Infleqtion Hits 30 Logical Qubits on Path to 1,000 by 2030

What happened

In September 2025, neutral-atom quantum company Infleqtion announced a redesigned architecture for its Sqale quantum computer and moved its logical-qubit roadmap forward. The updated plan promised a full-stack, fault-tolerant system with more than 1,000 logical qubits by 2030.216 The company said the new design already supported 12 logical qubits with error detection and loss correction. Its earlier roadmap had targeted 10 logical qubits with error correction in 2026, so the company described itself as ahead of schedule and set a new 2026 goal of 30.213

One year later, Infleqtion reported reaching that goal. At Quantum World Congress 2026, it said it had entangled 30 logical qubits using 80 physical qubits on Sqale. It called this a first for a commercial neutral-atom system and said it remains on track for 100 logical qubits in 2028 and 1,000 in 2030.15 Taken together, the two announcements show a company that has so far hit its published numbers on time. A closer look also shows how much depends on what counts as a "logical qubit," and that question is where the most important debate sits.

The architecture behind the roadmap

The 2025 redesign combined two capabilities that neutral-atom systems usually use separately. One is the ability to aim gates at individual atoms where they sit. The other is the ability to rearrange the atom array in the middle of a computation.216 Moving atoms gives a machine all-to-all connectivity, but each move adds runtime, lets errors build up, and risks losing atoms. Infleqtion's argument is that doing in-place entanglement alongside motion keeps the connectivity while cutting those costs.213 Quantum Computing Report said the design was first shown on a Sqale processor with a 114-atom array. The same demonstration included a constant-depth CNOT ladder and preparation of a [[16, 4, 4]] many-hypercube error-correcting code.13

In the 2026 experiment, the 80 atoms were split into 10 blocks of eight. Each block was prepared using a distance-3 [[8,3,3]] code and then run in a distance-2 [[8,3,2]] code. That gives a ratio of eight physical qubits for every three logical qubits.175 The circuit used 136 physical two-qubit gates, about 1,000 physical operations in total, and included four non-Clifford logical CCZ gates. Gates of that kind are required for a universal quantum computer.517 Infleqtion said its output signal was about 1,000 times stronger than the noise background.3

Two other details matter. First, the company said an AI model it calls GPT 5.6 Sol found a new logical "double-CZ" operation that needs four physical two-qubit gates where eight were needed before.58 Second, loss-correction software in its Superstaq stack rebuilt measurement results lost when atoms dropped out. That post-processing step roughly quadrupled the share of usable runs.5 Both are examples of software reducing the hardware burden, which is the hardware-software co-design pitch Infleqtion has been making for years.6

The Shor's algorithm claim and post-quantum cryptography

The most widely repeated part of the 2025 announcement was not the architecture. It was the company's claim to be the first to run Shor's factoring algorithm on logical qubits.216 Infleqtion ran a pre-compiled, simplified version of the algorithm to factor the number 15. It reported measurable fidelity gains over running the same task on physical qubits alone, under real-world noise.1116 CTO Pranav Gokhale presented the result as a reason to speed up the move to quantum-safe encryption, while acknowledging that the number factored was small.12

The result should be read for what it is. Factoring 15 with a pre-compiled circuit is a long way from threatening RSA-2048. Infleqtion's own release says a 2048-bit number would take classical computers an astronomically long time to factor.11 What the experiment does show is that error-encoded qubits beat unencoded ones on a task with cryptographic relevance. That is a real milestone, but it measures quality, not scale.

The more useful link to cryptography comes from scale. PostQuantum.com points out that Infleqtion's 2030 target of 1,000 logical qubits is close to the roughly 1,400 logical qubits in Craig Gidney's 2025 estimate for breaking RSA-2048. The site adds that the comparison only means something once Infleqtion publishes a logical error rate.17 Raw counts do not break encryption. Long, deep, error-corrected computations do. Still, a public company putting a date near that threshold supports the case that migrating to post-quantum cryptography is not a problem for some distant decade.

