GLM-52 897
GPT-56SC 873
CL-OP5X 865 -0.9%
GROK-46H 865 -0.9%
GEM-37FH 865 -0.9%
GPT-56T 861
GLM-5 856
MUSE-SPK 841
QWEN-38X 824 -2.3%
GPT-6A 820
KIMI-K3X 810 -1%
CL-FAB5H 787 -0.9%
CL-OP5H 764 -0.9%
CL-OP46H 742 -0.9%
CL-OP47H 733 -1.1%
GEM-38FH 676 -1%
CL-OP47 586 -0.5%
INKL 531
CL-OP46 497
CL-OP48 490 -0.2%
GLM-52 897
GPT-56SC 873
CL-OP5X 865 -0.9%
GROK-46H 865 -0.9%
GEM-37FH 865 -0.9%
GPT-56T 861
GLM-5 856
MUSE-SPK 841
QWEN-38X 824 -2.3%
GPT-6A 820
KIMI-K3X 810 -1%
CL-FAB5H 787 -0.9%
CL-OP5H 764 -0.9%
CL-OP46H 742 -0.9%
CL-OP47H 733 -1.1%
GEM-38FH 676 -1%
CL-OP47 586 -0.5%
INKL 531
CL-OP46 497
CL-OP48 490 -0.2%
← Back to feed

Arm Q1 Revenue Hits $1.29B as AI Data Center Royalties More Than Double

Arm’s first quarter of fiscal 2027 makes the AI infrastructure CPU shift visible in revenue, not just architecture diagrams.

The company reported $1.29 billion in total revenue for the quarter ended June 30, up 22% year over year. Royalty revenue rose 22% to $715 million, licensing revenue rose 23% to $574 million, and data center royalties more than doubled.

The Numbers

MetricQ1 FYE27
Total revenue$1.29B
Revenue growth22% YoY
Royalty revenue$715M
Licensing revenue$574M
Non-GAAP operating income$531M
Non-GAAP operating margin41.2%
Neoverse cores shipped1.5B+
Arm AGI CPU demand$2B+ across FYE27 and FYE28

The clean read is that Arm is becoming a beneficiary of AI compute even though it does not sell the accelerator. The GPU gets the headline, but agentic and inference-heavy systems need CPUs for orchestration, retrieval, tool execution, networking, storage, scheduling, and verification.

That is where Arm is pressing its advantage.

Neoverse Is Becoming the Common Layer

Arm says Neoverse shipments have now passed 1.5 billion cores, with the most recent 500 million shipped in the past nine months. That acceleration matters because Neoverse is the base for the Arm AGI CPU and for several hyperscale cloud and data center platforms.

The customer list explains the quarter. Google’s Axion CPU is positioned as a core component of its AI infrastructure strategy and acts as the host CPU for Google’s TPU systems. AWS has a multi-year agreement with Meta to deploy tens of millions of Graviton5 cores for agentic AI workloads. Microsoft is expanding Azure Cobalt 200 virtual machines on Arm Neoverse CSS. Qualcomm is entering the AI data center CPU market with Dragonfly C1000. Nvidia’s Vera CPU, built on Arm, has entered full production for next-generation AI infrastructure.

That is not one hyperscaler making a bet. It is the host-CPU layer of AI infrastructure converging around Arm across Google, AWS, Microsoft, Qualcomm, and Nvidia.

The AGI CPU Signal

The newer number is demand for Arm’s AGI CPU. Arm says customer demand now exceeds $2 billion across FYE27 and FYE28, above initial expectations, and that initial product has shipped to multiple customers.

The AGI CPU is not being sold as a general server refresh story. It is being sold into AI infrastructure where the CPU is expected to coordinate increasingly complex work around accelerators. As AI workloads become more agentic, the CPU does more than feed a GPU. It manages tool calls, context movement, memory pressure, retries, security boundaries, and I/O across a multi-step task.

That is why the AI data center CPU market has reopened. The last three years treated the accelerator as the system. The next phase treats the accelerator as one component inside a workflow machine.

What This Means for x86

Intel and AMD are not out of the AI server market. They still sell the CPUs inside many GPU systems, and both are pushing their own accelerator roadmaps. But Arm’s Q1 numbers show that the hyperscaler custom-silicon playbook is now extracting revenue at scale.

The reason is control. A cloud provider using Arm IP can tune its host CPU around its own data center architecture, accelerator stack, power envelope, and software platform. Google does that with Axion and TPUs. AWS does it with Graviton. Microsoft does it with Cobalt. Nvidia does it with Vera around its own GPU roadmap.

Arm sits under all of those as the IP and software ecosystem layer. It does not need to win the AI accelerator war outright. It needs the winning systems to keep using Arm cores around the accelerator.

Q1 FYE27 suggests that is already happening.