4FIRE @netcreat

Our Executive Briefing Center is open! It was great to host academic, government, enterprise, and additional partner tours at our headquarters in College Park, Maryland, this week, including delegations from Finland, Japan, Korea, the UK and across the U.S. Here’s a sneak peek at what visitors are seeing. Thank you again @FIU, SDT Inc., and everyone else who joined us this week. Thanks for sharing the excitement, tour guides: Rick Muller, Matthew Keesan, Andrew Piper, and Hiram Gonzalez. See you soon! #IonQ #IonQSuperion #QuantumIsNow #QWC2026 #QuantumWorldCongress #IonQEBC

keyboard_arrow_left Previous keyboard_arrow_right Next

I jumped into a great discussion with @genejchan , @Truthinthetape , and @sakacc about FTQC yesterday. By coincidence, I came across a report co-authored by Qolab’s Dr. John Martinis, published just two days before QWC, and it contains some points that may very well put an end to much of what we were discussing yesterday — so I thought I’d share it here. I’ve also summarized the report on my blog and pulled together some related context and references around it in both ENG and KOR. So feel free to read whichever version suits your taste — the original report, my summary, or both. 🙂 Are 30 Logical Qubits Really 30 Logical Qubits? [ENG] netcreat.substack.com/p/are-30-logic… [KOR] blog.naver.com/4-fire/2244230… quantum.microsoft.com/en-us/insights… Sharing my comment about FTQC "FTQC is a complete system that only works when many technologies are balanced and integrated together. Claiming overall progress based only on the number of logical qubits achieved is not very different from completing an engine and saying that the entire car is finished."

Chad Sakac @sakacc

2 days ago

@netcreat @ATURULLAO4 @genejchan @Truthinthetape Thanks and yup, that’s a good catch on the Genesis Q RfA. wish the bar for the more near term ops count was higher (10^5 is still not enough), but it’s good that they pushing us all in this direction. For those curious, this is here (look at pg 8) science.osti.gov/-/media/grants…

1 0 22 2K 1

@JimHODL 🙏👊

If I’m invited again, I’ll definitely make sure to find you and say hello next time! 😊 And speaking of coincidences, I had already written about this in my blog post, but when I copied and pasted it, my finger somehow slipped and left out the next-generation 10^7 hard operations part. 🫰 Maybe my finger knew you already had the URL and page number at your fingertips and figured there was no need for me to include it! 🤣

I believe the DOE’s newly announced Quantum Competition will ultimately help settle much of this debate. Among its requirements, the number of logical qubits is relatively modest. What matters far more is the introduction of a metric that is difficult to argue with: hard operations. When Apple introduces a new iPhone, countless mockups and display models may appear before the actual product reaches consumers. They may look impressive, but ultimately, it is the experience of using the device that determines its value. Quantum computing is approaching a similar moment. Consider the claim that 3 logical qubits can be created from just 8 physical qubits. And 10 times of this block makes 30 LQ. For the sake of argument, let us accept that this is possible. If those 8 physical qubits were virtually error-free—perhaps with error rates below 0.00000001%—then perhaps such an approach could work. But what are the actual error rates of that company’s physical qubits? When 8 physical qubits are repeatedly performing quantum operations, the probability of multiple errors occurring within the same error-correction cycle cannot simply be ignored. A code distance of 3 means that only one arbitrary physical error can be guaranteed to be corrected: (3 − 1) / 2 = 1. Two or more errors can exceed that correction capability. Now consider the 3 quantum error-correcting codes presented in IonQ’s Walking Cat demonstration. Their code distances are nine or ten—greater than the entire physical-qubit count in the example above. Does anyone seriously believe that IonQ is using 70 or 102 physical qubits per code block, along with sophisticated techniques such as beam-search hard decoding, soft decoding, magic-state factories, and cat-state factories, simply because it has no better alternative? Is it because its physical error rates are too high? Quite the opposite. The objective is to achieve a higher level of logical-qubit reliability, even with exceptionally low physical error rates. Creating a logical qubit is one thing. Creating a logical qubit that can reliably perform a substantial number of meaningful operations is quite another. And that is precisely why the DOE’s requirements are so important. ~100 logical qubits ~100,000 hard operations Scientifically relevant workflows ~1,000 logical qubits ~10 million hard operations Production-level scientific applications In the end, the market will not be decided by who can claim the largest number of logical qubits using the fewest physical qubits. It will be decided by what those logical qubits can actually accomplish. Customers will ultimately judge the value of these systems through real-world experience, while rigorous, independently validated testing under programs such as DARPA and the DOE’s Quantum Competition will provide a more objective basis for comparison. The era of impressive demonstrations and headline numbers is gradually giving way to the era of measurable performance. And I believe that moment is now approaching. Quantum is Now? Nope. Quantum has already begun. Btw, I like your presentation, Chad. Your energy is contagious, and you have an incredible ability to keep the audience fully engaged. I was really disappointed that I didn’t get a chance to say hello to you in person at Investor Day. Much appreciated for your works and updates.

