Subscribe
Share
Share
Embed
Quantinuum and RIKEN have upgraded Reimei to H2 and describe research projects combining quantum computing, supercomputers and AI. The episode also covers Diraq’s Sydney data-center deployment, QCi’s quarterly figures, Google’s preprint and IonQ’s manufacturing and evaluation plans.
Chapters
00:00 Today's headlines
00:58 How trapped-ion quantum computing works
02:15 Diraq's eight-qubit computer live in Sydney
03:58 QCi's revenue and acquired businesses
05:37 Google AI-designed programs cut resource use
07:02 IonQ's Gumi manufacturing plan with SDT
08:22 Reimei upgrade to H2 at RIKEN
11:53 Reimei partnership: substance versus ambitions
Source links
Quickly Quantum is an AI-voiced podcast, built and run by a real person. Nothing in this episode is financial advice.
Follow Quickly Quantum: subscribe on YouTube · Apple Podcasts · Spotify
More from Brian Lampert: Concrete Compute, the daily AI infrastructure briefing, and Space Stakes, the business of the new space race. Transcripts and every episode: quickly-quantum.kngoworld.chatgpt.site.
The daily quantum computing briefing you don't need a physics degree to follow. Every day, Quickly Quantum cuts through the hype to bring you the breakthroughs, funding rounds, policy moves, and research that actually matter, with plain-English analysis a smart non-physicist can follow. Hosted by Brian Lampert. AI-voiced, human-built, always skeptical of press releases. Nothing on this show is financial advice.
Also from Brian Lampert: Concrete Compute and Space Stakes.
Quantinuum and RIKEN have upgraded the Reimei quantum computer at RIKEN to System Model H2, The Quantum Insider reports. The partners now plan to combine quantum computing, supercomputing and AI for research in electronics, energy and life sciences, while also exploring access through Japan’s AI data centers. That puts a hardware upgrade beside two bigger ambitions: a fault-tolerant machine developed with Japanese industry, and services businesses might use. What can the upgrade support today, and how much of the commercial path is still exploratory? We’ll separate the work already underway from the plans still being explored. Welcome back to Quickly Quantum, your daily brief on the quantum frontier. It's Tuesday, November 10, 2026. Let's get into it.
Trapped-ion quantum computing stores information in charged atomic particles held in place by electromagnetic fields. Those particles are called ions. The information sits in stable electronic states of each ion, while the ions’ shared motion can transfer quantum information between them. Lasers help control the information: they can operate on an individual qubit, the basic unit of quantum information, or connect that information to the ions’ motion to create entanglement, a link between quantum states. That link is useful because it lets a quantum program work with relationships between qubits, rather than treating each one as an isolated switch. The method has a physical complication built into it. Ions in a shared trap repel one another, and adding ions brings more vibrations into the system. Those vibrations make it harder to prepare the system and carry out a computation. The page describes approaches in development that move ions between separate traps or connect ion chains through photons, particles of light; those approaches aim to extend how the system can be arranged. But the page’s basic picture is already enough to follow the hardware: charged particles are held, controlled with lasers, and made to interact through their shared motion. This explanation helps with the main story later in today’s show.
From the physics of ions to a system placed in a working data center: Diraq says its eight-qubit silicon-spin computer is live inside an Equinix facility in Sydney, Business News Australia reports. A qubit is the quantum equivalent of the bit that carries information in a conventional computer; Diraq builds its qubits in silicon using CMOS manufacturing processes, the processes used to make everyday computer chips. The company is targeting up to 256 qubits in the same machine within the next year. That’s a target, while the system running at the facility now has eight qubits. Business News Australia says the current machine draws less than 20kW and takes up 16sqm of floor space. Diraq expects a future utility-scale system to draw around 300kW and need as little as 5sqm. Those future figures are the company’s expectations, not the present system’s specifications. Diraq says customers can add computing power by upgrading the quantum processor, with minimal changes to the surrounding infrastructure. There’s a real deployment milestone here, but a data-center address doesn’t establish what useful computation the machine can perform. The article says the September announcement described a test of how the system performs in a live commercial data center; industry partners and customers would get a chance to see it in action once testing is complete. Digital Journal posted on X that ‘Australian startup Diraq is conducting the first field test worldwide putting a small quantum computer based on a silicon chip in a commercial data centre in Sydney.’ That’s Digital Journal’s characterization, and the distinction between hosting the system and demonstrating its computational value still matters. I think the achievement worth crediting is the infrastructure test: quantum hardware has to fit somewhere people can access it.
