A major shift in how the world's smartest investors view the semiconductor supply chain is underway, and it reveals something crucial about who will actually profit most from the AI boom. Billionaire hedge fund manager Philippe Laffont, who oversees a $33 billion public equity portfolio through Coatue Management, recently trimmed stakes in eight of his fund's ten largest positions. But he added to just two: Taiwan Semiconductor Manufacturing Company (TSMC) and Lam Research, a maker of chip fabrication equipment. This move signals a fundamental reordering of power in the semiconductor world.

The global semiconductor market generated $793 billion in combined revenue in 2025, up 21 percent from the prior year, marking the fastest growth the sector has seen in a decade. Artificial intelligence spending drove nearly all of this growth. Yet despite this explosive demand, Laffont's portfolio decisions suggest that not all semiconductor companies are equally positioned to benefit.

Where Does the Real Power Sit in the Chip Supply Chain?

The semiconductor industry operates in distinct layers. At the top are fabless designers like Nvidia and AMD, which create chip blueprints but don't manufacture them. Below them are foundries like TSMC, which actually build the chips. And at the foundation are equipment makers like ASML and Lam Research, which supply the machines that make chip manufacturing possible.

Laffont's strategic shift reveals a critical insight: the companies in the middle and bottom of this chain hold more leverage than the designers at the top. TSMC, the world's largest contract chipmaker, generated $122.9 billion in revenue in 2025, up 31.6 percent. The company manufactures chips for Nvidia, Apple, AMD, Qualcomm, and hundreds of others. When TSMC announces a bottleneck in advanced packaging capacity, the entire AI buildout slows. This gives TSMC enormous pricing power.

Lam Research, which supplies wafer fabrication equipment, saw its stock climb nearly 60 percent so far in 2026. The company reported strong earnings and better-than-expected guidance in April, yet the stock barely moved, indicating the market has already priced in stellar results. This suggests investors recognize the company's durable competitive advantage.

Why Are Hyperscalers Losing Leverage?

The huge spending by hyperscalers like Microsoft, Meta Platforms, Amazon, and Alphabet is transferring significant cash flow from cloud computing companies to manufacturing companies. Microsoft recently shared expectations that its capital expenditures will climb to $190 billion for the calendar year, with marked acceleration in the second half. Yet this massive spending creates uncertainty for the cloud giants betting on future returns, while creating tremendous opportunity for companies like TSMC and Lam Research.

Both TSMC and Lam Research have been able to raise pricing as demand soars. The same advantage applies to two new additions to Coatue's portfolio: ASML Holding, the sole manufacturer of extreme ultraviolet lithography machines essential for advanced chip production, and Micron Technology, a memory chip supplier. Each machine ASML produces costs approximately $200 million and contains over 100,000 components. Without ASML's machines, TSMC, Samsung, and Intel cannot make advanced chips at 3 nanometers or below.

Laffont sold chipmakers such as Nvidia and Broadcom, which are stuck between the manufacturers and the hyperscalers. Both are reliant on TSMC to manufacture their chips, as its leading-edge technology and scale make them the only viable partner to meet the demands of their hyperscaler customers. That puts TSMC in the driver's seat as it controls its capital expenditure budget and pricing. TSMC's equipment suppliers will benefit as well, as TSMC can pass costs on to the rest of the supply chain, as their demand is practically insatiable.

How to Identify the Real Winners in AI Infrastructure?

• Control of Bottlenecks: Companies that control scarce resources or manufacturing capacity can raise prices and maintain margins even as demand fluctuates. TSMC's advanced packaging capacity is currently the bottleneck for the entire AI chip industry.
• Durable Competitive Moats: Equipment makers and foundries have high barriers to entry due to massive capital requirements and technical expertise. ASML has no real competitors in extreme ultraviolet lithography, giving it near-monopoly pricing power.
• Visibility and Planning Horizons: Companies with longer-term visibility into customer demand can plan capital investments more confidently. Large cloud providers now plan infrastructure investments years in advance, giving suppliers like TSMC and Lam Research clearer demand signals than chip designers receive.

Vontobel Asset Management, an investment firm that conducted on-the-ground research across the semiconductor supply chain, noted that visibility in this cycle is longer than usual. Planning horizons extend closer to two years due to manufacturing complexity and long-term infrastructure investments by large cloud providers. This extended visibility is rare for the semiconductor industry and meaningfully informs how investors should think about portfolio positioning.

The firm's research teams spent time with management teams, engineers, and operators across the accelerated computing stack, from Broadcom in Silicon Valley to TSMC's manufacturing facilities in Taiwan and Arizona, and to key equipment players in Europe and Japan. The consistent message: end demand for AI compute appears to remain resilient not only through 2026 but likely into 2027 and, absent an external shock, remains structural for the multi-year horizon.

What Does This Mean for the Broader AI Infrastructure Buildout?

The semiconductor industry remains cyclical, and after growth of this magnitude, some moderation would be neither surprising nor unhealthy. Yet evidence across the semiconductor value chain continues to support durable demand. The bottlenecks remain concentrated in a small number of hard-to-replicate areas: advanced chip manufacturing, packaging, memory, networking, and power. As such, chips tied to computing, memory, and power, along with the equipment used to manufacture them, remain the clearest expressions of this structural shift.

Nvidia crossed $215.9 billion in fiscal year 2026 revenue, becoming the first chip firm to cross $100 billion in annual revenue. Yet despite Nvidia's dominance as a chip designer, Laffont's portfolio moves suggest that the real wealth creation in AI infrastructure will flow to the companies that manufacture the chips and the equipment that makes manufacturing possible. TSMC's position in the middle of every product on the semiconductor industry's list means that when TSMC announces a bottleneck, the entire AI buildout slows. For founders building AI-native products anywhere outside the US, Taiwan, and South Korea, supply chain concentration on TSMC represents the single largest systemic risk to watch through 2028.

TSMC's valuation reflects this advantage. Even with market-beating returns so far in 2026, its price-to-earnings ratio sits at an attractive 25, in line with Microsoft and Nvidia, and around where other hyperscalers trade despite the much more positive medium-term outlook in TSMC's business. That makes it a very attractive stock and deserving of Coatue's top position in the portfolio.