Apple is working on its first server chip for artificial intelligence, a project meant to support the heavy computing needs behind its newer AI features. The chip is known internally as Baltra, and Apple is aiming to start mass production by 2026.
What Apple Is Building
Baltra is described as Apple's first server chip for artificial intelligence. Unlike chips designed for individual devices, this project is focused on the server side of AI, where large amounts of computation are needed to run and scale services.
According to The Information, Apple is working with Broadcom on the chip's network technology. That detail matters because AI server infrastructure depends not only on raw processing power, but also on how efficiently systems can move data between components.
The source article says Baltra will handle the high computational demands of Apple's new AI features. Those features sit under the Apple Intelligence name, making the chip part of a broader effort to support AI services across Apple's ecosystem.
Why Baltra Matters For Apple Intelligence
Apple Intelligence is expected to reach a very large installed base. The source states that Apple aims to reduce costs and scale AI services for billions of devices using Apple Intelligence.
That goal helps explain why Apple would develop a dedicated AI server chip. If the company can design infrastructure around its own needs, it can tune the chip for the workloads its services require instead of relying only on more general options.
The article does not say which specific Apple Intelligence features Baltra will support. What it does make clear is the strategic direction: Apple wants server hardware capable of handling the computational pressure created by new AI features at massive scale.
Who Is Involved In The Project
Apple's silicon design team in Israel is leading development of Baltra. That places the project within Apple's broader internal chip design work, rather than presenting it as a purely external supplier effort.
Broadcom's role is tied to network technology, according to The Information. The source does not provide further detail on the arrangement, but it identifies Broadcom as a key partner in a part of the chip that is important for AI server systems.
Apple also plans to manufacture the chip using TSMC's advanced N3P production process. The source identifies TSMC as the manufacturing partner and N3P as the process Apple intends to use.
The Timeline Apple Is Targeting
Apple intends to complete Baltra's initial design within 12 months. That target comes before the broader goal of starting mass production by 2026.
The sequence suggests a project that is still in development, with design work remaining before manufacturing can begin at scale. The source does not provide a launch date for services powered by Baltra, nor does it specify when the chip would begin supporting Apple Intelligence in production.
For now, the most concrete milestones are the 12-month design target and the aim to start mass production by 2026. Those dates frame Baltra as a near-term infrastructure project rather than a distant research effort.
The Bigger Picture
Baltra shows Apple moving deeper into the infrastructure layer behind AI. The company already has new AI features under the Apple Intelligence banner, and the source says the chip is meant to support the computational demands those features create.
The practical stakes are cost and scale. Apple wants to reduce costs while scaling AI services for billions of devices, and a dedicated server chip is one way to align hardware design with that goal.
The available details are still limited. But the core picture is clear: Apple is developing an AI server chip called Baltra, working with Broadcom on network technology, relying on TSMC's N3P process, and aiming for mass production by 2026.