Thursday, September 17, 2026
Tech Beat
Sep 16, 2026, 10:02 PMEnterprise AI Hardware

Apple Eyes 2029 AI Server With M8 Ultra Chips and Nvidia NVLink

Apple is developing a 2029 AI server with two or four M8 Ultra chips, possibly linked by Nvidia NVLink Fusion, as Mac demand strains global memory supply.

Listen to this briefingAudio briefing

Summary

Apple is developing an enterprise AI server for an expected 2029 release, its first server in nearly two decades following the January 2011 retirement of Xserve. Two configurations would carry either two or four future M8 Ultra chips, extending the high-performance M-series Ultra silicon used in Mac desktops. The project began a year ago with support from then hardware engineering chief John Ternus, who succeeded Tim Cook as Apple CEO on September 1, 2026, after Cook’s 15 years in charge.

Booming Mac mini and Mac Studio sales show demand from AI developers. OpenAI has bought tens of thousands of the computers to train AI agents through trial-and-error reinforcement learning, while Anthropic has rented Mac minis from Amazon Web Services. Apple and Nvidia have discussed using Nvidia’s NVLink Fusion data center networking to connect M8 chips, although Apple could cancel the server or proceed without Nvidia. The plan also faces an industrywide memory chip shortage caused by rapid AI data center construction, which has raised prices for gaming devices, flagship smartphones and numerous Apple products.

Positives

  • Two planned configurations would offer either two or four future M8 Ultra chips for enterprise AI workloads.
  • OpenAI has bought tens of thousands of Mac minis and Mac Studios to train AI agents through reinforcement learning.
  • Anthropic has rented Mac minis from Amazon Web Services, demonstrating cloud demand for Apple hardware.
  • Nvidia’s NVLink Fusion could connect Apple’s M8 chips using established data center networking technology.
  • A 2029 launch would return Apple to enterprise servers nearly two decades after Xserve’s retirement.

Risks & concerns

  • Apple could cancel the AI server before its expected 2029 release.
  • Nvidia’s NVLink Fusion may be excluded even if the server project continues.
  • The memory chip shortage threatens Apple’s server plans and the Mac computers already attracting AI developers.
  • Memory constraints have increased prices across gaming devices, flagship smartphones and numerous Apple products.
Primary sourceAI - Ars Technicahttps://arstechnica.com/ai/2026/09/apple-reportedly-building-server-packed-with-m-series-ultra-chips-for-ai/
Read full article
Editorial note: Tech Beat summarizes and analyzes third-party reporting. The source link is the authoritative article. This page does not reproduce the full source text.

More From The Wire

Venture CapitalSep 16

Valor Gives Investors $8.5 Billion in SpaceX Stock, Avoids Market Sale

Artificial Intelligence and ClimateSep 16

Al Gore Says AI’s Biggest Data Center Risk Is Not Emissions

TransportationSep 16

House Passes Bill to Mandate Free AM Radio in New US Vehicles