How did AMD Instinct progress from MI300X to large Helios commitments?
AMD began by unveiling MI300X, which Azure announced as virtual machines, Oracle planned to offer and OEMs showed in systems. It put the product line on an annual upgrade cadence, reported more than $5 billion of Instinct revenue in 2024, and expanded deployments through Oracle, Meta and Microsoft production inference. That record preceded large multigeneration commitments from OpenAI and Meta, a Microsoft plan to scale Helios on Azure, and Anthropic's planned MI450 Helios deployment. AMD then launched MI400 and named Helios customers, after which HPE announced a $1.2 billion Vultr order for AMD Helios systems.
Trace the chain
- 1MI300X unveiledAnnouncement
AMD unveils the Instinct MI300X with 192 GB of HBM3, sampling in Q3
At its Data Center and AI Technology Premiere AMD introduced the MI300X, a GPU-only version of the MI300 family built on the CDNA 3 architecture with up to 192 GB of HBM3 memory, aimed at large language model training and inference, and said it would sample to key customers from the third quarter.
AMD Investor Relations ·
The MI300X is sampling to key customers starting in Q3.
Why it matters here. After AMD introduced the MI300X and said it would sample to key customers from Q3, AMD next announced the accelerator's availability alongside Azure, Oracle and OEM systems.
- 2MI300X launchesLaunch
MI300X and MI300A launched; Microsoft Azure, Oracle and the major OEMs line up
At Advancing AI 2023 AMD announced availability of the MI300X accelerator and the MI300A APU. Microsoft announced Azure virtual machines built on MI300X, Oracle Cloud planned MI300X offerings, and Dell, HPE, Lenovo and Supermicro showed systems using the parts.
AMD Investor Relations ·
Today, AMD announced the availability of the AMD Instinct™ MI300X accelerators – with industry leading memory bandwidth for generative AI and leadership performance for large language model (LLM) training and…
Why it matters here. MI300X availability, Azure virtual machines, planned Oracle offerings and OEM systems were followed by AMD's annual roadmap for MI325X, MI350 and MI400.
- 3Annual cadenceAnnouncement
AMD moves Instinct to an annual cadence: MI325X in Q4 2024, MI350 in 2025, MI400 in 2026
At Computex in Taipei AMD laid out a yearly roadmap for its AI accelerators: the MI325X with 288 GB of HBM3E in the fourth quarter of 2024, the CDNA 4-based MI350 series in 2025, and the MI400 series on a CDNA "Next" architecture in 2026.
AMD Investor Relations ·
New AMD Instinct MI325X accelerator expected to be available in Q4 2024 with up to 288GB of HBM3E memory
Why it matters here. AMD's MI325X, MI350 and MI400 roadmap set out successive accelerator generations before it reported more than $5 billion of Instinct revenue for 2024.
- 4$5 billion in 2024Financial
AMD reports more than $5 billion in 2024 Instinct revenue
AMD said it delivered more than $5 billion of AMD Instinct accelerator revenue during 2024. It said Data Center annual revenue nearly doubled as EPYC adoption accelerated.
AMD ·
Data Center segment annual revenue nearly doubled as EPYC processor adoption accelerated and we delivered more than $5 billion of AMD Instinct accelerator revenue.
Why it matters here. Having reported more than $5 billion in 2024 Instinct revenue, AMD next said MI350 was deploying at Oracle and previewed the Helios rack-scale system.
- 5MI350 and Helios previewLaunch
MI350 series rolling out at Oracle; Helios rack on MI400 previewed for 2026
At Advancing AI 2025 AMD said MI350 series accelerators were already deploying at Oracle Cloud Infrastructure with broad availability in the second half of 2025, and previewed Helios, a rack-scale system built on the coming MI400 series GPUs, Zen 6 EPYC "Venice" CPUs and Pensando "Vulcano" NICs.
AMD Investor Relations ·
AMD demonstrated end-to-end, open-standards rack-scale AI infrastructure—already rolling out with AMD Instinct MI350 Series accelerators, 5th Gen AMD EPYC™ processors and AMD Pensando™ Pollara NICs in hyperscaler…
Why it matters here. AMD said MI350 was deploying at Oracle and previewed the MI400-based Helios system, while Meta next reported broad MI300X deployment for Llama inference.
- 6Meta runs Llama on MI300XAnnouncement
Meta says MI300X is broadly deployed for Llama inference
AMD reported that Meta detailed broad MI300X deployment for Llama 3 and Llama 4 inference. The statement concerned AMD Instinct MI300X deployments.
