The Angstrom Era Begins
TSMC's transition to the A16 (1.6nm) process node marks a defining moment in semiconductor history. Dubbed the start of the "Angstrom Era," A16 represents TSMC's first foray into sub-2nm manufacturing using nanosheet gate-all-around (GAA) transistors combined with the company's Super Power Rail (SPR) backside power delivery architecture. This combination promises an 8-10% speed improvement at the same voltage as N2P (2nm), a 15-20% power reduction at equivalent clock speeds, and up to 1.10x chip density improvement.
The technological leap is significant. Super Power Rail moves the power delivery network to the backside of the wafer, freeing up front-side routing resources for signal interconnects. This is particularly beneficial for high-performance computing and AI chips, which require complex signal routes and dense power delivery networks. TSMC's approach directly competes with Intel's PowerVia backside power architecture, which is a key feature of Intel's 18A and 14A process nodes.
Nvidia Feynman: The Anchor Customer
Nvidia's next-generation AI architecture, codenamed "Feynman" after the renowned physicist Richard Feynman, is positioned to be the lead product on TSMC's A16 node. First hinted at during Nvidia GTC 2026, Feynman is expected to succeed the Vera Rubin architecture and represents Nvidia's most ambitious chip design to date.
Industry sources indicate that Feynman will be a multi-die platform consisting of GPU compute dies, CPU chiplets (codenamed "Rosa"), and custom HBM4 memory stacks, all interconnected through TSMC's advanced packaging technologies. The chip is expected to push power envelopes well beyond 1000W per accelerator, requiring innovative cooling solutions and power delivery architectures.
Nvidia CEO Jensen Huang has described Feynman as "unlike any chip that has come before," hinting at architectural innovations that go beyond simple transistor scaling. While Rubin GPUs are slated for later in 2026 on TSMC's N3 process, and Rubin Ultra follows in 2027, Feynman is targeted for a 2028 launch on the A16 node.
The $54B Capex Commitment
TSMC's capital expenditure for 2026 is projected to reach approximately $54 billion, reflecting the company's aggressive expansion across multiple nodes and geographies. This staggering investment covers nine phases of wafer fabs and advanced packaging facilities under construction simultaneously.
The capex is distributed across several priorities: A16 production capacity in Kaohsiung, Taiwan, and Fab 21 in Arizona; N2 (2nm) ramp with over 20 tape-outs already in progress; CoWoS advanced packaging expansion with 80% CAGR capacity growth from 2022 to 2027; and Co-Packaged Optics (CPO) samples targeting 50% power reduction and 90% latency improvement in datacenter interconnects.
TSMC's confidence is backed by committed customer purchase orders from hyperscalers. The company reported Q1 2026 revenue of $35.9 billion, up 35% year-over-year, with high-performance computing comprising 61% of sales.
Dual-Continent Manufacturing Strategy
For the first time, TSMC is executing a dual-continent manufacturing strategy for its most advanced process node. Initial A16 production will ramp in Kaohsiung, Taiwan, where process maturity and yield optimization can be achieved efficiently. Once trial production stabilizes, capacity will be shared with the Arizona Fab 21 Phase 3 facility.
The Arizona expansion is particularly significant. The first fab is already in production using 4nm technology. Equipment move-in for the second fab is planned for the second half of 2026, while construction of a third fab is underway. Work on a fourth fab and the site's first advanced packaging facility is expected to begin this year. TSMC anticipates a 1.8-fold year-on-year increase in Arizona output by 2026, with yields described as comparable to Taiwan — a meaningful credibility statement given historical concerns about non-Taiwan site yield performance.
Timeline and Volume Projections
TSMC's official roadmap indicates A16 mass production will begin in the second half of 2026, though wafer shipments for Nvidia are expected to ramp significantly in 2027-2028. The technology symposium held in May 2026 confirmed that N2 and A16 capacity is growing at a 70% CAGR through 2028.
For Nvidia, the Feynman platform represents a multi-year bet on TSMC's ability to execute. The company is also hedging its bets by exploring Intel's 18A and 14A nodes as potential secondary sources — a strategy that would diversify its manufacturing base and reduce dependence on a single foundry.
Supply Chain Implications
The Feynman platform is expected to boost Taiwan's entire semiconductor supply chain. Beyond TSMC, major OSAT companies including ASE, SPIL, and KYEC stand to benefit from advanced packaging and testing demand. ABF substrate suppliers such as Nan Ya PCB are expected to see order surges for GPU packaging substrates.
Cooling and power management are also critical enablers. As Feynman pushes power consumption to new heights, companies like Auras Technology, AVC, Delta Electronics, and Lite-On Technology are positioned to capture significant value from the AI server ecosystem.
The Geopolitical Dimension
TSMC's A16 expansion cannot be understood without considering the geopolitical landscape. The CHIPS Act has provided substantial funding for domestic semiconductor manufacturing, and TSMC's Arizona operations are central to U.S. ambitions for supply chain resilience. The completion of Fab 21 Phase 3 will mark the most advanced semiconductor manufacturing facility in the United States, helping to diversify geographic risk and ease concerns about concentrated advanced chip production in Taiwan.
With a market capitalization exceeding $1.87 trillion and a 70%+ share of the global foundry market, TSMC's strategic decisions shape the entire industry. The A16 node and the Feynman platform represent the next chapter in the AI computing revolution, built on a foundation of record capital investment and global manufacturing expansion.