AI has been the only headline that mattered for years. Nvidia. Data centers. The capex buildout that made investors feel like geniuses if they were long the stack. Maverick Capital’s co-CIOs Ben Silver and David Tykocinski think the easy phase is at risk of ending, and they’re explicitly watching for an “air pocket.” In Tykocinski’s framing, the gap between infrastructure spending and the moment that spending actually translates into productivity is exactly where volatility breeds.

That bet is not coming from a place of doubt about AI itself. It’s more like a timing and plumbing problem: after infrastructure ramps, the next constraints may shift. Tykocinski lays out why that matters. In the early days, when demand could be served within existing industry production capacity, the biggest growth sat downstream, in physical outputs tied to GPUs. Once you cross those thresholds, bottlenecks move upstream to fabrication, tooling, and even the more obscure inputs that show up on Japanese stock exchanges. Now he argues that migration is about to swing back downstream, toward the infrastructure and application layer, where AI becomes less of a “thing companies build” and more of a “thing that transforms how they work.”

To understand why Maverick believes that matters, zoom out to how tech cycles usually create winners. Tykocinski says the hallmark of the AI trade has been an inversion of prior tech decades. Historically, value accrued at the software application layer, close to end users, with companies like Salesforce, Google, and Meta. AI flipped that script, pulling returns toward hardware and infrastructure, including Nvidia’s GPUs, hyperscaler data centers, and the energy ecosystem powering them. The key monetization skill, in his view, is tracking where the bottleneck is moving over time. The investor takeaway is straightforward: if the constraint moves, the stock winners often change with it.

But the “air pocket” thesis is also rooted in what enterprises are actually doing with AI. Tykocinski points to a shift from the earlier narrative that large language models would replace existing enterprise systems. What’s happening instead is closer to integration. In the world of AI agents, he says it is about plugging an LLM into preexisting enterprise workflows and stacks, not wiping them out. That changes which components become critical. He says that brings more value closer to the edge and closer to the end user, making things like CPUs, databases, and edge infrastructure more relevant again.

If that sounds like “rotation” talk, it is, but Maverick isn’t only rotating within AI. Silver has been hunting in the place where money appears to be leaking out. His background is health care, and he previously ran Maverick’s health care book before being elevated to co-CIO. Right now, he says the sector “sounds like a vacuum,” with capital flowing out of health care and into AI. Silver’s response is not despair. He’s looking at life science tools, the equipment and consumables used to discover and manufacture drugs. It’s an unglamorous, consolidating corner of the market that has, in his words, become “left for dead.” He argues two tailwinds are converging and not fully priced in: reshoring and AI-driven drug discovery.

On reshoring, Silver points to the push to move pharmaceutical manufacturing back to the U.S., driven by trade policy and national security concerns. That requires a massive buildout of domestic capacity, meaning capex. And he expects the spending to start showing up in company earnings within three to six months. On AI, he links the dots from machine learning used in drug discovery to more drugs being discovered, which then leads to more drugs manufactured and a longer runway for consumables. In his view, that combination could make life science tools both an “AI winner” and a “modern mercantilist winner.”

Then there’s the catalyst hedge funds love: M&A. Silver says the sector’s consolidators include three to five major players, and that they have spent decades acquiring smaller firms. With cohorts of $5 billion to $10 billion companies trading at depressed valuations, he sees potential takeout risk acting as a floor. That, combined with the operational drivers from reshoring and AI, is the shape of the trade he’s building.

Of course, Maverick is also clear-eyed about what could break these theses. For Tykocinski, the deepest AI risk is China. He argues that many of the infrastructure plays that have driven equity appreciation, such as lasers, optics, analog semiconductors, and specialty materials, are historically vulnerable to commodification dynamics that hollowed out earlier generations of hardware companies. He points to why value historically accrued at the application layer: that’s where IP existed. Root hardware and materials, he says, are more subject to commodification over time, and investors may be underpricing structural risk from Chinese competition.

Silver’s concerns are described as more systemic. He points to the difficulty of the American political system making rational long-term decisions when short-term political incentives dominate. That structural problem shows up across policy areas affecting markets, and he says he is watching the U.S.-China geopolitical dynamic closely, particularly its implications for the technology supply chain.

Silver and Tykocinski both were elevated to co-CIO roles in 2021 by Lee Ainslie, the Maverick founder who began the firm in 1993 and built it into a long/short equity shop. Ainslie chose them together, not just one successor, based on what he observed about their sector leadership and “chemistry together.” Their strategy has an edge in that structure: it is built for rotation, not devotion to a single macro narrative. For executives and investors watching their own portfolios get “AI-shaped,” the stake is simple. If the value migration changes, the winners can change quickly, and the air pocket can be real even for believers.