And while Diraq’s headline is a live installation, Quantum Computing Inc.’s headline is revenue: The Quantum Insider reports QCi recorded third-quarter revenue of $9.5 million, exceeding the $9.2 million generated during the entire first half of 2026. The same outlet reports revenue of $384,000 a year earlier, approximately $1.3 billion in cash, cash equivalents and investments at the end of September, and a contract backlog of $39.1 million. The composition changes how I read that revenue. According to The Quantum Insider, acquired businesses generated approximately $8.8 million, or 93%, of the quarter’s revenue. The acquired operations include photonic components, quantum communications equipment and advanced semiconductor packaging. Photonics uses light to process or transmit information. In the company’s unaudited pro forma figures, which assume the acquisitions occurred on Jan. 1, 2025, combined nine-month revenue was $28.4 million in 2026, compared with $44.8 million a year earlier; the corresponding third-quarter decline was approximately 34%, The Quantum Insider reports. So the reported sales grew with the acquisitions, while the comparison that treats those businesses as present in both periods points the other way. QCi CEO Dr. Yuping Huang told the outlet through a company news release, ‘Execution is what makes that opportunity last.’ For me, that’s the right word to carry forward: execution, especially whether the expanded business can turn its reported sales into a stronger quantum product mix.
A different kind of resource story comes from Google researchers: The Quantum Insider reports their AI-designed quantum programs cut combined qubit-and-gate requirements by roughly elevenfold for a fusion-related simulation, eightfold for photochemistry and a 2.5-fold reduction for a drug-metabolism model. The work is described in a study posted on the preprint server arXiv. The cost measure combines the number of logical qubits with the number of T gates. Logical qubits use error correction to protect information from faults in the underlying hardware; T gates are operations that are expensive to implement on error-corrected quantum computers. The researchers used Lean, a system that checks mathematical proofs, to specify what programs had to do and how their resource use would be counted. AI agents proposed programs, and each solution had to include proofs of correctness and cost. That matters because the researchers can check a program without asking a conventional computer to simulate all of its quantum behavior. But the scientific programs still exceed capabilities projected for early fault-tolerant quantum computers, according to The Quantum Insider. The paper says many proposed applications require billions to quadrillions of logical operations, compared with roughly a million operations projected for early fault-tolerant processors. I think this is a useful algorithmic contribution, with a clear boundary: savings on these examples don’t yet put the target calculations within reach of the machines projected for early use.
MK reports that Scott Millard, IonQ’s CEO, said, ‘Korea, which has a solid supply chain and research ecosystem, is a very attractive partner for IonQ.’ The article says IonQ chose SDT as a strategic manufacturing partner in September and decided to build a dedicated manufacturing facility in Gumi. It describes this as the first time IonQ operates a manufacturing facility outside the U.S. mainland. The article also says IonQ agreed in March to link quantum computers to supercomputers at the Korea Institute of Science and Technology Information. Millard described quantum memory modules as devices that can store quantum information and then take it out again if necessary. He said SDT’s manufacturing work requires supply chains including cryogenic and magnetic-field technology. He also argued that specialization and division of labor will expand across the quantum industry, with professional manufacturing partners taking a larger role where processes and quality can be standardized. There’s a two-track strategy in the account: IonQ says its Skywater Technology acquisition in July is intended to increase its U.S. manufacturing capabilities, while it focuses on SDT to enter overseas markets. I’d treat the manufacturing partnership as a stated plan with a clear industrial rationale; the article doesn’t give a schedule for the Gumi facility.