@HannaSuds Thanks Hanna, hope you’re well.

$IONQ ’s Bigger Picture [ENG]netcreat.substack.com/p/ionqs-bigger…

4FIRE @netcreat

2 days ago

$IONQ 가 양자컴퓨터 제조사를 넘어 양자산업의 공급자가 되려 한다면? m.blog.naver.com/4-fire/2244223… 이번 QWC 슬라이드를 처음 봤을 때는 IonQ가 그동안 인수한 기업들을 하나의 기술 스택으로 정리한 자료라고 생각했다. 그런데 Hybrid Classical Integration 계층에 AMD·Arm·Broadcom이라는 세 기업이

4 6 58 9K 6

$IONQ 소프트 디코딩 논문 완전 해설: 580배 오류 억제부터 논리 큐비트 수명 13.66초까지 youtu.be/AQEJZLjvzgk?si…

$IONQ 가 양자컴퓨터 제조사를 넘어 양자산업의 공급자가 되려 한다면? m.blog.naver.com/4-fire/2244223… 이번 QWC 슬라이드를 처음 봤을 때는 IonQ가 그동안 인수한 기업들을 하나의 기술 스택으로 정리한 자료라고 생각했다. 그런데 Hybrid Classical Integration 계층에 AMD·Arm·Broadcom이라는 세 기업이 등장하고, 그 주변에 Nexus Photonics와 Vector Atomic이 배치된 것을 보면서 다른 가능성이 떠올랐다. 이미 NVIDIA는 NVQLink와 CUDA-Q를 통해 자사의 기존 컴퓨팅 자원과 다양한 양자컴퓨터를 연결하는 생태계를 구축하고 있다. 그렇다면 IonQ는 NVIDIA와의 협력을 유지하면서도, 다른 기존 컴퓨팅 기업들이 양자산업에 참여할 수 있도록 관련 기술을 공급하는 또 하나의 사업 모델을 준비하고 있는 것은 아닐까? AMD의 FPGA가 다른 양자컴퓨터 제조사의 제어 시스템과 연결되고, Arm 기반 시스템이나 Broadcom의 네트워킹 기술이 다양한 양자 장비와 결합하는 과정에서 필요한 광학 기술을 Nexus가 공급할 수도 있다. 그리고 이러한 부품의 제조와 고객별 맞춤 설계를 SkyWater가 담당한다면, IonQ는 자사의 QPU를 판매하지 않는 영역에서도 양자산업에 참여할 수 있다. 이는 IonQ가 이미 공식적으로 발표한 Merchant Supplier 전략과도 방향성 면에서 연결된다. NVIDIA가 NVQLink를 통해 NVIDIA와 다양한 양자컴퓨터를 연결하는 생태계를 구축하고 있다면, 어쩌면 IonQ는 Merchant Supplier로서 AMD·Arm·Broadcom 등 기존 컴퓨팅 산업의 주요 기업들과 협력하며 또 다른 형태의 Pick & Shovel 기업으로 발전하려는 것일 수 있다. 이 경우 IonQ의 사업은 자체 양자컴퓨터의 성능과 판매량에만 의존하는 것이 아니라, 다른 양자컴퓨터 제조사의 기술 발전과 기존 반도체 기업들의 양자산업 진입 과정에서도 새로운 기회를 찾을 수 있게 된다. 그리고 IonQ가 이러한 전략을 실제로 구상하고 있다면, Nexus Photonics는 단순히 양자컴퓨터 내부의 광학 시스템을 소형화하기 위해 인수한 기업 이상의 의미를 가질 수 있다. 기존 컴퓨팅 산업과 양자산업에 모두 적용할 수 있는 광학 기술을 확보하고, 이를 SkyWater의 제조 역량과 결합해 외부 고객에게 공급하기 위한 중요한 기술적 기반이 될 수 있기 때문이다. 어쩌면 Nexus는 IonQ의 양자컴퓨터를 더 잘 만들기 위해 인수한 기업인 동시에, IonQ가 직접 만들지 않는 다른 양자컴퓨터에서도 매출을 창출하기 위해 확보한 기업이었을지 모른다. 물론 이것은 어디까지나 필자의 가설이다. QWC 슬라이드의 기업 배치가 이러한 사업 계획을 의미한다는 공식 설명은 아직 확인되지 않았으며, AMD·Arm·Broadcom과 Nexus가 공동으로 범용 양자 인터커넥트를 개발하고 있다는 증거도 없다. 현재 확인되는 사실은 Nexus가 양자 및 기존 컴퓨팅 산업에 활용할 수 있는 광학 기술을 보유하고 있고, IonQ가 SkyWater를 통해 외부 양자기술 기업에 반도체 제조 및 집적 기술을 공급하는 사업을 추진하고 있다는 것이다. 앞으로 관련 기업들의 공동개발 발표와 실제 제품 출시를 지켜보면서 이번 가설이 현실과 얼마나 가까워지는지 확인해 볼 생각이다. 그때가 되면 QWC에서 공개된 이 슬라이드 한 장을 다시 꺼내 볼 이유가 생길 것이다. 자세한 내용은 해당블로그에서. 영어버전은 곧 올라갈겁니다.

