The race for artificial intelligence is in full swing and is making the fortune of Nvidia, which has become one of the most valuable companies in the world thanks to its dedicated chips. However, Apple remains surprisingly distant from the GPU giant, preferring to rent servers through third parties like Amazon or Microsoft. This situation is not the result of chance, but the legacy of a tumultuous relationship dating back more than 20 years.
A lively start in 2001
While Steve Jobs publicly celebrated the arrival of Nvidia GPUs in Macs in 2001, behind the scenes tells a different story. In a memorable meeting that same year, Jobs directly accused Nvidia of copying technologies owned by Pixar, which he then owned. Faced with a denial from an Nvidia executive who even suggested that it was Pixar that should be sued, Jobs adopted an icy attitude by conspicuously ignoring him for the rest of the meeting.
In the years that followed, the relationship only deteriorated. For the Nvidia teams, Apple represented a particularly demanding client who was not even in their top 10 in terms of revenue. Nvidia CEO Jensen Huang was reluctant to allocate significant resources to Mac projects. On Apple’s side, obtaining the collaboration of Nvidia engineers for quality control was an obstacle course.
The “Bumpgate” incident and the consequences for AI
The breaking point came in 2007-2008 with the “Bumpgate” scandal. The GeForce 8600M GT GPUs powering MacBook Pros presented major failures, also affecting Dell and HP machines. Nvidia procrastinated before admitting liability, forcing Apple to extend the warranty on affected machines. Analysts at the time said that 40% of Macs were affected, that’s huge. This episode provoked the anger of Steve Jobs and partly pushed Apple towards AMD, a competitor more flexible both in terms of prices and in the customization of chips. The situation became even worse in the 2010s when Nvidia attempted to obtain royalties from Apple, Samsung and Qualcomm, suspecting these companies of using its patented technologies in their smartphones. This legal battle, which Nvidia lost to Samsung, further widened the gap with Apple.
Today, this historical distrust is manifesting itself in the field of AI. A revealing example: in 2018, when John Giannandrea took over as head of AI at Apple, he preferred to equip his teams with Google chips rather than directly purchasing Nvidia hardware, despite the latter’s overwhelming domination of the market (70 % to 95% according to Technology Magazine).
Apple is currently developing its own server AI chip in collaboration with Broadcom, called “Baltra”, scheduled for 2026. This chip will be manufactured by TSMC with the N3P process, the same technology that will equip the iPhone 17 Pro. This strategy is part of Apple’s desire for technological independence, already illustrated by the development of Apple Silicon chips.
Interestingly, Nvidia executives today say this rivalry is mostly one-sided, saying they are open to collaboration. Recent partnerships concerning Vision Pro or artificial intelligence prove that their relationship “is not entirely hostile”, as underlined The Information.
