Arm Achieves Record Financial Growth and Shifts to Production Silicon Strategy with the Launch of Arm AGI CPU for Hyperscaler and AI Markets
Arm published a shareholder letter which confirmed that the company achieved record financial performance during both its 4th quarter and the complete fiscal year which ended on March 31 2026. The company generated quarterly revenue of 1.49 billion which contributed to its total yearly revenue of 4.92 billion. The company achieved its third consecutive year of growth which exceeded 20 percent. The financial disclosure showed that royalty revenue for the complete year reached 2.61 billion because of increasing smartphone and AI usage in both cloud and edge computing environments. The final quarter of the year established a new licensing revenue record at 819 million while the total annual licensing revenue reached 2.31 billion.
The company has transitioned its operational model by launching production silicon manufacturing. Arm Everywhere introduced the Arm AGI CPU at its March 24 event which serves as a dedicated processor for agentic AI infrastructure. The new silicon product delivers double the rack performance of x86 based platforms. Data center operators can achieve capital expense reductions of 10 billion per gigawatt because of this efficiency. The personal superintelligence roadmap will be developed through Meta's partnership with this program which targets 3 billion users.
Market demand for the AGI CPU is already high with 2 billion in orders projected across 2027 and 2028. The CPU is being integrated into accelerator based systems by major technology firms which include Cerebras OpenAI Positron and Rebellions. The European cloud provider Verda will utilize its hardware for AI orchestration. The Arm AGI CPU is now available through vendors such as Lenovo Supermicro ASRock and Quanta who offer commercial systems that support it. The unified software ecosystem between customers who select IP licensing or finished silicon will be maintained through support from NVIDIA Samsung and TSMC together with 50 other companies.
Arm now holds 50% of the CPU compute share among top hyperscalers. Google recently announced it is replacing x86 host processors with custom Arm based Axion CPUs for its next generation TPUs. This shift provides 2x better performance per watt and up to 80% better inference performance per dollar. NVIDIA also introduced Vera which is an Arm based CPU designed for agentic AI. Vera enables 50% faster performance and 2x better efficiency compared to standard x86 CPUs when it works with NVIDIA GPUs. Microsoft has deployed Cobalt CPUs across Azure regions to power workloads for Databricks and Snowflake.
The custom silicon business at AWS which includes Arm based Graviton cores alongside Trainium and Nitro is now running at an annual rate of 20 billion with triple digit growth year over year. Anthropic is utilizing tens of millions of Graviton cores for generative AI workloads. In the enterprise sector SAP is moving core database applications to Arm starting with AWS Graviton and expanding to internal Arm AGI CPUs. Cloudflare is also deploying the architecture across its global network to handle security and traffic management. With 350 billion chips shipped and 22 million developers the company is leveraging its massive footprint to move AI from the cloud to the edge using a common architecture.
