Advanced Micro Devices (AMD) is still struggling to keep up with red-hot demand for artificial intelligence chips, according to CEO Lisa Su. Speaking in Taipei, Su said AI demand is outpacing the industry's ability to manufacture and assemble enough chips, and that AMD remains supply constrained—including on memory—even after boosting output this year.
The comments underscore a persistent bottleneck in the AI boom: while demand for powerful processors used in data centers and AI models has exploded, the complex manufacturing and packaging needed to produce them hasn't caught up. Su said AMD is aiming for a "substantial" increase in supply in 2027, signaling that the crunch could last for years.
Why supply is still tight
Making advanced AI chips isn't just about the silicon itself. It also requires cutting-edge memory, like high-bandwidth memory (HBM), and advanced packaging that stacks components together. These steps are highly specialized and capacity is limited, with a handful of suppliers controlling the market.
AMD, like rivals, has been working to secure more manufacturing capacity and memory supply. But Su's remarks suggest that even with recent increases, the company can't fully meet demand. The memory shortage is particularly notable, as it affects not just AMD but the entire industry.
This isn't a new problem. The AI chip supply chain has been stretched for over a year, and companies like Foxconn and others have reported surging AI-related revenue. The demand is real, but the ability to deliver is the limiting factor.
What it means for investors
For everyday investors, this news is a double-edged sword. On one hand, it shows that demand for AI chips is still booming, which is positive for AMD and the broader semiconductor sector. On the other hand, it means AMD may not be able to fully capitalize on that demand in the near term, potentially capping revenue growth.
When a company is supply constrained, it often means it has to turn away orders or delay deliveries. That can frustrate customers and give competitors a chance to win business. For AMD, the key question is whether it can ramp up supply fast enough to keep pace with Nvidia and other rivals.
The 2027 timeline is important. It suggests that the industry expects the supply-demand gap to persist for a while. Investors should watch for updates on AMD's manufacturing partnerships and memory supply deals, as those will be critical to closing the gap.
This situation also has broader implications. The AI boom has driven massive investment in data centers, and companies like DayOne Data Centers are tapping public markets to fund expansion. If chip supply remains tight, it could slow the build-out of AI infrastructure, affecting everything from cloud computing to autonomous driving.
The bigger picture
AMD's comments come at a time when the AI trade is under scrutiny. Investors have poured billions into AI-related stocks, but there are questions about whether the spending will translate into profits. Supply constraints add another layer of uncertainty.
For now, AMD's message is clear: demand is strong, but supply is the bottleneck. The company is working on it, but the fix won't happen overnight. As Su put it, the industry is "still supply constrained," and the path to a substantial increase in supply runs through 2027.
For investors, this means patience is key. The AI story is still intact, but the timeline for AMD to fully benefit may be longer than some hoped. Keep an eye on AMD's earnings calls and any announcements about new supply agreements or capacity expansions.
In the meantime, the broader market will continue to watch how the AI supply chain evolves. As industrials and other sectors feel the ripple effects of data center demand, the chip shortage remains a central theme in the tech and investing world.
