Analyst Ming-Chi Kuo insists Apple will double down on artificial intelligence, by mass-producing specialized AI server chips in 2026 and filling new AI data centers with it in 2027.
Artificial Intelligence is a big part of the modern tech landscape, and Apple is no different to others in spending heavily in the field. Part of that expenditure is on servers and infrastructure, and Apple's efforts will be quite visible in the coming year if an analyst is correct.
TF Securities analyst and long-time Apple observer Ming-Chi Kuo posted to X on Tuesday about a podcast appearance, which included a discussion about Apple's AI work. In his notes, he forecasts that Apple's own-designed AI chips for servers will finally enter mass production in the second half of 2026.
Kuo goes on to add that the chips will quickly find a use in Apple's empire. Data centers that will use those chips will start to be constructed in 2027, and will be set to start operation in 2027.
Based on this timeline, Kuo believes this means Apple anticipates a significant growth in on-device AI demand in 2027.
In rumor circles, Kuo is well regarded for his various claims and speculative comments, and is generally considered an accurate source of information thanks to his regular supply chain checks. However, just as any other analyst or leaker, there's always the possibility that what Kuo predicts could be delayed or incorrect.
AI servers
Apple's intentions to build out its data center infrastructure are already well known. In February 2025, Apple announced a four-year, $500 billion investment plan in U.S. manufacturing. The effort would create 20,000 new American jobs.
Part of the announcement involved a facility in Houston, Texas, specifically made for the production of AI servers. Intended for use as part of Apple's Private Cloud Compute system, the AI servers were previously manufactured outside the United States.
At the time, Apple also said that it would continue "expanding data center capacity" in North Carolina, Iowa, Oregon, Arizona, and Nevada.
It is unclear but it seems Kuo's claims of data center construction in 2027 may be for new data centers specifically made for AI processing. The effort is on top of Apple's existing data center collection.
By October, Apple confirmed that the Houston plant was up and running, and had started to ship American-made servers built to power Apple Intelligence. The shipments were ahead of schedule, allowing Apple to implement the hardware sooner rather than later.
AI chips and existing processing options
Apple is much more secretive about its AI server chip production, but there have been reports about that too.
In May 2024, this took the form of an internal project codenamed "ACDC," which involved developing Apple Silicon specifically for AI server farm processing.
By December of the same year, Apple had reportedly partnered with Broadcom to create a processor called Baltra for release in 2026. While Apple is developing its own Apple Silicon processors and probably doesn't need much help with that, the work with Broadcom could be in creating "chiplets," smaller sections that are combined to make a larger single chip.
While Apple is working on chips specifically intended for AI server processing, its existing AI server work is already using Apple's own chip designs.
In September 2024, software chief Craig Federighi explained that Private Cloud Compute relied on producing servers that used Apple Silicon. The decision enabled Private Cloud Compute to deal with queries in a way that maintained security with end-to-end encryption.
A switch to chips specifically designed for AI servers should provide more benefits to Apple, including performance improvements, power savings, and thermal management considerations beyond the capabilities of normal Apple Silicon chips.
