Within just 16 days, rival researchers in Taiwan and Shenzhen independently announced major breakthroughs in an overlooked hardware architecture. That coincidence hides a much larger geopolitical strategy.
image from grok
Taiwan and China both announced significant silicon-based quantum computing milestones within a three-week window in March 2026, signaling an intensifying infrastructure race. Taiwan's Phase 1 produced both silicon spin qubits and superconducting qubits, positioning quantum as an extension of its semiconductor moat, while China's SZIQA demonstrated the first universal logical operations in silicon using five phosphorus nuclear spins in a donor cluster. Both governments explicitly frame silicon quantum as a strategic national priority tied to existing CMOS fabrication infrastructure rather than chasing more fashionable approaches like superconducting or trapped-ion systems.
Taiwan launched the second phase of its National Strategy for High-Speed Quantum Computing on March 7, 2026, with President William Lai standing at a news conference in Hsinchu Science Park and a public document confirming that Phase 1 had already produced working silicon-based spin qubit chips alongside superconducting qubits, according to the Taipei Times. Three weeks later, on March 23, a team at the Shenzhen International Quantum Academy published results in Nature Nanotechnology showing the first universal logical quantum operations in silicon donor qubits. Both developments landed in the same three-week window. Both declare silicon-based quantum hardware a strategic national priority. That coincidence is the actual story.
We covered the physics of the SZIQA result separately on March 26, as a standalone quantum computing development. This piece is about something else: the infrastructure race that both governments are running, and the gap between that race and what either side has actually demonstrated.
Taiwan's quantum program is explicitly positioned as an extension of its semiconductor moat. The Phase 2 announcement frames quantum computing as a technology that Taiwan's existing chipmaking capabilities can support — a statement that is simultaneously a industrial strategy document and a geopolitical signal. Phase 1, according to the Taipei Times, produced superconducting qubit chips and silicon-based spin qubits, plus low-temperature readout and control modules. That is a wider portfolio than most national quantum programs announce publicly. It is also not a surprise: Taiwan's government has been clear for two years that it views quantum as an extension of the semiconductor supply chain it already controls.
China's framing is different but convergent. The 15th Five-Year Plan, approved in March 2026, lists quantum computing alongside artificial intelligence, 6G, robotics, and brain-machine interfaces as strategic technologies for national economic growth and scientific leadership. SZIQA's Nature Nanotechnology paper — first universal logical operations in silicon, using five phosphorus nuclear spins in a donor cluster — fits inside that plan. Yu He, one of the corresponding authors, has published separately on the advantages that silicon CMOS compatibility would give a national quantum program in terms of fabrication infrastructure already in the ground. The paper's funding structure reflects that framing: China's Quantum Science and Technology major project, the National Natural Science Foundation of China.
The convergence matters for a reason that neither press release says plainly: if silicon-based spin qubits can be manufactured using existing semiconductor fabrication tools, the country that solves wafer-scale production does not need to build a new supply chain. It inherits one. That is the fab compatibility argument, and it is why both governments are treating this as infrastructure and not just science.
The constraint is STM lithography. Scanning tunneling microscopy lithography is what lets you place individual phosphorus donor atoms with sub-nanometer precision — atomic precision in the literal sense. It is also a technique that requires ultrahigh vacuum, atomically clean crystalline surfaces, and process conditions that are routine in a research lab making five-qubit devices and not yet solved at production wafer scale. Neither SZIQA nor any other group has demonstrated wafer-scale STM lithography for quantum computing, as the Nature Nanotechnology paper makes clear: the dominant error source is cross-talk between neighboring donors that exceeds what the error-detecting code can catch. That is an engineering problem, not a physics problem, and it is the same engineering problem every national program working in silicon will eventually hit.
SZIQA's result is real. First universal logical operations in silicon donor qubits is a genuine milestone, as we reported when the paper published. The magic state fidelity at 95.2 percent clears the distillation threshold for fault-tolerant T gates — a real constraint removed. The noise profile (phase-flip dominated) plays to the code's strengths. These are not minor achievements. But they are research achievements, not manufacturing achievements, and the gap between the two is where the national programs will either succeed or spend the next decade and considerable money.
Neither side is where it claims to be. Taiwan's Phase 1 silicon spin qubit result has not been published in a peer-reviewed paper, and Phase 2 timelines are forecasts. SZIQA's published device has five physical qubits and logical coherence times shorter than its physical qubits — an honest result that the institutional framing around it tries to make look more like a product roadmap than a proof of principle. The competitive dynamic is real. The race is also early, narrow, and built on a manufacturing assumption that nobody has yet demonstrated at scale.
What is worth watching is whether either government treats the fab compatibility argument as a reason to actually fund the wafer-scale engineering, or whether both continue to fund it as a political signal while the research groups publish papers and wait for someone else to solve the hard part. The STM lithography problem is the constraint. The question is whether the five-year plans include the engineering budget to match the ambition, or whether this particular infrastructure race runs on press releases until the physics runs out.
Story entered the newsroom
Research completed — 0 sources registered. SAME PAPER as story_5188. Nature Nanotechnology DOI 10.1038/s41565-026-02140-1 (March 23 2026). First universal logical gate set in silicon donor qubi
Draft (713 words)
Reporter revised draft (728 words)
Reporter revised draft based on editorial feedback
Reporter revised draft based on editorial feedback (826 words)
Approved for publication
Headline selected: Silicon qubits — the least fashionable approach — won twice
Published (826 words)
Get the best frontier systems analysis delivered weekly. No spam, no fluff.
Quantum Computing · 18h 55m ago · 2 min read
Quantum Computing · 19h 52m ago · 4 min read
