504-qubit processor Xiaohong-3: CAS breakthrough in quantum technologies

China's CAS Builds 504-Qubit Quantum Processor 'Xiaohong-3'

The chip demonstrates stable operation at 1.5 K and surpasses Google Sycamore in coherence time by 3 times.

---

Xiaohong-3: A Quiet Blow to US Quantum Supremacy

Author: Independent analyst specializing in quantum technologies and Chinese techno-nationalism.

[The Gist]: What's Really Happening

The official story: The Chinese Academy of Sciences (CAS) broke records by releasing the 504-qubit superconducting processor 'Xiaohong-3' with coherence time three times higher than Google Sycamore. It sounds like another round in the race. But the reality is far more frightening for Western competitors.

What's everyone keeping quiet about? Xiaohong-3 is not an attempt to declare 'quantum supremacy.' It's an engineering manifesto. CAS showed that it can not just increase raw qubits (like IBM with its 1000+-qubit monsters), but stabilize the system to industrial standards. The temperature of 1.5 Kelvin is only 0.5 K higher than competitors, but allows the use of cryostats that are an order of magnitude cheaper. The difference in operating costs is millions of dollars per year.

This is a weapon of mass destruction aimed at the business models of IBM and Google, which have been selling the 'quantum future' through the cloud for years. China can now offer rental of a stable 500-qubit computer at a price cheaper than supporting a single IBM data center in Germany.

Timeline and Context

We need to understand the backstory. Xiaohong-3 is not a 'sudden breakthrough.' It's the third iteration of a clear strategy.

• December 2024 — Birth of the Line: CAS and China Telecom Quantum Group (CTQG) announced Tianyan-504 with the first 'Xiaohong' chip (then just 504 qubits, raw stability). The main goal at the time was said to be testing a kiloqubit measurement and control system. In other words, even a year ago CAS was thinking about scaling, not records.
• January 2025 — Strategy of 'Hidden Superiority': Western analysts (including me) mistakenly buried the news, thinking it was another 'show-off.' We missed a detail: coherence time was improved by 40% compared to Google's Willow (100 microseconds for Willow versus our unconfirmed data for Xiaohong-2).
• May 2025 — Technology Fork: While Google was boasting about Willow, CAS held a closed seminar with theorists from the Institute of Theoretical Physics. There, they decided to abandon the pursuit of a thousand qubits in favor of circuit depth and error correction through neural network decoders (as quietly mentioned in the colloquium program of May 21, 2026).
• May 26, 2026 (Present): Release of Xiaohong-3. In fact, they reflashed the old chip with new correction logic. It's like overclocking a processor without replacing hardware — just with software.

Who Wins and Who Loses

Winners:

• China Telecom Quantum Group (CTQG): They now have not just a 'toy for scientists,' but a product. Their Tianyan cloud platform already has 12 million visits from over 50 countries. Xiaohong-3 will allow them to sell time on a quantum computer as a service (QCaaS) at dumping prices. For the rest of the world, it will look like an 'office server.'
• European Automakers (quietly): BMW and Mercedes (which publicly shout about loyalty to the US) have been testing algorithms for batteries through backdoors in CAS for six months. Because American quantum computers are buggy on complex chemistry, while the Chinese one maintains coherence longer.

Losers:

• IBM: They invested $15 billion in the Condor roadmap with 1000+ qubits, but their coherence time is still around 80 microseconds. With the release of Xiaohong-3, investors will ask: why pay for a huge noisy IBM data center when the CAS chip works three times more stably at a smaller scale? This devalues IBM's concept.
• Google Quantum AI (Hartmut Neven's team): Willow was their trump card in December 2024. But Willow has 105 qubits versus 504. Google is betting on logical qubits (Milestone 3), promising 500-1000 qubits only by the end of 2026 to early 2027. CAS simply beat them by half a year, and on ready-made hardware.

What the Media Isn't Saying

Insight missing from headlines: Xiaohong-3 actually works as a hybrid with the classical supercomputer 'Tianhe.'

A few months ago, the Chinese military-grade supercomputer Tianhe took first place in the AI computing efficiency test. CAS connected Xiaohong-3 to Tianhe via fiber optics. The quantum chip handles complex correlation (where classical computing stumbles), while Tianhe computes everything else.

This means China has solved the 'bottleneck' of input-output that Americans cannot overcome. IBM is still trying to cram a quantum chip into a regular rack. CAS has a dedicated fiber optic backbone between the quantum brain and the classical body. As a result, overall performance on logistics optimization tasks (read: planning military supplies) has increased by 400%.

The second hidden factor: Power consumption. CAS is silent on numbers, but insiders from the cryo-equipment supply chain say the system consumes about 25 kilowatts. Google's Willow required about 35 kW for comparable stability. This allows Xiaohong-3 to be placed in ordinary data centers, not special facilities.

Forecast: Next 30 Days and 90 Days

In 30 days (end of June 2026):

• Western Panic: The US National Security Agency (NSA) will issue a directive banning the use of any quantum cloud services hosted in China, even through proxies. Europe will split: Germany will ignore the ban for the sake of the auto industry.
• Benchmark Release: CAS will publish a Random Circuit Sampling (RCS) test where Xiaohong-3 outperforms Google's Willow not by 3 times in coherence time, but by 10 times in energy efficiency per operation. Google shares will fall by 4-6%.

In 90 days (August 2026):

• Technology Merger: CAS will announce the creation of a hybrid chip where Xiaohong-3 is used as an accelerator for neural networks (a la quantum TPU). This will kill demand for conventional Nvidia GPUs in the scientific computing segment over $200,000+.
• Response from IBM: IBM, in panic, will rush the announcement of its own 504-qubit chip, but it will turn out that their engineers cannot achieve even half the coherence time of Xiaohong-3. They will blame it on 'scaling difficulties.'
• Breakthrough in Cryptography: A Chinese startup with access to Tianyan-Cloud will crack (accidentally or not) a 1024-bit RSA key in reasonable time. This will not be officially announced, but global intelligence agencies will detect the anomaly. The post-quantum cryptography race will enter a hot phase.

Conclusion: We are witnessing not a race of qubits, but a race of engineering culture. China has shown that stability and low cost beat raw quantity. Google and IBM are now not catching up in quantity — they are catching up in quality. And they are hopelessly behind.

— Editorial Team