Back to News
BroadcastYouTube

qBitTensor Labs Live — October 1, 2026

October 1, 20261:10:22

Fresh off the Exploit Summit in Montreal, the team marks a shift from proving things out to scaling them: RSA-512 and RSA-530 both fall with RSA-560 up next, a white paper puts the cost of breaking RSA-512 at roughly $7 versus about $100 for the 2015 benchmark, Hardening Quantum Proof Milestone 3 is broken twice in under 12 hours by two different approaches, Enigma's first IP licensing deal is being explored, 90%+ of Open Quantum traffic runs through SN48 with the first revenue buybacks arriving in October, the team recaps IEEE Quantum Week, Quantum World Congress and the Clemson Quantathon, longtime community member Ryan Bequette (JB Rings) joins to introduce Carbon — the new surrogate-model physics subnet he won a BitStarter slot for on the Exploit main stage — and the team shares early takeaways on gamma.

Fresh off a busy run of events, ending with the Exploit Summit in Montreal (September 28–29), the October 1st broadcast covers Enigma first, then steps back to look at how SN48 and Open Quantum fit together, recaps three weeks on the road, and brings longtime community member Ryan Bequette, known to most as JB Rings, on to talk about his new subnet and his experience at Exploit. Bob set the tone with a slide he added at four in the morning: a lot of what the team talks about now has moved from proving things out and building them to scaling them.

Enigma: What Makes Something "State of the Art"?

  • Bob opened with the five-minute pitch he gave at Exploit. State of the art isn't the best that can be done. It's closer to a centralized staffing decision: a committee funds a problem up to a point and then stops, and wherever it stopped becomes the new state of the art. Talent is everywhere, in basements and apartments all over the world, even if the labs are concentrated in a few places. Enigma removes the staffing decision.
  • From Const's Exploit keynote: incentive computing is where you specify a valuable result, how to verify the result, and a reward for producing it, and you leave the methods open. Bob's take: you couldn't describe Enigma any better.
  • Prizes on Enigma grow until someone achieves the result and unlocks them. Winning results get published and feed into the next milestone, just past the new state of the art. There are no participation trophies. Big results earn big rewards.
  • At Exploit, plenty of people who don't take part in Enigma already knew it as "the one with the big prizes."
  • It all runs on Bittensor, prize pools included. Many subnets still hold prizes in owner-custodied wallets you have to trust. Enigma's are tracked on-chain in treasury wallets, and the site reads the chain to show where everything is. The team thanked the Church of Rao for building treasury wallets, and as far as they know Enigma may still be the only subnet using them.

What Makes a Good Challenge, and What's Being Added

  • The original criteria on the technical side: the problem is hard (each step further is much harder than the last), scalable (it can be shrunk to a size that's practical to verify), verifiable, and randomizable (each submission gets a fresh instance). On the market side: it's understandable enough to draw attention, it has solid partners (ideally ones who would license the resulting IP), and the solutions are valuable. Finding challenges that meet all of these is harder than most people expect.
  • Arbos's feedback on the subnet scored Enigma seven out of nine, and it would have been eight with AI in the subnet. With that and the themes at Exploit, three new priorities are on the list: AI challenges next, other Bittensor subnets as challenge partners (conversations started at Exploit), and gamma, which could come into play as the team works with subnet partners.

Breaking RSA: 512 and 530 Fall, 560 Is Next

  • RSA-512, the key size that protected real communications until about 1999, was gating for months and has now been solved. RSA-530 went next, faster than Bob expected. RSA-560 is the next target.
  • ShorShot: the aim is milestones that take real effort without being out of reach. Back at 460 and 480 bits, some people said the milestones were impossible and asked for more time and CPU. The community pushed through anyway. The rules stay the same: a four-hour window, 24 CPUs and an RTX 6000 Pro. The 530 win brought in new techniques that look like they came from outside the project.
  • "The record is 896 bits, so who cares about 560?" Enigma scales problems down on purpose. If a challenge were about cutting inference cost by 5%, you wouldn't make miners run millions of dollars of inference to prove it. You'd give them a smaller, representative problem. The 896-bit factorization used 2,048 GPUs for 10 calendar days, about 30 GPU-years. Enigma gives a single GPU for four hours, and 30 GPU-years is roughly 65,000 times longer than that. Bob wouldn't be surprised if techniques found on Enigma end up pushing the large-scale record forward.
  • RSA is a good challenge because the math is easy to generate and verify, and it gets a lot of attention. It was ideal for building the framework, and it brings people into Bittensor. RSA-1024 works as a canary in a coal mine. Nobody expects classical methods to break it, but if someone does claim the quarter million, it could be an early warning that quantum computers are getting close to 2048. Large standing prizes that go unclaimed for a long time are good for the subnet and for attention.

