AMD's Advancing AI 2026 brings Zen 6 EPYC Venice launch and Azure rack deal
For three years, “AI chips” meant one word: Nvidia. This week in San Francisco, AMD gets its best shot at changing that.
What is confirmed
Advancing AI 2026, AMD’s flagship event, runs July 22-23 with Lisa Su on the keynote stage. The confirmed slate is already significant: the commercial launch of EPYC Venice server processors on the new Zen 6 architecture, plus an update on the Instinct MI455X and MI500 accelerator roadmap.
Azure’s first rack-scale bet on AMD
The bigger news came from Microsoft. Azure publicly committed to deploying AMD’s Helios racks, 72 MI455X GPUs per rack, for AI inference, with new VM families arriving in the second half of 2026. It is Azure’s first public bet on AMD racks at this scale.
That commitment matters more than a benchmark slide ever could. Hyperscaler rack deployments are where AI chip market share is actually decided, and until now that game has been effectively a one-vendor sport.
The rumor everyone came for
Analysts expect a major new customer announcement on stage, with Anthropic flagged as the most likely name. If a frontier lab publicly commits to AMD silicon, the “Nvidia has no real competition” era officially ends: not because Nvidia stops leading, but because the largest buyers demonstrate they have a credible second source.
Why this reaches your subscription price
Chip competition is not an industry inside story. Cheaper compute means cheaper AI for everyone downstream: training runs, inference serving, and eventually the price of the chatbot subscription in your pocket. Those economics are quietly decided in rooms like the one in San Francisco this week.
By the end of the event, the question will have an answer: did the monopoly crack, or does “AI chips” still mean one word?
Sources
ANOTHER News is published by ANOTHER, an AI-native content agency. Daily coverage also runs on Instagram.
Related
DFSX claims an HBM-free AI chip with double NVIDIA's system bandwidth on 14nm
Dongfang Suanxin stacks DRAM directly on top of the compute die with 3D hybrid bonding, so its chips need no HBM, the memory hardest for Chinese firms to obtain under export controls. The company claims a TY64 SuperNode built from 14nm DF2000 chips delivers 960 TB/s of memory bandwidth against 576 TB/s for NVIDIA's GB200 NVL72. The numbers are self-reported and unverified.
Musk's Davos warning that AI will run out of electricity is aging into consensus
Speaking at the World Economic Forum on January 22, 2026, Elon Musk said electrical power is fundamentally the limiting factor for AI deployment, warning that with US generating capacity growing only 3 to 4 percent a year, the industry could soon produce more chips than it can turn on. Months later, data-center power deals and grid bottlenecks dominate the AI infrastructure conversation.
HRL's 18-qubit silicon chip runs its own error detection inside the cryostat
On July 29, 2026, HRL Laboratories published in Nature an 18-qubit silicon spin quantum processor that runs itself, with no racks of room-temperature control electronics. A custom cryogenic CMOS controller with 70 million transistors, drawing under 3.5 watts at the 4 Kelvin stage, runs quantum error detection with zero room-temperature latency while the qubits sit at about 150 millikelvin. Control errors came in about 10 times lower than previous demonstrations for this qubit type, with roughly fivefold error suppression as more qubits joined the code.