Intel’s Peculiar Rally: A Most Improbable Event

by

Shares of Intel (INTC +6.05%) experienced a rather noticeable upward trajectory today, climbing a respectable 5.5% as of 12:52 p.m. EDT. One begins to wonder if the market is aware of something we aren’t. Possibly a subtle shift in the Earth’s magnetic field, or a particularly persuasive flock of pigeons. (It’s always the pigeons.)

The general semiconductor sector, as is its wont, decided to participate in this upward movement, displaying a generally enthusiastic, if somewhat inexplicable, bullishness. Intel’s rise, however, appears to be linked—at least partially—to the activities of Advanced Micro Devices (AMD +8.81%), who announced a deal with Meta Platforms (META +0.13%). A deal, it should be noted, of truly staggering proportions.

Intel, meanwhile, has also been engaging in a spot of deal-making itself. A smaller deal, admittedly, but one that may contribute to the general air of… well, not exactly optimism, but perhaps a slightly less profound sense of existential dread.

Loading widget...

AMD Dances with Meta, While Intel Attempts the SambaNova

Today, AMD announced a multi-year agreement with Meta for six gigawatts of AMD’s Helios rack systems. Six gigawatts. To put that into perspective, that’s roughly the energy output of several small suns, or approximately the amount of electricity required to power a moderately sized obsession with collecting porcelain thimbles. (The connection remains elusive.) Meta also received a performance-based warrant for 160 million AMD shares – roughly 10% of the company. It’s remarkably similar to a deal AMD struck with OpenAI in October, suggesting a pattern. A pattern that may or may not involve sentient algorithms and a global shortage of paperclips.

Some might interpret this as a competitive blow to Intel, but it seems more likely that Meta is simply diversifying its sources of GPU-based systems. Like a particularly cautious squirrel stockpiling nuts. Investors seem to be viewing this as a signal of massive demand for AI computing, which, in turn, could benefit everyone. Including, presumably, the aforementioned squirrel. And with Meta committing to this enormous amount of compute, it might free up Intel to sell to other players, given the industry’s current, shall we say, constrained supply.

Intel, not to be outdone, announced its participation in a $350 million funding round for AI chip startup SambaNova, along with a new product partnership. The press release, as these things often are, was a masterclass in corporate optimism.

SambaNova and Intel announced a planned multi-year strategic collaboration to deliver high-performance, cost-efficient AI inference solutions for AI-native companies, model providers, enterprises, and government organizations worldwide, built around Intel Xeon-based infrastructure.

SambaNova also unveiled its new SN50 inference chip, which it claims is five times faster than “competitive” inference chips, and can run AI inference at “three times lower cost” than GPUs. Naturally, this claim is subject to the usual caveats regarding marketing hyperbole and the inherent unreliability of all things digital. (It’s a statistical certainty, really.) Softbank, incidentally, will be the first customer to implement the SN50 in its data centers in Japan. Softbank, of course, invested $2 billion in Intel last summer. A coincidence? Possibly. But in a universe governed by quantum mechanics and the capricious whims of market forces, one tends to suspect patterns where none exist.

Agentic AI Should Lift Intel This Year (Probably)

The demand for AI compute is accelerating, and supply is, to put it mildly, strained. GPUs, memory, and even CPUs are in short supply. Intel, it appears, is particularly affected, having recently raised its server CPU prices by 10% to 15%, and reportedly being sold out for the year. This, naturally, is a situation that requires careful analysis. (And a large supply of calming tea.)

In this new era of agentic inference, Intel may have a promising growth area, even without leading AI GPUs. But, of course, CEO Lip-Bu Tan also recently noted that Intel is still going to build AI GPUs as well. A sensible plan, one might think, until one considers the sheer logistical complexity of simultaneously pursuing two mutually exclusive strategies. (It’s like trying to herd cats while simultaneously inventing a perpetual motion machine.)

CFO David Zinsner is scheduled to speak at the Morgan Stanley Technology, Media, and Telecom conference next week on March 4, where he may share more details about this new partnership. One can only hope he brings a detailed map, a compass, and a healthy dose of skepticism.

Read More

2026-02-24 22:33