A White Paper on RSA-512

  • Omar compared Enigma's RSA-512 results with the 2015 benchmark paper. That paper broke RSA-512 in about four hours on AWS, using far more CPUs than Enigma allows, at a cost of about $75 (roughly $100 in today's dollars). By the team's calculation, breaking RSA-512 with the techniques found on Enigma now costs about $7.
  • The team thinks that's worth a paper and is writing one up. It will also project how far the methods could go with unlimited hardware and time, the kind of setup behind the record-breaking factorizations. The paper forecasts that result rather than running it.
  • Bob's offer on the fly: Argonne and Oak Ridge have both offered time on their supercomputers for quantum-adjacent work, as long as it's published and they're cited. If you're serious about scaling these approaches up, reach out. A Bittensor subnet breaking the largest RSA key ever factored would be good for the whole ecosystem.

Hardening Quantum Proof: Milestone 3 Falls Twice

  • Milestone 3 had stood for months. It fell while Bob was on the road. Because the new web-submission path was missing a gate, a second solution got in too, and it also passed. Two completely different approaches broke the milestone within a day of each other. Bob downloaded both to make sure one hadn't copied the other.
  • The full prize went to the first winner. The second solver, who the team knows to be the head of a quantum program at a large pharmaceutical company, was offered a choice: keep the source private for a future milestone, or open-source it for the $10,000 baseline prize. They chose to open-source it. They had been submitting through the web portal, which doesn't need a Bittensor wallet, and then said they would go register one. Bob called that a good sign of outsiders forming a positive view of the ecosystem.
  • RSA scores high on every criterion except value, since it isn't very licensable. Hardening Quantum Proof is the reverse. It's harder for outsiders to understand (it's about proving or disproving quantum advantage with peaked circuits), but the breakthroughs are novel and very licensable.
  • The first licensing deal is being explored. Full disclosure: it's with Quantum Rings, the team's company outside Bittensor. A few of the techniques in the submitted solutions could improve Quantum Rings' core technology. Rather than use the open-source license, Quantum Rings would reimplement the approach and, to be safe, license it. They're testing now to see whether it gives the performance boost they want. If it does, that's Enigma's first small licensing deal.

The Road to Revenue

  • Omar walked through the arc. The early milestones were meant to be achievable, to get miners used to the system. Difficulty went up from there, some prizes went weeks without being won, and the growing pools drew outside attention. Web submissions are an early sign of people from outside Bittensor taking part.
  • Where things are now: approaching small IP sales, with milestones hard and relevant enough that people care about the results. Next comes the hope that other companies take notice. The end goal is large, repeatable IP sales for results that companies urgently need, which is when the full revenue cycle kicks in.
  • Why so many small steps (480 → 500 → 512)? Each one builds the ladder miners need to climb. A day-one RSA-512 challenge probably still wouldn't have been solved. At least one of the Hardening Quantum Proof Milestone 3 winners built on work from the previous two milestones. That's the flywheel turning.

What's Next for Enigma Challenges

  • The team has a backlog of future challenges that meet the technical criteria, plus a list of interested parties: universities, national labs, defense contractors and other Bittensor subnets. The work now is matching the two, ideally launching challenges directed by a partner so the results are more likely to be licensed.
  • The team could move faster without partners, but partners bring both reach and licensing potential. A defense contractor's first question was what it would end up costing them. The answer: licensing terms can be agreed in advance, and they don't pay unless the milestone is passed.