Dan Klamm @DanKlamm

3 days ago

“We want to work with everyone who is here.” - @NiccoloDeMasi emphasizing @IonQ_Inc’s role building a trusted quantum ecosystem for all. #QuantumWorldCongress

2 9 67 6K 2

“Countries want to grow ecosystems with @IonQ_Inc. So do states. So do universities.” @NiccoloDeMasi #QuantumWorldCongress

“We want to work with everyone who is here.” - @NiccoloDeMasi emphasizing @IonQ_Inc’s role building a trusted quantum ecosystem for all. #QuantumWorldCongress

@DesFrontierTech you say nothing. eh? 😀

It’s official: @IridiumComm shareholders have voted and approved the acquisition by Rocket Lab ✅ By joining forces, we’ll be creating a vertically-integrated space powerhouse that designs, builds, launches, and operates its own constellations to bring critical communications capability to millions of users worldwide. Full details and next steps: bit.ly/3T53qq1

Quantum value will compound where capital and talent compound. @NiccoloDeMasi @IonQ_Inc #QuantumWorldCongress

FIU chooses IonQ’s Superion 256. I explore its growing customer base and 2027 revenue scenarios for 10, 20, or 30 systems—not company guidance. youtu.be/DgyQwHOzzGo?si… cpachoi.wixsite.com/ryanlab/post/f… Turn on English caption/translation. $IONQ #QuantumComputing

$IONQ $QNT System Deployments for those keeping count: IonQ: 1. NVIDIA Accelerated Quantum Research Center (Superion) 2. University of Cambridge (Superion) 3. Florida International University (Superion) 4. QuantumBasel (Superion & Tempo) 5. Horizon Quantum (Superion) 6. KISTI (Tempo) 7. EPB Quantum Center (Forte) 8. UChicago (System not Identified) 9. Air Force Research Lab (System not Identified) 10. UK NQCC (Quartet) 11. Cyberagentur (MinION) 12. SDT Inc. (Superion) Quantinuum: 1. Riken (H1 & H2) 2. Singapore NQO (Helios) 3. Oracle (Helios)

$IONQ Sells Superion 256 to FIU — A Bigger Picture Is Emerging blog.naver.com/4-fire/2244220… IonQ has signed an agreement to sell its Superion 256 quantum computer to Florida International University (FIU), with on-site installation expected by late 2027. FIU aims to establish a regional hub for quantum research, education, and industry collaboration. Interestingly, FIU is also a co-owner of Florida LambdaRail, which partnered with IonQ to develop a statewide quantum-secure network using QKD. m.blog.naver.com/4-fire/2242672… This highlights a potential path for IonQ to expand customer relationships from quantum security into quantum computing, networking, and quantum memory. Meanwhile, IonQ's integration of Oxford Ionics' technology with SkyWater's semiconductor manufacturing capabilities is laying the foundation for scalable production beyond laboratory-built systems. Its roadmap from Superion 256 to Superion 10K could give customers greater confidence in the platform's long-term viability, reducing uncertainty around early investments. Recent announcements involving SDT, NVIDIA, and FIU suggest growing interest in deploying quantum hardware directly at customer facilities rather than relying solely on cloud access. The bigger picture: IonQ is building an ecosystem where quantum security, computing, and memory can reinforce one another, while its semiconductor-based manufacturing strategy and scalable product roadmap give customers greater confidence to invest in quantum infrastructure today rather than wait for the technology of tomorrow. Good work, Team @IonQ_Inc

$IONQ Hybrid soft decoding. m.blog.naver.com/4-fire/2244220… IonQ’s new soft decoding approach improves quantum error correction without adding physical qubits or changing the underlying code. By using BP convergence information from its existing Beam Search Decoder, IonQ selectively accepts reliable results, rejects unreliable ones, and performs a second decoding pass only when necessary. The results: 580× lower logical error rate in Q54 memory simulations, rejecting just 0.1% of results. 210× lower logical measurement error rate with only a 0.5 percentage-point increase in EDM restart probability. 13.66-second logical memory lifetime from reanalyzing experimental data from IonQ’s earlier breakeven study. The first two results are simulations; the third uses existing experimental data. The key takeaway: IonQ can improve logical reliability by making smarter use of its existing decoder, potentially benefiting logical operations and magic-state factories without requiring additional physical qubits.

