AMD ships MI400 GPUs and Helios rack to challenge Nvidia
New CapabilitiesOpenAI, Meta, Anthropic, Microsoft, and Oracle commit to AMD's AI chips as Helios racks begin shipping
July 23rd, 2026: AMD launches MI400 and Helios at volumeNew here? Follow stories to track developments over time. Create a free account to get updates when stories you care about change.
Overview
AMD began shipping Helios racks in Q3 2026, with OpenAI, Meta, Microsoft, Oracle, and Anthropic all committed as launch customers. A deal with Anthropic on July 22 added 2 gigawatts of MI450 GPU capacity and an AMD equity investment of up to $5 billion, pushing total committed deployments to 14 gigawatts. Each Helios rack lists at $5.25 million, roughly 40% above Nvidia's competing second-generation Rubin rack.
AMD's Q2 results, reported August 4, showed data-center revenue of $6.7 billion, up 107% year over year, on record total revenue of $11.5 billion. The stock fell about 9% after hours. Gross margin came in at 54%, below the 56% analyst target, as Helios ramp costs weighed on near-term profits.
Why it matters
A credible second supplier of AI chips means lower prices and faster availability for the compute that powers nearly every AI product people use.
Questions about this story
No questions yet — be the first to ask.
Key Indicators
Voices
Curated perspectives — historical figures and your fellow readers.
Play
Exploring all sides of a story is often best achieved with Play.
Higher or Lower
A number from this story, against one from elsewhere in the news — guess which is bigger, then keep the chain going. 5 rounds, 3 strikes; a miss costs a strike and resets your streak.
Keyboard: ↓/L lower · ↑/H higher
0 points — sign up to put that on the leaderboard.
Timeline
Order five events from this story, oldest at top. Each in the right slot scores 1 — neighbours within one slot count too. Your previous result — green ✓ for exact slots, yellow ~ for off by one. Cards now in true chronological order.
Sign up to save your score and track a streak across stories.
Connections
Sixteen names from the news. Find the four hidden groups of four. Four mistakes max.
Sign up to keep a daily streak — a new puzzle lands every day.
Exit debate?
Your progress in this debate will be lost.
- 1 Two AI personas square off on this story.
- 2 You predict who'll win each round — correct picks earn XP.
- 3 One crossfire question is yours to fire. Pick it carefully.
Couldn't generate a topic
Select Your Champions
Choose one persona for each side of the debate
DEBATE TOPIC
Choose personas with different perspectives for a more dynamic debate.
Select debater for this side:
No debate personas available right now.
Select debater for this side:
No debate personas available right now.
Who's Got This Round?
Make your prediction before the referee scores
The referee scores both sides on
Round Results
Set the Crossfire
Pick the question both personas must answer in the final round
Debate Oracle! You called every round!
Sharp Instincts! You know your debaters!
The Coin Flip Strategist! Perfectly balanced!
The Contrarian! Bold predictions!
Inverse Genius! Try betting the opposite next time!
XP Breakdown
Prediction History
People Involved
Organizations Involved
The main rival to Nvidia in data-center AI accelerators, now shipping the MI400 series and Helios rack at volume.
The maker of ChatGPT and one of two anchor customers whose commitment validated AMD's AI hardware at scale.
The maker of Claude AI and a major AMD Helios launch customer.
Timeline
October 2025 July 2026
-
AMD launches MI400 and Helios at volume
Latest Product LaunchAt the Advancing AI 2026 keynote in San Francisco, AMD details volume production of the MI400 series, the Helios rack, and the EPYC 'Venice' processor, naming OpenAI, Meta, Microsoft Azure, and Oracle as customers.
-
AMD invests $5 billion in Anthropic and adds 2-gigawatt GPU deal
PartnershipAMD and Anthropic announced a deal for up to 2 gigawatts of MI450 GPU capacity in Helios racks, with AMD committing a $5 billion equity investment in Anthropic. The two companies will also collaborate on using Claude to optimize workloads for AMD's ROCm software stack; the first gigawatt of deployment is planned for the first half of 2027.
-
Meta expands AMD partnership to 6 gigawatts
PartnershipMeta enlarges its AMD commitment to 6 gigawatts, bringing the combined OpenAI and Meta total to 12 gigawatts of planned deployments.
-
AMD reveals full MI400 lineup at CES
Product RevealLisa Su presents five MI400 accelerators and the Helios rack, with the flagship MI455X rated at up to 40 petaflops of FP4 compute. Production is targeted for the second half of 2026.
-
OpenAI commits to 6 gigawatts of AMD GPUs
PartnershipAMD and OpenAI announce a multi-year deal for 6 gigawatts of Instinct GPUs, with warrants for up to 160 million AMD shares. It is the first anchor commitment for the coming MI400.
Historical Context
2 moments from history that rhyme with this story — and how they unfolded.
AMD EPYC breaks Intel's server grip (2017)
AMD launched its first EPYC server processors under Lisa Su, targeting a data-center market where Intel held well over 95 percent share. Early buyers were skeptical of switching. AMD competed on core count, memory, and price.
Adoption was slow at first as customers tested reliability and software support.
Over several years EPYC took more than a third of the server CPU market, proving AMD could unseat an entrenched leader.
The MI400 follows the same playbook against Nvidia that EPYC ran against Intel: a credible challenger chipping away at a near-monopoly over years, not months.
Nvidia's CUDA lock-in builds a moat (2006 onward)
Nvidia released CUDA, a software platform that made its GPUs the default for scientific and later AI computing. Developers built their tools around it. Rivals with competitive hardware still struggled because the software ran on Nvidia.
CUDA drove GPU adoption in research and finance well before the AI boom.
It became Nvidia's deepest advantage, locking in customers even when competing chips matched on raw specs.
AMD's hardware can match Nvidia on memory and compute, but the harder fight is software. Whether developers port workloads to AMD's ROCm platform will decide if MI400 sells.
