Latest / Quickly Quantum / NEC Exits Its Quantum Hardware Program as Fujitsu Bets on Diamond Spin — September 8, 2026
Transcript
- Today on Quickly Quantum: one Japanese tech giant just walked away from building its own quantum computer — while its biggest domestic rival unveiled a brand new one, built out of diamond. Same country, same week, completely opposite bets. So which one's the smarter play? That's exactly what we're digging into today. Before that, in the headlines: Singapore and Luxembourg are teaming up on quantum, AI, and space, and a University of Surrey team says a superfluid helium qubit could cut errors a hundredfold — on paper, anyway. Welcome back to Quickly Quantum, your daily brief on the quantum frontier. It's Tuesday, September 8, 2026. This is one of those days where two stories are genuinely fighting for the top slot, so we're running them both, back to back, then asking what they add up to together. Let's get into it. Singapore and Luxembourg are looking to get closer on quantum computing, AI, and space and satellite technology. Singapore's prime minister, Lawrence Wong, raised the plan during an official lunch with Luxembourg's prime minister, Luc Frieden, on September 7th — Frieden's first visit to Singapore since taking office. Luxembourg opened its first resident embassy in Singapore, a move Wong called a signal of deeper engagement with Southeast Asia. Here's the question to watch: which concrete projects and commitments follow this announcement? This is two small, trade-dependent financial hubs betting they can lean on each other rather than try to out-invest the US, China, or the EU alone. Whether an actual joint project follows the handshake is the real test, and it's one you and I will have to wait on. Here's a much stranger one: Quantum Zeitgeist reports that researchers at the University of Surrey have proposed a qubit — quantum computing's basic unit of information — built from superfluid helium-3, a liquid state so slippery it flows without any friction at all. If you're picturing liquid helium doing something strange, you're right. They're calling it SHOQ, the Superfluid Helium Oscillator Quantum, and Quantum Zeitgeist reports the team's calculations predict error rates a hundred times lower than a standard superconducting qubit. So does frictionless liquid helium actually make a better qubit than metal circuits on a chip? The math says maybe — the trick is that helium is charge-neutral, so the qubit should be naturally shielded from the electromagnetic noise that trips up conventional chips. Now, a caveat: this is a proposed design backed by calculations; those predictions still need experimental testing. Our main story today: NEC's exit — and the question it leaves hanging over every other quantum hardware roadmap in the field. According to Nikkei Asia's reporting, NEC has discontinued its effort to build its own physical, gate-based superconducting quantum computer — that's a chip running actual quantum circuits, not software that just mimics one. The program wound down at the close of the company's fiscal year, this past March, and Nikkei Asia, which first reported the story, says NEC's own reasoning was blunt: commercialization was judged too far off to justify the capital it would take to keep chasing it. This isn't some outsider giving up early. NEC pioneered this exact field — back in 1999, it demonstrated the world's first solid-state superconducting qubit, twenty-seven years before deciding to walk away from owning the hardware it helped invent. That's the uncomfortable part of this story: the company that proved the modality could work is the one now ending that hardware program. Think about what building this kind of machine actually costs you. You need dilution refrigerators running near absolute zero, a dedicated chip fabrication line, and years of error-correction research, all before a single dollar of revenue shows up. NEC's math said that runway was too long. So where's the money going instead? Quantum annealing — a narrower style of quantum hardware built specifically for optimization problems rather than general-purpose computing — plus quantum-inspired classical emulation, which is software that borrows quantum math to speed up ordinary computers without needing a quantum chip at all. That reallocation reshapes the map inside Japan. Fujitsu and RIKEN, Japan's national research institute, remain names to watch in superconducting quantum hardware. NEC's decision raises a question for the remaining developers: what would make that investment worthwhile? Meanwhile, American and Chinese competitors are still pouring capital into physical hardware at a pace NEC just decided it couldn't justify matching. Here's the caveat that matters most: NEC's language — that the payoff was 'too long' to chase — is the company's own framing, relayed through Nikkei's reporting, and it's the company's stated rationale, not an independently audited verdict on the whole industry. NEC also isn't exiting quantum research broadly; it's retreating from one specific, capital-intensive bet — owning the physical chip — while keeping annealing, quantum-inspired computing, and software services running. But here's the question I can't shake: if NEC's own return-on-investment math says hardware ownership doesn't pencil out yet, is that a signal other well-capitalized players should be sitting with? This is the same tension we've flagged before on this show — IBM needs quantum to be the answer to its innovation story, and whether the market's buying that long-term is still an open bet. NEC just publicly decided the runway was too long for its own balance sheet. That doesn't settle whether IBM or anyone else is wrong — it does mean the market should probably start asking every company still funding its own quantum chip line the exact