Apple knows that it needs to get really serious about AI server chips. There are signs that the company is willing to spend big under John Ternus to buy its way out of its AI problem.
We already know that Apple is looking at deals such as one with PrismML to improve its on-device processing and Siri’s capabilities. Apple also wants to further its server-based AI processing too.
According to an article published on Wednesday by The Information, Apple is showing signs it’s ready to switch acquisition tactics. Instead of deals in the hundreds of millions of dollars, it’s prepared to do more in the range of billions of dollars, rivaling the deals to get Beats and PA Semi.
So far in 2026, this has included an agreement to buy Q.Ai in January. That deal is valued at $2 billion, making it the second largest behind the $3 billion purchase of Beats in 2014.
That Q.Ai deal nets Apple a machine learning company from Israel, that specializes in interpreting speech based on a person’s facial micro-movements. PA Semi ultimately led to advancements in the A-series processor, leading to the full roll-out of Apple Silicon.
It won’t be the only big-money deal on the table. During the second quarter earnings, CFO Kevan Parekh warned that Apple was shifting from its long-term policy of being net cash neutral, balancing debt with cash reserves.
The indication means Apple is more willing to spend those cash reserves on big acquisitions.
New broom, new priorities
Over time, Apple has been bringing its design work in-house, and that includes chips. It’s what eventually resulted in the Apple Silicon line, after all.
However, Apple hasn’t done that many major moves to bump up its chips by acquiring tech and talent that much. That may change.
The article offers little new information, aside from the $2 billion valuation of the Q.ai purchase. But it does use Apple’s history for its main thesis that Apple will be doing something different to deal with AI processing.
It calls out Apple’s purchase of startup PA Semi for $278 million in 2008 as being instrumental to its chip designs. However, there’s been little in the way of massive purchases for chip manufacture and design since then.
The CEO transition to the hardware-centric John Ternus, as well as the promotion of chip chief Johny Srouji into being the hardware engineering overseer, could bring with it a new ethos to acquisitions. Ternus may be more willing to take risks to bolster Apple’s technology by making bigger acquisition plays.
One factor is Apple’s focus on developing chips for consumer use, namely power-efficient devices. This is massively different from making chips intended for servers, like its ongoing Baltra AI server chip project.
The Wednesday article proposed that Apple’s current chip-design knowledge is lacking when it comes to these high-power versions. One way to plug that knowledge hole is to acquire it.
A renewal of its partnership with Broadcom for the production of custom chips is certainly going to help in the short term, at least until 2031. But, Apple could always acquire its way to not need Broadcom’s expertise when it comes to designs after that partnership ends.
Currently, Apple’s AI server strategy seems to involve using the M7 Ultra and its massive AI capabilities as the base, but that chip may not see the light of day until 2029. That’s a long time in the tech industry.
At the moment, Apple is using Nvidia processors to help drive its cloud compute infrastructure. It’s a situation that raises questions about privacy and security, even when it’s the use of Apple-owned servers.
Ideally, Apple’s AI server chip will be its answer to ditching Nvidia in those servers at some point. It will take time to get there, but an acquisition or two could help cut down the wait for a result.
