Could there be a $10 trillion company? Tech investing star Laffont believes so.

Philippe Laffont said the world could see its first $10 trillion company within 15 years, assuming global market cap grows from about $120 trillion to $200 trillion and the largest company reaches 5% of that total. He highlighted AI infrastructure and supply-chain names as attractive, noting Nvidia's forward P/E of 19.66 and calling it relatively cheap on 2027 buy-side earnings. The comments also pointed to a reshuffling of mega-cap leadership, with SpaceX at a $1.77 trillion debut valuation and OpenAI and Anthropic potentially approaching $1 trillion IPO valuations.

Analysis

The setup is less about a single “winner” than a persistent capital-recycling regime: AI capex is migrating from model headlines into the boring-but-valuable constraints that sit one layer upstream. That favors the picks-and-shovels complex—power distribution, grid hardware, semiconductor tools, and memory—because hyperscalers can defer software spend, but they cannot defer data-center power, cooling, or wafer-fab capacity if the buildout is to continue through the next 24-36 months.

The second-order effect is margin compression in the most crowded part of the stack. GPU leaders and frontier-model names may keep the narrative premium, but as the ecosystem matures, pricing power should shift toward suppliers with bottleneck exposure and long lead times; that creates a better risk-adjusted asymmetry in equipment and electrification names than in the headline AI beneficiaries. The most overlooked beneficiary is likely the grid-interconnect chain: if power remains the binding constraint, orders can remain elevated even if AI sentiment wobbles, because the backlog is driven by physical delivery schedules rather than quarterly hype.

The main risk is timing: the market is already pricing a multi-year AI spend cycle, so near-term upside in the obvious names may be capped unless estimates keep moving up. A genuine reversal would require either a capex pause from hyperscalers, a faster-than-expected easing in power bottlenecks, or a deterioration in enterprise AI monetization that forces investors to question ROI. Over the next several months, any disappointment in cloud capex commentary should hit the most valuation-sensitive semis first, while the infrastructure names should hold up better due to backlog visibility.

Key Decisions for Investors

• Overweight GEV and ETN versus NVDA on a 6-12 month horizon: long the power-constraint winners with more durable backlog conversion; risk/reward favors names tied to grid bottlenecks over crowded AI sentiment beta.
• Long LRCX/AMAT basket against short a portion of NVDA into any post-earnings strength: if AI capex persists, tools should capture sustained wafer-fab spend with less multiple risk than the GPU leader.
• Add MU as a delayed-beneficiary trade over 3-9 months: memory pricing should tighten if AI buildout stays hot, offering better cyclicality leverage once inventory normalizes.
• Use AVGO as the highest-quality compounder within the AI infrastructure complex: accumulate on dips for 12-month upside, with lower execution risk than frontier AI platforms and stronger cross-sell optionality.
• Tactically reduce exposure to AAPL/GOOGL on rallies and rotate toward the supplier chain: these names are less directly levered to the capex impulse and may lag if the market continues rewarding bottleneck exposure.