Intel and Advanced Micro Devices are signing longer-term purchase agreements with Chinese server customers as data center processor prices surge, according to a Reuters report citing people familiar with the talks.
Server CPUs have spent years as the dependable, easy-to-source layer of an AI data center, unlike GPUs or memory chips. That status just flipped, and Chinese buyers are now locking in supply without any guarantee on price.
The timing raises the stakes. Intel reports second quarter earnings Thursday, July 23 after the closing bell, and this report lands hours ahead of that call.
Investors will be listening for whether management addresses the pricing dynamic directly or leaves it to analysts to connect the dots.
Related: AMD just landed its biggest AI deal yet
The deals trade price certainty for guaranteed supply
The agreements typically lock in purchase volumes but not prices. Most cover roughly a year of supply, though Intel and AMD have discussed commitments running two years or longer with some customers.
That structure favors the seller, not the buyer. A customer accepting volume terms without price protection is betting that availability matters more than cost.
CPU buyers rarely had to make that trade before this year, which is what makes the shift notable rather than routine.
Intel and AMD are locking Chinese server customers into longer contracts as CPU prices in China climb more than 40% since January.
Cheng Chia Huang / Getty Images
The shortage built quietly, then showed up in the price
Server CPU prices in China have climbed more than 40% since the start of the year, with some products rising over 10% month on month.
Intel’s Xeon server processors have also faced lead times of up to six months for some orders.
Those numbers point to a supply gap that formed well before this week’s contracts became public.
China is one of the world’s largest server markets, driven by rapid construction of data center racks, AI computing clusters and national computing infrastructure, which has intensified competition for both companies’ chips.
This cycle looks different from the 2021 chip shortage, when the bottleneck sat mostly in auto-grade semiconductors and foundry capacity, not general-purpose CPUs. Server processors were rarely the constrained part of an AI build.
Agentic AI workloads changed that math, since they lean on CPUs for orchestration, storage and inference alongside the GPU work that gets most of the attention.
AMD has raised its long-term forecast for the global server CPU market to more than $120 billion by 2030, citing demand tied to agentic AI workloads.
Intel CEO Lip-Bu Tan told analysts in April that demand was running ahead of supply for Xeon chips specifically, pointing to a multiyear agreement with Google as one of several long-term contracts signed in the first quarter.
Wall Street flagged this before the story broke
KeyBanc analyst John Vinh raised his price target on AMD to $725 from $530 and lifted his Intel target to $155 from $110 earlier this month, both moves tied to tightening server CPU capacity. That was before this week’s China reporting surfaced.
Susquehanna analyst Christopher Rolland made a similar call, raising his Intel price target to $115 from $80 while citing stronger-than-anticipated server CPU demand.
Two analysts reaching the same conclusion from different angles is a stronger signal than either call alone.
Back in January, Vinh had already said Intel was largely sold out of server CPUs for 2026 and was considering an average selling price increase of 10% to 15%, according to CNBC.
The China contracts reported this week look less like a new development and more like confirmation of something analysts had already priced in.
More Intel:
- Intel and Google deepen AI ties for chip design
- Intel’s stock split history (& prospects) explained
- Intel CEO gives investors a reality check
Intel (INTC) shares closed on July 22 at $102.62, near their 52-week high of $142.35 set on June 30. The stock barely moved on the July 23 report.
Investors are saving their reaction for Intel’s second quarter results, due after Thursday’s closing bell.
The AI shortage is moving down the stack
This is the same pattern that hit memory chips earlier in the AI buildout, when DRAM and NAND prices spiked as hyperscaler orders outran capacity.
That dynamic has now reached the CPU, the most basic component in any server rack, not just the accelerators getting most of the attention.
Server processors also sit outside the export restrictions Washington has placed on advanced AI GPUs bound for China.
Chinese buyers face a price problem here, not an access problem, which is a distinction most coverage of the chip war overlooks.
Two catalysts will test how durable this pricing power really is. Intel’s earnings call Thursday evening could confirm or complicate the shortage narrative, and AMD reports in early August with its own server CPU numbers due for scrutiny.
For investors, the bigger takeaway is that AI-driven scarcity is no longer a GPU story alone. It is working its way through every layer of the stack, and the next pricing surprise may not come from Nvidia’s side of the market at all.