The Chinese business customers of Intel and AMD are recording abnormally long queues when it comes to ordering server central processing units, the process that indicates how far the world-wide artificial intelligence boom is transforming the semiconductor provision chain. Over the past few weeks, both companies have privately advised the Chinese market that stocking of the major server CPUs is limited, and delivery schedules extend far beyond what the market has become used to.
In the case of Intel especially, the company has advised some customers that lead times in delivery of some of the processors may take a time of up to six months. Time delays are not a trifling issue to businesses which operate data centers, cloud systems, or workloads that are intensive in AI. Though special AI accelerators are hitting the news, CPUs continue to play a central and dominant role in computing infrastructure. The absence of them slows down whole deployment plans.
The price has also been translated into the supply strain. People acquainted with the conditions report that the Intel server products in China are currently costing an average of over 10 percent above the normal price, but the end price will majorly depend on the individual customer contracts as well as the long-term supply agreements. To the businesses that are already struggling with the increased prices of memory, storage and energy, the increment in the cost further strains their margins.
China is not an outlier market of Intel. It contributes over one-fifths of the total revenue of the company worldwide, and its demand in the market has been resilient despite the broad geopolitical and trade disputes. The fourth and fifth generation processors of Intel are particularly constrained, as they still widely appear in enterprise servers and cloud data centres. These models are said to be rationed, whereby Intel is prioritizing some of their customers and applications leaving others behind.
The backlog is significant. Orders made many months ago are yet to be fulfilled and certain customers are now being advised to wait up to half a year. Such uncertainty can be disruptive in an industry where infrastructure planning is typically done on regular quarterly cycles. Hardware introductions are delayed, software updates are pushed off and capacity extensions plans are downgraded.
Although the general perception of AMD is that it is somewhat more flexible in the recent years, the same pressures do not spare the company. Chinese clients have also been informed by the company about shortages in supply. The lead times in the delivery of some AMD server products have been said to have stretched to a period of eight to ten weeks. Although its worst-case timelines are shorter than Intel, these delays remain significant enough to cause the projects relying on strict time schedules to go wrong.
What is interesting with this situation is that it extends beyond the common consideration of AI-specific chips such as GPUs. The artificial intelligence infrastructure being deployed on a worldwide scale has created a scramble on the entire computing stack. Although some focus has been on graphics processors and specialized AI accelerators, CPUs are still required to coordinate, manage, and perform general-purpose tasks. With the increase in AI implementation, demand on the mainstream computer also increases with the demand on advanced accelerators.
A good parallel is offered by memory chips. The cost of DRAM and other memory chips has soared with a data center scramble to create an AI-capable architecture. The same trend is currently happening with CPUs, as shortages are an indicator that production capacity is not able to match demand. The messages that have been dispatched to Chinese customers over the past few weeks indicate that such pressures have escalated instead of abating.
Intel has admitted the problem publicly. The company sounded the alarm during its January earnings call, and subsequently in a statement that the rapid AI adoption had necessitated high demand in traditional compute. It pointed out that the inventory levels were projected to be lowest in the first quarter but it further pointed that the company was taking aggressive steps to address the situation. The supply is anticipated to be improved by Intel in the second quarter 2026, which implies that relief will not come immediately. The company wrote that its inventory was lowest in the first quarter, however, it is in the process of getting aggressive and that it will see an increase in supply by the second quarter of this year until 2026.
AMD, in its turn, has provided restatements of the statements expressed in its own earnings call, pointing at the fact that it has increased supply capacities to address high demand. The company has also put a lot of investment in increasing its manufacturing relationships and managing its product mix but these also require time to be reflected on the ground as easily available inventory.
In the context of the larger industry, the episode underscores an age-old lesson: semiconductor supply chains are delicate despite the calamities of the pandemic years past. AI-driven demand cycles are more focused and narrower compared to the conventional enterprise refresh cycles. Once several hyperscalers and cloud providers begin aggressively at once, they can easily leave the suppliers strained.
It is especially difficult in terms of timing, as far as Chinese technology companies are concerned. Most are increasing their investments in local data centres, artificial intelligence research and cloud services. The problem of CPU shortages is a threat of dragging down this momentum particularly to smaller companies that are not capable of making purchases because of their lack of buying power to get priority. Big organizations can absorb price increases and long queues, but new ventures and middle-sized firms might need to either postpone growth or re-architecture their systems based on the hardware they have.
It also has a strategic aspect. The long-term use of imported CPUs at times of shortage is likely to empower the Chinese towards the creation of alternatives. A homegrown processor is not yet competitive at all performance levels, but the long-term uncertainty in supply by foreign company may hasten exploration and implementation in some applications.
Simultaneously, it would be a simplistic notion to perceive the situation solely in the context of geopolitics. The cause of this is not withholding, but global demand. This is a common pressure being experienced by Intel and AMD across the regions, although the effects may be especially pronounced in a market that is as large and dynamically moving as China.
