The AI rally reboot

The AI trade is back, and Meta Platforms is leading the charge.

The Facebook parent just struck another major deal with AI cloud infrastructure company CoreWeave, committing $21 billion on top of the $14.2 billion it had previously invested. CoreWeave’s data centers, packed with hundreds of thousands of Nvidia GPUs, give Meta the compute power needed to scale AI models quickly.

That infrastructure push comes just as Meta rolls out its new AI model, Muse Spark, looking to regain momentum after last year’s underwhelming Llama debut. Early signs are encouraging: Muse Spark reportedly outperformed Google’s Gemini on some benchmarks and held its own against OpenAI and Anthropic. Meta shares jumped 2.61% this afternoon as investors cheered the company’s efforts to catch up to the competition.

Intel gains ground

Meta isn’t the only tech stock gaining momentum—in fact, it’s got a long way to go to catch up to Intel.

Intel was once an also-ran in the AI race, but that all changed after news broke that the company is expanding its long-standing partnership with Google, which has relied on Intel processors since its earliest server builds nearly three decades ago. Now, Intel’s new Xeon 6 chips are set to power AI training and inference workloads, helping Google scale its AI infrastructure more efficiently

But demand isn’t just coming from Big Tech. Elon Musk recently tapped Intel as the first major chipmaker in his Terafab project, an ambitious effort to produce 1 terawatt of AI and robotics compute annually—roughly double current US demand.

Put it all together, and Intel is reasserting itself as a credible AI infrastructure player, chipping away at a market long dominated by Nvidia. Shares have climbed 28.5% over the past five days, and just hit a five-year high.

The real economy behind AI

But while AI is once again the talk of the town, Lowe’s CEO Marvin Ellison is reminding investors that the old-fashioned economy still does the heavy lifting—even for AI itself.

Ellison noted that thousands of data centers are now under construction, and all of them require real-world labor to bring online. That’s driving a renewed focus on skilled labor: Ellison announced that the company’s foundation will invest $250 million over the next decade to train workers in trades like plumbing, carpentry, and electrical work. BlackRock is backing a similar push, committing $100 million to expand the skilled trades pipeline.

So before you get swept up in the exciting, shiny AI trade, remember what the HALO trade has been trying to tell us: the physical economy still underpins it all. Or, as Ellison succinctly put it: “AI can’t climb a ladder.”—SY