Apple has struck a major deal with Google to support future AI features. At the same time, new reporting suggests Apple plans to ramp up its own AI infrastructure. The shift points to a bigger strategy: use outside help now, while building more control over core AI hardware for the long term.
Apple has faced criticism for moving slowly on AI and for recent software issues. Still, one part of the company keeps delivering: its chip team. That group has pushed iPhone, iPad, and Mac performance forward year after year. It also helped Apple bring more key components in-house, including cellular modems like C1 and C1X, plus a wireless connectivity chip called N1.
Now Apple appears to be applying that same playbook to AI servers.
Kuo points to a 2026 chip timeline
Ming Chi Kuo wrote about the timeline in a post on X.
“Apple’s self-developed AI server chips are expected to enter mass production in 2H26, and its own data centers are expected to begin construction and operation in 2027, which may indicate that Apple anticipates significant growth in on-device AI demand starting in 2027.”
That suggests a two-step rollout. First, Apple gets the server chips into production in the second half of 2026. Next, it brings new data centers online in 2027.
Why Apple would lean on Google now, then push its own hardware
Ming Chi Kuo also framed the Google partnership as a short-term move driven by pressure.
“Two short-term challenges in Apple developing its own AI, which force Apple to collaborate with Google:
(1) This year’s WWDC AI demonstration cannot afford to fail again.
(2) User expectations for AI service quality have risen with the development of cloud AI.”
That view ties the collaboration to execution risk and rising expectations. Cloud AI has raised the bar, so Apple needs to show progress fast. At the same time, Apple still wants deeper control, because AI will sit at the center of devices, software, and the user experience.
What happens between now and 2027
If Apple starts mass-producing AI server chips in 2026, it will likely not wait for brand-new data centers to use them. Apple can deploy early batches inside existing facilities at a smaller scale, then expand once new sites come online in 2027.
The big takeaway stays consistent. Apple is partnering with Google to cover near-term gaps, while its silicon team works toward a future where Apple owns more of the AI stack.