AMD ·
Meta detailed how Instinct MI300X is broadly deployed for Llama 3 and Llama 4 inference.
Why it matters here. Meta's broad MI300X deployment for Llama inference was followed by Microsoft's report that MI300X powered production proprietary and open-source models on Azure.
- 7Azure production modelsAnnouncement
Microsoft says MI300X powers production models on Azure
AMD reported that Microsoft said Instinct MI300X was powering proprietary and open-source models in production on Azure. The statement concerned Azure production workloads.
AMD ·
Microsoft announced Instinct MI300X is now powering both proprietary and open-source models in production on Azure.
Why it matters here. Microsoft reported MI300X production workloads on Azure; OpenAI next announced a multiyear agreement to deploy 6 gigawatts of AMD Instinct GPUs across generations.
- 8OpenAI: 6 gigawattsAnnouncement
OpenAI agrees to deploy 6 gigawatts of AMD GPUs; AMD issues a warrant for up to 160 million shares
AMD and OpenAI announced a multi-year agreement under which OpenAI will deploy 6 gigawatts of AMD Instinct GPUs across generations, starting with 1 gigawatt of MI450 GPUs in the second half of 2026. AMD issued OpenAI a warrant for up to 160 million shares at US$0.01 that vests with deployment milestones and AMD share-price targets.
AMD Investor Relations ·
The first 1 gigawatt deployment of AMD Instinct MI450 GPUs is set to begin in the second half of 2026.
Why it matters here. OpenAI's multiyear agreement covered 6 gigawatts of Instinct GPUs across generations, and Meta next agreed to deploy up to 6 gigawatts using a custom MI450-based GPU.
- 9Meta: 6 gigawattsAnnouncement
Meta signs for up to 6 gigawatts of AMD GPUs with a custom MI450-based part and a matching warrant
AMD and Meta agreed a multi-year, multi-generation partnership to deploy up to 6 gigawatts of Instinct GPUs. Shipments for the first gigawatt were expected to begin in the second half of 2026 using a custom GPU based on the MI450 architecture, and AMD issued Meta a performance-based warrant for up to 160 million shares.
AMD Investor Relations ·
AMD has issued Meta a performance-based warrant for up to 160 million shares of AMD common stock, structured to vest as specific milestones associated with Instinct GPU shipments are achieved.
Why it matters here. Meta's agreement for up to 6 gigawatts, including a custom MI450-based GPU, was followed by Microsoft's commitment to ramp Helios at scale on Azure.
- 10Microsoft ramps HeliosAnnouncement
Microsoft commits to ramp Helios at scale on Azure
AMD said Microsoft would ramp AMD Helios at scale on Azure. The deployment was intended to power frontier-model inference for Microsoft, its AI customers and Azure AI services.
AMD ·
Microsoft to ramp AMD Helios™ at scale on Azure to power frontier model inference for Microsoft, its AI customers and Azure AI services.
Why it matters here. Microsoft said it would ramp Helios at scale on Azure for frontier-model inference, and Anthropic next announced plans to deploy up to 2 gigawatts of MI450 GPUs in Helios systems.
- 11Anthropic: 2 gigawattsAnnouncement
Anthropic to deploy up to 2 gigawatts of MI450 GPUs in Helios, starting in the first half of 2027
AMD and Anthropic announced a partnership to deploy up to 2 gigawatts of Instinct MI450 series GPUs in AMD Helios rack-scale systems, with the first gigawatt beginning in the first half of 2027.
AMD Investor Relations ·
with deployment of the first gigawatt beginning in the first half of 2027
Why it matters here. Anthropic's planned up-to-2-gigawatt MI450 deployment in Helios was followed by AMD's MI400 and Helios launch, which named Anthropic among the organisations choosing Helios.
Helios commitments
- 12MI400 and Helios launchLaunch
MI400 series launched at Advancing AI 2026; Helios customers named
AMD launched the Instinct MI400 series, with the MI455X in Helios racks for frontier AI and the MI430X for sovereign AI and HPC, and named OpenAI, Anthropic, Meta, Microsoft, Oracle, HUMAIN, Tensorwave, Vultr and Cirrascale among the organisations choosing Helios.
AMD Investor Relations ·
Leading AI labs and cloud providers are choosing AMD Helios for its open, full-stack performance.
