https://www.wsj.com/tech/ai/who-owns-the-future-of-ai-05610081?mod=djemTECH
Perplexity:
Here’s the investment read-through of the article:
The biggest implication is that AI is shifting from a pure “model race” to an ecosystem race, where the winners may be the companies controlling compute, chips, cloud infrastructure, and distribution rather than only the chatbots themselves. That means investors should think beyond headline AI app names and look harder at the supply chain and infrastructure layer.
Highest-conviction AI infrastructure
- TSMC — probably the cleanest “picks-and-shovels” exposure if AI chip demand keeps rising, since the article points directly to manufacturing strain and capacity pressure.
- NVIDIA — still the most direct beneficiary of AI compute spend, though the trade is more crowded and valuation-sensitive than in earlier phases.
- Broadcom — attractive if you want exposure to AI networking, custom silicon, and infrastructure rather than just GPUs.
- Arista Networks — benefits from the data-center networking buildout that follows AI cluster expansion.
- Micron — memory demand can become a second-order winner when AI deployments scale.
Cloud and platform layer
- Amazon / AWS — the article suggests AWS is regaining importance in the AI race, so Amazon has become a more credible AI infrastructure owner again.
- Microsoft — still strategically strong because of its AI distribution and enterprise reach, though the article’s tone suggests the “winner-take-most” thesis is less clear than it was a year ago.
- Alphabet — worth watching because of AI research strength plus cloud and advertising integration, but execution risk remains.
Secondary beneficiaries
- Vertiv — data-center power, cooling, and thermal management can compound as AI clusters grow.
- Eaton — electrical infrastructure and power distribution are underrated AI beneficiaries.
- GE Vernova — if AI data centers keep driving incremental power demand, utility and grid buildout names can matter more.
- Super Micro Computer — a high-beta play on AI servers, but more volatile and operationally risky.
Higher-risk private-market signals
- Anthropic — the article frames it as the presumptive front-runner in model competition, but it’s private, so most investors can only access it indirectly through partners and infrastructure suppliers.
- OpenAI — still a major force, but the article hints that growth may be normalizing, and the company’s structure adds governance complexity.
- SpaceX-related compute partnerships — interesting strategically, but not a straightforward public equity theme; the article mentions SpaceX renting compute to Anthropic, which is more of a signal than a direct investment path.