SN48 + Open Quantum: The Open Infrastructure Layer for Quantum

  • Chutes is one of the key backends behind OpenRouter, and that relationship was both the template and the aspiration for Open Quantum and SN48. OpenRouter was acquired for $7.5 billion.
  • Every technology stack ends up with a neutral infrastructure layer: Linux for operating systems, Docker and Kubernetes for servers, Terraform for cloud, OpenRouter for AI models. Those companies exit for billions. The team believes it's building quantum's version.
  • How it works: developers connect from any framework to the Open Quantum cloud, which routes to leading quantum computers through SN48, or directly for customers who don't want to touch crypto. Generative AI can run the quantum executions for you. Along the way, the platform builds a large, rich dataset of inputs, outputs and execution data.
  • Free credits and discounted compute run only on the public tier, which is decentralized and goes through SN48. Users who pay full price get centralized, private execution with no decentralized component. Some customers need that.
  • Why subsidize? AI is rewriting every part of the software industry except quantum. The quantum market is growing from about $3.5 billion to $20 billion over the next few years, and AI still struggles with it: a recent study found it is less than 40% accurate at mapping real business problems into quantum algorithms. Subsidized compute drives adoption and builds the dataset.
  • Traction: jobs, total users, active users and revenue are all up and to the right. More than 90% of traffic runs through SN48, from nearly every country with IT infrastructure (a few can't be served). 23% of September's new users came from referrals.
  • Funding the business: an Exploit panel on funding businesses from subnet emissions reached a clear conclusion. Emissions can fund the development of a technology, but pulling out enough to pay staff and run a business will crash the subnet. Quantum Rings, the traditional business, is opening its next funding round. Meanwhile, the first revenue buybacks come to SN48 in October, funded by Open Quantum e-commerce buying time on SN48. A couple of months have already accumulated, and the books take time to settle because the businesses are intertwined.

On the Road: IEEE Quantum Week, Quantum World Congress, Clemson

  • IEEE Quantum Week (Toronto): Quantum Rings exhibited and talked Open Quantum with users, students, academics and industry. Plenty of recent sign-ups trace back to the conference, many drawn by the heavily subsidized public tier powered entirely by Bittensor. BlueQubit won an award for its peaked-circuits work. Hayk told many who came to his booth about the Enigma collaboration and the money behind it. Other booths had "partner of the IBM ecosystem" plaques, and he wanted an Enigma one. More is in the works with BlueQubit, to be revealed soon. There was also a lot of interest in classical-quantum hybrid workloads such as variational algorithms, which the team will look at for Enigma challenges or SN48.
  • Quantum World Congress (University of Maryland): a more DC-centric crowd. Bob's observation: it's getting harder to tell hype from reality, as publicly traded quantum companies face growing pressure to claim advantage and faster roadmaps, putting the marketing narrative ahead of the real one.
  • South Carolina Quantathon (Clemson University): a 24-hour challenge. Students got quantum circuits with their runtimes on the Quantum Rings simulator and had to predict runtime without running the circuits. It follows the MIT hackathon and the Colorado School of Mines capstone. The goal is upfront billing for simulators on Open Quantum, instead of charging at the end of runs that can take minutes to days. In the mentors' room (a Clemson PhD student, someone from Oak Ridge, people from two universities, and an Amazon engineer), a casual mention of Enigma turned into a walkthrough on the big screen. One mentor spent the next 24 hours pulling more people in to hear about it. Hundreds of thousands of dollars in prizes, and someone winning $30,000 for quantum work, gets attention.

Exploit Summit & Carbon, with Ryan Bequette

  • Exploit's big announcement for this community: a new subnet, Carbon, from longtime supporter Ryan Bequette (JB Rings). Carbon focuses on surrogate models, a class of models that could provide huge benefits in physics simulation. The aim is to make them valuable to real engineering teams for their simulation work, bringing outside teams to Bittensor and creating value. It's at carbonphysics.ai, with the same handle on X. There's no slot yet. One should be registered in the next week or so.
  • Ryan presented on the Exploit main stage and won the contest for a BitStarter-registered subnet. In his words, it was "way out of my comfort zone," but he'd have regretted not trying.
  • The event itself: Ryan walked into quiz night on the first evening and was surprised to see Bob already wheeling and dealing. Everyone you know from the ecosystem is there, and Ryan recommends going to anyone who wants to get more involved. Bob's take: people in person are often nothing like their online presence, in both directions.
  • Const's keynote felt like a well-produced NVIDIA keynote. Ryan said it started slowly, then got to the exciting part: incentivizing subnets to bring outside capital and attention into the ecosystem, and subnets building on each other.
  • The emission cut came up constantly. The subnet was one of eight that had emission cut about four weeks earlier, and it was the first thing most people asked Bob about. He didn't have an inside story. He's grateful the ecosystem exists, and he accepts that in an ecosystem moving fast, things get broken and fixed again. Things are running smoothly now.
  • Ryan's other big theme: people are ready for real value on the network and tired of open-ended experimenting. He pointed to talks on this, including one by Max from Score.

