSpocalypse, a decade of over-indexing on debt-fueled software companies that won’t make it.

That correction is already underway in SaaS, and it’s spreading to horizontal AI. Expect a wave of consolidation, markdowns, and zombie unicorns dropping dead as their backers stop pumping money into them. The venture model pushes everyone to chase the same handful of winners, but I’d argue for a smarter distribution of capital toward the companies that will actually carry us into the 2030s. Most will still have AI in them; trying to build the next OpenAI is a losing proposition. Gartner forecasts that agentic AI in supply chain management alone will grow from under $2 billion in 2025 to $53 billion by 2030.

Infrastructure builders and the picks-and-shovels play

On an earnings call earlier this year, Blackstone CEO Stephen Schwarzman pointed to the firm’s exposure to physical assets—logistics, real estate, transportation, and communications infrastructure—as well-insulated from disruption, and called Blackstone “extraordinarily well positioned for an AI-enabled future.” It’s no surprise the firm is now the largest investor in AI-related infrastructure in the world. The infrastructure builders are more promising than the horizontal players. This is what VC and PE keep missing: the picks-and-shovels play and betting on vertical AI over horizontal.

Part of why these companies get overlooked is that they’re hard to hype. Unless you’re an OpenAI or an xAI with a Sam Altman or an Elon Musk at the helm, standing out is difficult, and right now the market is priced to perfection, with the expectation that every one of these bets will hit. As the storytellers behind The Big Short knew, that’s never how it goes.

Real-world AI tends to make for a better story anyway. It isn’t about burning more compute to answer a query, or about speeds and feeds. It’s tangible. It’s a robot sorting waste or a sensor preventing a power grid failure. The companies doing this work are solving difficult, real-world problems, and they’re flying under the radar with investors and the media alike.

That gap is the opportunity.

Why the horizontal AI layer is the riskier bet

The uncomfortable truth is that the foundation-model layer is starting to look like a commodity. When a dozen well-funded labs race to build roughly the same capability, pricing power erodes and differentiation collapses into a compute-spending contest that only the largest balance sheets survive. The application layer works the opposite way. A company solving a specific, physical problem, whether that’s routing recyclables, balancing a grid, or keeping a fulfillment center running, builds an advantage that compounds with every deployment, because hard-won domain knowledge can’t be replicated by raising another mega-round. That’s the inversion investors keep missing. In AI, the boring, vertical, real-world layer may turn out to be the defensible one.

Where investors and founders should look next

AI investment and adoption are just getting started. Gartner predicts that by 2030, over 80% of enterprises will deploy industry-specific AI agents in support of critical business objectives, up from less than 10% in 2025. VCs and future entrepreneurs have a massive opportunity over the next few years, whether there’s an AI wildfire or not.

Right now, investors are going all-in on the perceived winners, doubling and tripling down with obscene amounts of late-stage capital. That approach either hits big or misses big. At the moment, it’s hitting, so enjoy it, but there’s a huge opportunity in funding profitable businesses that can go public and solve real-world problems. It’s the healthier play, fiscally and in valuation terms. You know the founders who walk into Shark Tank asking for $500,000 in exchange for 5%? That’s roughly where we are with horizontal AI.

I’m trying to find the optimal balance between environmentalist and capitalist; perhaps they’re not mutually exclusive. The entrepreneurs worth backing are the ones building existing things in new and sustainable ways—Patagonia recycling its own materials and repairing what it sells, Framework making laptops you can actually fix, AMP using AI to pull recyclables out of the waste stream. None of them is trying to be the next OpenAI, and that’s the point.

Instead of trying to be the next Bezos or Altman, be inspired to have a vision where capitalism and environmental sustainability can co-exist. The next decade belongs to the founders solving tangible problems in the physical world, from the forest floor to the sun.

That’s the Big Shift. The only question is who sees it coming.