Detection versus correction: where the coverage splits

Most coverage of the 30-qubit result follows Infleqtion's framing. Several outlets repeated the "first neutral-atom company" claim, the 1,000-fold signal, and the AI-discovered gate with little critical comment.789 Interesting Engineering's headline went further and called the system an "error-corrected architecture."9

PostQuantum.com takes a more skeptical view and offers the most important correction to that framing. It notes that the 30 logical qubits rely on error detection. In these experiments, the [[8,3,2]] code can detect any single fault but corrects only an atom loss whose location is already known. The site stresses that Infleqtion's original 2026 target specified error correction.17 It also follows the roadmap's history. In February 2024, Infleqtion described a fully error-corrected 100-logical-qubit machine able to run circuits more than a million layers deep. A March 2026 SEC filing paired 100 logical qubits in 2028 with one million sequential logical operations. The September 2026 announcement repeated the 100-qubit target without either condition.17 The same analysis points to a proposal from Microsoft and Qolab for a stricter definition of a "scalable logical qubit," which requires real-time decoding and a universal set of fault-tolerant gates. It also notes that the U.S. Department of Energy says it is not interested in proposals for machines that can only detect or mitigate errors.17

The term "QuOp" is also used loosely. The company's 1 KiloQuOp refers to about 1,000 physical operations, while its stated next goal, the "MegaQuOp," refers to roughly a million reliable logical operations.17 These are different quantities, and using the same label for both makes the gap between them look smaller than it is.

The fair reading is that Infleqtion has delivered on the count it promised and on the date it promised. That record holds: two logical qubits in 2024, 12 in 2025, and 30 in 2026.17 The quality condition attached to those numbers, however, has been relaxed along the way. Both of these things are true, and investors and policymakers should give more weight to the second.

Why it matters for the quantum industry

Infleqtion now operates as a public company. HPCwire reported in September 2025 that it planned to merge with the special purpose acquisition company Churchill Capital Corp X at a pre-money valuation of about $1.8 billion, with expected gross proceeds of more than $540 million.16 Its shares began trading on the New York Stock Exchange in February 2026.17 Roadmap milestones are now material disclosures. The 30-qubit result was filed in an 8-K alongside FY26 revenue guidance of about $45 million. Quantum Zeitgeist reported record second-quarter 2026 revenue of $13.5 million.7

The company also says it has paying users. Three customers are running logical-qubit circuits on Sqale, among them the Wellcome Leap Quantum for Bio program. In that program, a quantum neural network is trained on GPUs and run on the quantum processor for biomarker discovery.15 HostingJournalist reported that Eaton uses Sqale through a private-cloud arrangement. It also reported that Infleqtion is under contract to install a system at the Illinois Quantum & Microelectronics Park in 2027, designed to expand beyond 50 logical qubits.10 Integration with NVIDIA's CUDA-Q places the quantum processor as an accelerator alongside classical GPUs, not as a standalone replacement.10

For the wider field, the result strengthens the case for neutral atoms. Reconfigurable arrays and low encoding overhead are emerging as real advantages over modalities that need many more physical qubits for each logical one. HostingJournalist also noted that Infleqtion's qLDPC work with NVIDIA used about six physical data qubits per logical qubit.10 The company appears to be investing in error correction specifically: it signed a memorandum of understanding with Riverlane in late September 2026 to advance error correction and fault-tolerant computing in the UK.

The road from 30 to 1,000

HostingJournalist summed up the situation well: 30 logical qubits is still 30. The 2028 and 2030 numbers are vendor roadmaps, not commitments, and the work between here and there spans error correction, control electronics, optics, and classical orchestration.10 On this analysis, the deciding factor will not be the next logical-qubit count. It will be whether Infleqtion publishes logical error rates and demonstrates real-time error correction at depth. Without those, a 1,000-logical-qubit machine would be a big number with uncertain capability. With them, the 2030 target becomes a credible date for cryptographically relevant quantum computing, and organizations that have not started moving to post-quantum encryption should start now.