Why it matters here. The MI400 and Helios launch named Vultr among the organisations choosing Helios, preceding HPE's announcement of a Vultr order for Helios systems.
- 13Vultr Helios orderAnnouncement
HPE announces $1.2 billion Vultr order for AMD Helios systems
Hewlett Packard Enterprise announced a $1.2 billion order from Vultr to deploy AMD Helios AI Rack by HPE systems at Vultr's cloud data centres in the United States.
Hewlett Packard Enterprise via Business Wire ·
HPE (NYSE: HPE) today announced a $1.2 billion order from Vultr, the world’s largest privately-held cloud infrastructure company, to deploy AMD Helios™ AI Rack by HPE systems at Vultr’s cloud data center locations in…
Why it matters here. HPE's announcement of a $1.2 billion Vultr order showed a disclosed cloud deployment commitment for AMD Helios systems.
Changes to this chain
- Proposal #185 accepted (publish:amd-instinct-ai-gpus): replace step 12: [cause] MI400 and Helios launch -> amd-instinct-ai-gpus#110: The MI400 and Helios launch named Vultr among the organisations choosing Helios, preceding HPE's announcement of a Vultr order for Helios systems.; insert after 12: [cause] Vultr Helios order -> amd-instinct-ai-gpus#1756: HPE's announcement of a $1.2 billion Vultr order showed a disclosed cloud deployment commitment for AMD Helios systems.; answer: AMD began by unveiling MI300X, which Azure announced as virtual machines, Oracle planned to offer and OEMs showed in systems. It put the product line on an annual upgrade cadence, reported more than $5 billion of Instinct revenue in 2024, and expanded deployments through Oracle, Meta and Microsoft production inference. That record preceded large multigeneration commitments from OpenAI and Meta, a Microsoft plan to scale Helios on Azure, and Anthropic's planned MI450 Helios deployment. AMD then launched MI400 and named Helios customers, after which HPE announced a $1.2 billion Vultr order for AMD Helios systems.
- Proposal #144 accepted (verify:v2): why of step 11 -> Anthropic's planned up-to-2-gigawatt MI450 deployment in Helios was followed by AMD's MI400 and Helios launch, which named Anthropic among the organisations choosing Helios.
- Proposal #10 accepted (verify:v1): why of step 1 -> After AMD introduced the MI300X and said it would sample to key customers from Q3, AMD next announced the accelerator's availability alongside Azure, Oracle and OEM systems.; why of step 2 -> MI300X availability, Azure virtual machines, planned Oracle offerings and OEM systems were followed by AMD's annual roadmap for MI325X, MI350 and MI400.; why of step 3 -> AMD's MI325X, MI350 and MI400 roadmap set out successive accelerator generations before it reported more than $5 billion of Instinct revenue for 2024.; why of step 4 -> Having reported more than $5 billion in 2024 Instinct revenue, AMD next said MI350 was deploying at Oracle and previewed the Helios rack-scale system.; why of step 5 -> AMD said MI350 was deploying at Oracle and previewed the MI400-based Helios system, while Meta next reported broad MI300X deployment for Llama inference.; why of step 6 -> Meta's broad MI300X deployment for Llama inference was followed by Microsoft's report that MI300X powered production proprietary and open-source models on Azure.; why of step 7 -> Microsoft reported MI300X production workloads on Azure; OpenAI next announced a multiyear agreement to deploy 6 gigawatts of AMD Instinct GPUs across generations.; why of step 8 -> OpenAI's multiyear agreement covered 6 gigawatts of Instinct GPUs across generations, and Meta next agreed to deploy up to 6 gigawatts using a custom MI450-based GPU.; why of step 9 -> Meta's agreement for up to 6 gigawatts, including a custom MI450-based GPU, was followed by Microsoft's commitment to ramp Helios at scale on Azure.; why of step 10 -> Microsoft said it would ramp Helios at scale on Azure for frontier-model inference, and Anthropic next announced plans to deploy up to 2 gigawatts of MI450 GPUs in Helios systems.
- Step labels rewritten by the editor
Timelines in this chain
- AMD Instinct: From MI300X to Helios and the Gigawatt Deals
- HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4E
- US Export Controls on Nvidia AI Chips to China: A100 to H20 to H200
- Nvidia Data Center GPUs: Hopper, Blackwell and Vera Rubin
A chain is written by the editor from entries already published on the timelines above, each of which was read against its source before publication. The chain changes when those entries do. Data and terms.