return-on-investment question NEC just answered for itself. Our second lead today: Fujitsu's answer to the same problem — a computer that doesn't need to run quite as cold. Fujitsu, working with Delft University of Technology and QuTech, its quantum research institute, unveiled what they're calling the world's first working prototype of a diamond-spin quantum computer. There's something almost too on-the-nose about a company building a quantum computer out of diamond, but stick with me. Instead of superconducting circuits, this design uses tin-vacancy centers — a tiny defect in a diamond crystal's structure, involving a tin atom and vacancies in the diamond lattice, whose spin can encode quantum information — built into photonic circuits, meaning the qubits talk to each other using light instead of direct electrical wiring. Now, the number that actually matters: this prototype runs at minus 271.6 degrees Celsius. That's warmer than a typical superconducting quantum computer, which needs minus 273.13 degrees — just a hair above absolute zero. That gap sounds tiny on paper, but in cryogenics, warmer means cheaper, less finicky refrigeration, which is a real practical advantage if it holds up at scale. Fujitsu says the prototype was demonstrated running on its own Hybrid Quantum Computing Platform, and, according to the company, without requiring any additional specialist knowledge to operate. This builds on joint research Fujitsu started with TU Delft and QuTech back in 2020, and it slots into a roadmap Fujitsu has stated publicly: 250 logical qubits — that's error-corrected qubits, built by combining many raw physical qubits into one reliable enough to trust — by fiscal year 2030, and 1,000 logical qubits by fiscal 2035. The company says it plans a multi-module version of this diamond-spin machine by 2027, and is also exploring integrating the approach with superconducting hardware down the line. Underneath the headline, Fujitsu also detailed three specific engineering pieces behind this: a way to convert standard quantum circuit instructions into the light, microwave, and radio-frequency signals the diamond-spin qubits actually need; a bonding and thinning process that fuses diamond substrates onto silicon and alumina and shaves them down from hundreds of micrometers to just a few hundred nanometers thick; and a photonics fabrication technique that extracts single photons from these tin-vacancy centers during readout. That's real engineering, not just a naming exercise. Vivek Mahajan, Fujitsu's chief technology officer in charge of System Platform, framed the appeal this way: the diamond-spin approach, he says, 'not only offers exceptional scalability in its own right, but also has the potential to be integrated with superconducting quantum computers to further extend their capabilities.' Now, the more interesting quote, to me, comes from the research partner. Dr. Kees Eijkel, general director of QuTech, called this prototype 'a major milestone' in the collaboration — but he didn't stop there. He also said, and I'll quote him directly: 'Demonstrating the scalability expected of diamond spin quantum computing remains a long and challenging journey.' That's the head of the actual research institute behind this work telling you, on the record, that scaling this thing up is still unproven. So how do we weigh that? This is a company press release announcing a lab-stage prototype — it hasn't been independently benchmarked against trapped-ion or superconducting front-runners on the metrics that decide a race, like error rates per gate. Fujitsu's own timeline puts a multi-module prototype in 2027, with a 250-logical-qubit machine not due until fiscal 2030 — and roadmaps in this industry slip constantly, which isn't a knock on Fujitsu specifically, just the field's base rate. But the underlying idea is genuinely interesting: photonic connectivity, using light to link modules, is one of the more credible paths to solving the scaling problem that trips up nearly every other qubit type, because you're not limited by how many wires you can physically route into a fridge. For now, file this as promising, unproven, and worth revisiting when Fujitsu says the next version is due. So what do these two stories add up to, side by side? Two Japanese giants, one week apart, staring at the same long, capital-intensive road to a quantum computer that actually pays for itself — and making different choices about where to invest. NEC decided the payoff was too far away to keep funding hardware ownership. Fujitsu, in that same week, put fresh money into a warmer, optically-networked qubit design it believes can scale past the superconducting chips everyone else is racing to build. Time for the Hype Check: I'm calling this pairing a six. NEC's move is real and well-corroborated — a founding company walking away from its own invention is a genuine data point. Fujitsu's prototype is real hardware with real partners, but it's an unbenchmarked lab result, and QuTech's own director is the one calling the scaling path 'long and challenging' — the substance is there, the certainty about where it leads is not. If Fujitsu's multi-module diamond-spin prototype actually ships on schedule in 2027, that'll tell us whether this warmer, optically-linked approach is a real contender or just another credible architecture that never scales past the lab. Today's episode is a good one to send to the colleague who thinks every quantum headline is the same story — this week it clearly wasn't. Follow Quickly Quantum wherever you listen, and I'll catch you back here tomorrow. 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! I also host Space Stakes: the business of the new space race, every day. What actually flew, what the contract is really worth, and who has customers. Find it wherever you get your podcasts.