Arm is strategically shifting from a pure IP licensing model to directly producing and selling its own chips, starting with the AGI CPU, to capture more value in the AI era.
The global semiconductor supply chain is critically fragile, with single points of failure like ASML and its German mirror supplier posing a significant risk to the entire industry.
CPUs are essential and irreplaceable for managing, orchestrating, and scheduling advanced AI agentic workloads, a role that GPUs cannot fulfill.
Arm's AGI CPU offers a superior value proposition to competing x86 data center chips, providing twice the performance-per-watt.
While it is strategically important for the United States that Intel succeeds, Arm's foundry decisions will be pragmatic, based on which partner—be it TSMC, Intel, or Samsung—offers the most competitive and cost-effective technology.
Pre-2022
For 33 years, Arm Holdings operated exclusively under an intellectual property licensing business model, providing blueprints for other companies to manufacture chips.
Around 2015
Haas notes this as the period when the value of China's semiconductor imports surpassed its energy imports, highlighting a key driver for initiatives like 'Made in China 2025'.
2021-2022
According to Haas, investment decisions made by foundries during this period, before the AI demand surge, are the root cause of the current shortage in semiconductor manufacturing capacity.
2022-2023
Haas points out that the largest memory suppliers, including Micron, Samsung, and SK Hynix, were all unprofitable, contributing to the current DRAM shortage that is constraining AGI CPU production.
Recent Past (5 weeks)
Haas reveals that the order forecast for Arm's AGI CPU rapidly doubled from $1 billion to $2 billion, indicating a dramatic and recent acceleration in customer demand.
Current
Haas has launched Arm's first-ever chip, the AGI CPU, and is now managing extreme demand that exceeds manufacturing capacity from TSMC and is further constrained by DRAM memory shortages.
▶Arm's Strategic Pivot to AI HardwareMay 2026
Haas is spearheading Arm's historic shift from its 33-year-old IP licensing model to becoming a direct chip supplier with its first product, the AGI CPU. This move is driven by the massive demand from AI workloads and a strategic partnership with Meta, which found it more efficient to buy the chip directly from Arm rather than licensing the IP.
This pivot positions Arm to capture a much larger share of the value chain in the booming AI data center market, but it also introduces new risks related to manufacturing, supply chain management, and direct competition with its own licensees.
▶The Fragility of the Global Semiconductor Supply ChainMay 2026
Haas repeatedly highlights the precarious nature of the semiconductor value chain, pointing to single-source chokepoints like ASML for EUV machines and a German supplier for essential mirrors. He also notes how current capacity shortages stem from pre-AI boom investment decisions and are exacerbated by constraints like DRAM availability.
Haas's focus on these vulnerabilities underscores the strategic importance of supply chain diversification and resilience, justifying Arm's consideration of multiple foundry partners like TSMC, Intel, and Samsung in the US.
▶The CPU's Resurgence in the AI Era
While GPUs dominate AI training, Haas argues that CPUs are indispensable for the management, orchestration, and scheduling of emerging 'agentic' AI workloads. He positions Arm's AGI CPU as uniquely suited for this role, claiming it offers double the performance-per-watt of x86 competitors.
This narrative attempts to carve out a critical, high-value niche for Arm's new products in the AI data center, countering the perception that GPUs are the only chips that matter and directly challenging Intel's x86 dominance.
▶Navigating Geopolitics and Foundry CompetitionMay 2026
Haas balances a stated desire for US-based Intel to succeed with a pragmatic, performance-driven approach to manufacturing, keeping TSMC, Intel, and Samsung as options. He acknowledges the geopolitical drivers behind national semiconductor strategies, such as China's 'Made in China 2025' initiative, which was spurred by its massive import dependency.
Haas's commentary reveals a strategy of geopolitical hedging, leveraging competition between foundries to secure capacity and favorable terms while acknowledging the national security implications of semiconductor manufacturing.