Since CEO Lisa Su’s Taiwan CompuTex keynote on June 3, stock in Advanced Micro Devices (NASDAQ:AMD) stock has gone nowhere.
CompuTex was supposed to be Su’s big homecoming. She was born in Tainan, once the island’s capital, and emigrated to the U.S. at age 2. But the company’s key AI demo crashed, live on stage.
Su wound up upstaged by Nvidia (NASDAQ:NVDA) CEO Jensen Huang, ironically a second cousin, who was treated like a rock star.
There was more bad news in the last week. Nvidia’s share of the PC graphics card market has ballooned to 88%, against 12% for AMD. AMD was also downgraded by Morgan Stanley (NYSE:MS), from “overweight” to “equal weight.”
The Case for AMD Stock
AMD stock still has fans, like Sandeep Rao of Leverage Shares, who sees it living in a distinct part of the market than Nvidia. About 40% of AMD revenue comes from Ryzen PC chips.
It also has an embedded computing unit, the former Xilinx, which AMD bought in 2022. These are markets where Nvidia barely participates. Susquehanna analyst Christopher Rolland recently boosted his price target on AMD to $200.
A lot depends on the extent to which Ryzen is AI-capable. AMD is rushing out its latest version, the 900X3D, to get ahead of the opportunity.
Ryzen is where the rubber really meets the road for AMD this year. The market battle taking place at stores like Best Buy (NYSE:BBY) and online at Amazon. Com (NASDAQ:AMZN) won’t just be about feeds-and-speeds.
It will be built on brand marketing, on price, and on how hard resellers push AMD against PCs using Intel (NASDAQ:INTC) or ARM Holdings (NASDAQ:ARM) designs.
AMD’s past will define its short-term future. That’s not a bad thing.
The Squeeze on AMD
AMD stock is always being squeezed by competitors.
Nvidia has squeezed AMD out of the data center. AMD gets by in data center niches, supporting virtual machines in Microsoft’s (NASDAQ:MSFT) Azure cloud, for instance. These solutions, announced late last year, are just starting to roll out.
AMD is also seeking niches in what I’d call non-Nvidia AI. It has an open source compiler called Phoenix that can create AI functions that run locally, without a cloud. It has increased the speed with which is uses cached memory, to make such small language models valuable.
Gaming is what leads the market forward. It’s where Nvidia first made its name. But just as gaming can lead to opportunities, it can point toward problems.
One of the more popular accessories at CompuTex this year were gaming PC cases with multiple fans. I have already seen them at Atlanta computer stores.
Gaming PCs need multiple fans because fast PCs generate a lot of heat, and if the heat isn’t vented, the chip can breakdown. The problem has been around for 20 years. Back then, it was dealt with by reducing the power going into the chip, bringing circuit lines closer together under Moore’s Law.
But Huang’s Law, which accelerates computing speed faster than Moore’s Law, does so by making chips bigger. This increases the heat load. Could Huang’s Law be self-defeating?
The Bottom Line
I may be a voice in the wilderness, but the current “big AI” vision requires enormous amounts of energy and puts out enormous amounts of waste heat.
This is less of a problem for AMD than for Nvidia, but it’s still a problem. That’s because AMD is tied to the Microsoft PC ecosystem, and Microsoft is all-in on Big AI solutions.
What the market really needs is what I call “Little AI.” These are solutions that can run on clients, with minimal upgrade, and deliver real value worth paying for. It’s up to the software market to drive AMD forward, and we may have to wait some time for that.
In the near term, I expect AMD stock to go nowhere.
As of this writing, Dana Blankenhorn had a LONG position in NVDA, INTC, AMZN and MSFT. The opinions expressed in this article are those of the writer, subject to the InvestorPlace.com Publishing Guidelines.