That's the rest of today's quantum news. Now, the story I want to slow down for. The Reimei upgrade is about putting more capable quantum hardware into a hybrid research setup, while Japan’s partners pursue a longer path toward fault tolerance and commercial access. A hybrid platform combines quantum computing with conventional high-performance computing, or HPC, and AI, each contributing different kinds of computation. The Quantum Insider reports that Quantinuum and RIKEN completed Reimei’s upgrade to System Model H2. Reimei is operated as part of the JHPC-quantum project, funded by NEDO. The upgrade, the outlet says, enhances the project’s quantum-HPC platform and enables researchers to investigate more complex quantum systems that are increasingly difficult to simulate using classical computing alone. The partners plan to use quantum computing, supercomputing and AI for work across electronics, energy and life sciences. What does that mean in actual research? The article describes two projects already operating on Reimei before the public launch of its services. In one, the partners use RIKEN’s JHPC-quantum GPU supercomputer, ROQUO, to train an AI model on quantum data generated by Reimei for magnetism simulations. Preliminary results suggest the model can make more accurate predictions than AI models trained on classical simulations alone, and more precise predictions than the quantum computer itself. Those are preliminary results, and the comparison is specifically about the predictions in that work. The second project combines Reimei, RIKEN’s Fugaku supercomputer and NVIDIA’s evolutionary AI agent running on ROQUO to simulate a complex photoactive biomolecule. The aim is unprecedented scale and detail, according to the partners’ announcement. Better understanding of how biomolecules respond to light could inform the design of more efficient solar cells, among other industrial opportunities that could arise from advances in biomolecular engineering. RIKEN’s Division Director, Quantum-HPC Hybrid Platform Division, Dr. Mitsuhisa Sato, said the upgraded Reimei would let researchers tackle more complex and larger-scale scientific problems that were difficult to address with the previous H1 system. He said a newly signed Memorandum of Understanding would pave the way for expanding research into quantum computing for AI for Science. That gives the upgrade a concrete research role: it sits inside a platform already running projects that combine quantum-generated data, AI and supercomputers. There’s also a second, longer-term track. The Quantum Insider reports that Quantinuum and RIKEN will work with Japanese universities and domestic components suppliers toward a fault-tolerant quantum computer developed with Japanese industry. Fault tolerance means protecting computation from errors that arise in the hardware. The partners say they’ll draw on their experience integrating and operating Reimei to inform that development. Quantinuum CEO Dr. Rajeeb Hazra said, ‘Japan’s quantum ambitions require advanced capabilities today and a sustained commitment to developing what comes next.’ The same announcement describes a separate agreement with SoftBank to explore connecting quantum computers directly with AI data centers in Japan. The partners will assess infrastructure readiness and examine how connected systems could support commercial quantum-AI hybrid computing services. This remains an exploration, with the stated aim of broadening access beyond the research community. HPCwire posted on X, ‘Quantinuum and RIKEN Expand Quantum-AI Supercomputing Partnership.’ That captures the research-platform dimension; the SoftBank work adds a possible route toward business access, but the partners are still assessing it.
Here’s the complication: the upgrade is complete, but the fault-tolerant computer and the commercial services are goals the partners are working toward, not capabilities this announcement says are already in operation. That leaves the strongest evidence in the work described: the researchers have begun projects combining Reimei with AI and supercomputers, and the partners say the upgrade enables them to investigate more complex quantum systems. The magnetism result is especially interesting because the AI model trained on Reimei-generated data is reported to make more accurate predictions than models trained on classical simulations alone, and more precise predictions than the quantum computer itself. That doesn’t establish a general advantage across scientific problems; it describes preliminary findings in this simulation. But it does show a plausible division of labor: a quantum device can contribute data to a hybrid workflow even when the final predictive role belongs to an AI model. I think that is the strongest case for this partnership today. The useful unit of progress is the whole workflow, not the quantum processor in isolation: Reimei supplies quantum data, while supercomputers and AI take part in the research. The boundary matters just as much. The release’s language about future fault tolerance and commercial access points to work still ahead. SoftBank, RIKEN and Quantinuum are assessing infrastructure readiness and examining possible services; no launched service is described. The “Made with Japan” computer is also a development goal involving universities and domestic suppliers. Those ambitions make the partnership strategically interesting, while the evidence today is the H2 upgrade and the research projects, not a demonstrated fault-tolerant system or a commercial quantum-AI service. I’m cautiously encouraged by the research examples, especially the magnetism work, because they name a specific role for quantum-generated data within a larger computing process. I’d become more confident as the partners report what those projects establish beyond preliminary results, and as their stated commercial exploration produces evidence about infrastructure readiness and access. The release gives a sensible sequence: use the upgraded platform for research now, draw on integration experience for a future fault-tolerant machine, and assess whether quantum computers can connect to AI data centers. Time for the Hype Check. The completed H2 upgrade and the described research projects give this partnership real substance, while fault tolerance and data-center services remain ambitions still being explored, so I’d rate it a six on the substance scale. Japan’s research projects on Reimei are already underway, and the SoftBank group is assessing infrastructure readiness; the outcome will show whether this partnership can connect those research workflows to broader access.
If you want a clear read on what quantum systems can do today, follow Quickly Quantum wherever you listen. This has been Quickly Quantum, an AI-voiced podcast, created and built by a real human using today's cutting-edge technology. Nothing you heard on this show is financial advice. I'm Brian Lampert, and I'll catch you all tomorrow — take care!
Watching on YouTube? Hit subscribe, and tomorrow's briefing will be waiting for you. I also host Concrete Compute: a daily briefing on the AI buildout. The datacenters, the megawatts, and who actually pays for them. Find it wherever you get your podcasts.