AI Infrastructure Stocks: Where the Real Money Is Being Made

The first wave of AI investing was about identifying the companies building AI models and chips. Nvidia became the obvious winner, and its stock went up more than 5,000 percent from Jason Bodner’s October 2019 recommendation at $4.50 split-adjusted. But Bodner argues in his Accelerated AI presentation that the next wave will be different. The biggest winners will not be the companies building AI models but the companies building the physical infrastructure that makes AI possible.

This is the picks-and-shovels thesis. During a gold rush, the people who sell picks and shovels make more money than the miners. In the AI boom, the picks and shovels are the networking switches, the optical chips, the power systems, and the cooling solutions that hold AI data centers together.

The Infrastructure Stack

AI infrastructure is not a single category. It is a stack of interdependent components:

Silicon. The GPUs and custom AI chips that do the computing. Nvidia dominates this layer, but Broadcom designs custom AI silicon for OpenAI, Meta, Anthropic, and Google. These custom chips are optimized for specific workloads and are becoming increasingly important as hyperscalers move away from generic GPUs.

Networking. The switches and routers that let thousands of GPUs communicate. Broadcom dominates this layer. As Bodner puts it: “No Broadcom, no communication. The AI goes dark.” In a data center with thousands of GPUs, the networking layer is the bottleneck, and it is where the photonics transition is most urgent.

Optical components. The chips and transceivers that convert electrical signals to light. Broadcom debuted the industry’s first 400G optical chip. This is the exact technology that enables photonics to replace copper. For more on this component, see our Broadcom optical chips article.

Power. AI data centers consume enormous amounts of electricity. One ChatGPT question burns 10 times more electricity than a Google search. A single AI data center eats as much power as 100,000 homes. The power infrastructure, from generation to distribution to cooling, is a massive investment category. Bodner previously recommended Vertiv Holdings, a power and cooling infrastructure company, a full year before Goldman Sachs published its first AI energy report. Vertiv gained over 1,900 percent.

Photonics. The technology that replaces copper wires with laser beams. Light travels at 124,000 miles per second, carries more data, barely uses power, and does not generate heat. Bodner calls this “holy grail tech,” and the structural evidence supports it. Nvidia has invested $7 billion in photonics companies. AMD is building a $280 million photonics research hub. The Optical Interconnect Alliance includes Nvidia, AMD, Broadcom, Microsoft, Meta, and OpenAI.

The Free Ticker: Broadcom (AVGO)

Bodner reveals Broadcom for free in the presentation. Broadcom’s AI revenue was $10.8 billion last quarter, up 143 percent year over year, and the company is guiding to $16 billion next quarter. BlackRock owns over 12 million shares. Fidelity, Vanguard, Morgan Stanley, and Norway’s sovereign wealth fund are all loading up.

Broadcom is the most directly exposed large-cap stock to the AI infrastructure buildout because it occupies three critical positions in the stack: custom AI silicon, networking switches, and optical chips. For a full analysis, see our Broadcom stock article.

The Three Bonus Picks

Behind the $179 paywall, Bodner identifies three smaller companies with higher upside potential:

  1. A secret weapon inside every major AI data center. Two of the three biggest AI spenders are customers. Nvidia just tapped it as a key partner. Sales up 35 percent last quarter. 10x smaller than Broadcom.

  2. The purest photonics play. Hardware moves data as light between data centers across cities and states. New product cuts networking power use 70 percent. Sales over $6 billion. 20x smaller than Broadcom.

  3. Builds parts that turn electrical signals to light inside AI clusters. Went on an acquisition spree for photonic firms. Nvidia signed a multi-billion partnership. Jensen Huang calls it “the next trillion-dollar company.” Sales up 42 percent year over year.

These are the picks that Bodner believes could deliver the kind of returns that Akamai (+16,497 percent) and Equinix (+37,000 percent) delivered during the internet fiber-optic transition. For more on the historical pattern, see our Acceleration Curve article.

The Broader Context

AI infrastructure spending is accelerating. The Stargate project, a joint venture between OpenAI, Oracle, and SoftBank, is deploying $100 billion. Meta is building a Manhattan-sized data center. Amazon’s Project Blue is under construction. OpenAI filed an S-1 for a September IPO at a $1 trillion-plus valuation. Anthropic is preparing an October listing approaching $1 trillion. All of this capital is flowing into the physical infrastructure that AI requires.

For a different angle on AI infrastructure investing, see our ASI Fund review, which covers Alexander Green’s picks-and-shovels approach at The Oxford Club, including nuclear power and collaborative robotics.

Considerations

The infrastructure thesis is sound, but the valuations of AI infrastructure stocks have risen significantly. Broadcom trades at a premium that reflects its AI growth, and the smaller photonics plays Bodner teases are already surging (133 percent, 320 percent, and 210 percent in recent months). The risk is that expectations are elevated and any slowdown in AI spending could trigger a correction. The opportunity is that the infrastructure buildout is in its early stages and will continue for years.

If you want to explore Bodner’s full thesis, you can access the Accelerated AI presentation through Brownstone Research.

This is not financial advice. Always do your own research before investing.