Gamma: Early Takeaways

  • How the team understands it so far: gamma is an on-chain mechanism for subnets to transact with each other. Swap your alpha into gamma, which buys and burns the target subnet's alpha, and spend it there on storage, inference, IP development and so on. Where the other idea Const raised was about bringing dollars into TAO, gamma is about keeping TAO in TAO.
  • Ryan: it lets subnets tune owner versus miner emission and pay for things they need, like Carbon's costly simulation data, while being transparent with their community about it.
  • Bob's three positives: (1) it's on-chain, so unlike "we sold a million dollars of alpha for investments," spending through gamma can be verified; (2) it keeps value in TAO rather than leaving to pay for things; (3) it makes value creation measurable. A buy-and-burn alone doesn't prove revenue, but spending through gamma into a subnet does, which could let Bittensor show the outside world how much value it creates.
  • The one downside so far: once something goes to gamma, 100% of it is burned. One subnet reported $15 million in revenue from selling miner services and only $2 million in buybacks. Gamma would likely prevent that, but businesses should still have some way to declare a profit margin.
  • The details of the implementation will matter, but announcing it ahead of time lets subnets line up their strategies with the ecosystem's.

Community & Q&A

  • What's the bigger picture with Breaking RSA? At first it was about proving out the technology with a challenge that's easy to understand. From here, Bob says honestly, it's mainly a way to get eyes on the subnet and the ecosystem. A licensing deal is possible but not expected, though ShorShot notes a surprise breakthrough is always the hope. Bob joked that a cease-and-desist from the DOD would be the best thing that could happen to the subnet.
  • The ~50% alpha price drop: emission was cut about four weeks ago with the root key, alongside other good subnets, apparently because of a view that some subnets' designs gave them an accidental edge. The team thinks it was a rash move that hurt the ecosystem more than it hurt them. They've made small changes in response, and emission is back on. The big down candle came from one holder who said he'd waited long enough, then liquidated, and others piled on. Ryan: that's how things cascade in dTAO. You can't give investment advice, but Bob noted the price sits just below the settlement price on liquidation, because as an older subnet all the TAO from past burns remains in the reserve pool.
  • Miner sentiment has never been higher: new miners keep arriving excited, milestones are falling almost weekly, and with IP deals and a white paper underway, Omar expects the flywheel to turn the other side around too. ShorShot: some participants spend months on challenges as a hobby, and the platform can run and validate essentially any code in one clean, efficient system.

Community Sentiment

  • RSA-1024: plenty of excitement about the huge prize, and some pushback that it's "just marketing hype." The team agrees it is marketing, and thinks it could still draw outside talent. Nobody is breaking it in four hours on an RTX 6000, "otherwise we're all in a lot of trouble."
  • The Department of Energy put out a challenge paying per milestone toward building a useful quantum computer. The team pointed out on X that the model looks familiar, which confirms the vision and the market.
  • Marketing: a common complaint is that the vision needs more explaining. The team largely agrees and is working on how to pitch it better to new investors while giving existing ones fresh content.
  • The decentralized research lab narrative got a lot of positive reaction, and the team expects to lean into it more.
  • Bob's wrap-up: there's more red than blue in sentiment than usual, and the team is including both. Filtering out the RSA-1024 debate, which should settle down, the message is that people want more marketing and like the new narrative, and those two go together. Licensing the Hardening Quantum Proof IP, even to the team's own company, and burning the revenue would be a good first proof of the model, "even though it's awfully incestuous." He thanked everyone who stayed through the emission freeze, and to holders who left: no hard feelings, and he's sorry the situation couldn't be resolved faster.