@HannaSuds , I think you’ve made a very important point, and I’d like to add another perspective that may help explain NVIDIA’s broader strategy. Jensen Huang is an exceptionally business-savvy leader. Throughout NVIDIA’s history, he has demonstrated a remarkable ability to recognize emerging markets and position NVIDIA at the center of them. I believe quantum computing is another example of that strategy. When Jensen made his controversial comments in January 2025, NVIDIA had not yet introduced NVQLink. My interpretation is that he wanted to slow down the quantum narrative while NVIDIA was still preparing its own position in the emerging ecosystem. I don't think he wanted to kill the quantum computing industry. He may simply have wanted to buy NVIDIA some time. Two months later, at GTC 2025, he brought together leaders from across the quantum industry, helping to shift the conversation toward a future in which NVIDIA would play a central role. Then, in October 2025, NVIDIA introduced NVQLink. By GTC 2026, the focus had moved from bringing the industry together to deploying the infrastructure that would connect quantum processors with NVIDIA's accelerated computing platforms. That transition is significant. NVIDIA doesn't need to manufacture every quantum processor or capture the entire quantum computing market. Its strength lies in the picks-and-shovels strategy: providing the essential infrastructure that allows NVIDIA to benefit regardless of which quantum hardware architecture ultimately succeeds. And I believe Jensen has preserved that position remarkably well in quantum computing. Consider IonQ's recent CPU-based real-time quantum error decoding research. IonQ demonstrated that a single conventional CPU could handle the decoding workload of a simulated large-scale fault-tolerant quantum computer with remarkably little execution delay under its assumed operating conditions. This is a significant achievement for IonQ because it suggests that expensive GPU infrastructure may not always be necessary for quantum error correction. But I don't believe this threatens NVIDIA's broader quantum strategy. IonQ's own research shows that decoding delays increase as physical error rates rise. Other quantum architectures may face very different decoding requirements, potentially making GPU acceleration far more valuable. More importantly, quantum error correction is only one part of the quantum-classical computing ecosystem. NVIDIA GPUs can also play essential roles in large-scale quantum simulation, AI model training, hybrid quantum-classical algorithms, and AI-assisted quantum hardware control and calibration. Even if IonQ can perform its error decoding efficiently on a CPU, NVIDIA still has numerous opportunities to provide the computational infrastructure surrounding IonQ's quantum processors. In other words, IonQ's success does not necessarily come at NVIDIA's expense. This brings me to the latest announcement. NVIDIA has already collaborated with companies such as Quantinuum and QuEra. It could have continued building its research platform around those existing relationships. Instead, it has chosen IonQ's Superion 256 as the first quantum processor to be installed at its Accelerated Quantum Research Center. My interpretation is that NVIDIA wants to work with partners capable of delivering meaningful results as its hybrid computing infrastructure moves from research toward practical deployment. And there is another important distinction. IonQ is not simply being invited into a dedicated quantum computing facility. Its Superion 256 is scheduled to operate alongside NVIDIA's GB200 NVL72 infrastructure through NVQLink and CUDA-Q. This is about bringing quantum computing directly into the existing classical supercomputing environment rather than building an entirely separate computing ecosystem. That, to me, is what makes this announcement particularly significant. I also see IonQ's collaboration with EPB in Chattanooga and its joint research with Oak Ridge National Laboratory and NVIDIA as potentially complementary pieces of a much broader strategy. EPB provides an environment for deploying and testing quantum computing and networking technologies in real-world settings. ORNL brings national-laboratory expertise in high-performance computing and scientific applications. And researchers such as NVIDIA's Yuri Alexeev, who spent 13 years at Argonne National Laboratory, bring valuable experience connecting quantum algorithms with large-scale classical computing infrastructure. His participation in the recent IonQ–ORNL–NVIDIA research is particularly interesting in this context. Together, these developments suggest a possible path from quantum algorithm research and hybrid software development to real-world deployment and, eventually, integration into large-scale AI and HPC infrastructure. That is how I interpret NVIDIA's strategy. Jensen does not necessarily need to know which quantum hardware company will ultimately dominate the market. He needs to make sure NVIDIA remains indispensable to the infrastructure that connects them all. And by bringing IonQ's Superion 256 into NVIDIA's own accelerated computing environment, he is taking another concrete step toward that goal. Hanna, I hope you don't take this as a counterargument. I was actually trying to build on your point, but I got a little carried away and ended up repeating myself. 😂 At the end of the day, I think we're saying the same thing: NVIDIA needs quantum computing, and quantum computing needs NVIDIA. And right now, $IONQ and $NVDA have a lot to gain from each other. 🤝 netcreat.substack.com/p/ionq-nvidia-…

Visiting @IonQ_Inc’s new Executive Briefing Center for the first time! Love this space.