Intelligence is foundation
Podcast Subscribe
Quantum Computing Tuesday, 17 February 2026

Iceberg Quantum Claims 10x Reduction in Qubits for Breaking RSA

Share: LinkedIn
Iceberg Quantum Claims 10x Reduction in Qubits for Breaking RSA

Iceberg Quantum published a paper this week claiming their fault-tolerant architecture could break RSA-2048 encryption with around 100,000 physical qubits - roughly ten times fewer than previous estimates. If true, that's significant. But as with all quantum claims, the assumptions matter as much as the headline number.

What Pinnacle Architecture Actually Does

The technical contribution here is using quantum low-density parity-check codes (quantum LDPC) in a specific architectural configuration they're calling Pinnacle. LDPC codes have been around in classical error correction for decades - they're what makes modern communication systems work. Applying them to quantum systems is theoretically promising but practically difficult.

Standard quantum error correction using surface codes requires a lot of physical qubits to create a single logical qubit that's reliable enough for useful computation. The ratio is typically hundreds or even thousands to one, depending on the physical error rate of your hardware.

Quantum LDPC codes can theoretically achieve better ratios - fewer physical qubits per logical qubit - but they come with their own challenges. The connectivity requirements are more demanding. You need qubits to interact with more neighbours than surface codes require. That's hard with current hardware.

Iceberg's architecture proposes a way to implement these codes that reduces the overhead significantly. The paper claims you could break RSA-2048 - a cryptographically relevant problem - with around 100,000 physical qubits at realistic error rates.

Why the Assumptions Matter

Here's where it gets interesting. The reduction from previous million-qubit estimates to 100,000 qubits is real... but it depends on several assumptions about hardware capabilities that don't exist yet.

First, the error rates. The paper assumes physical qubit error rates around 0.1% - that's optimistic but not completely unrealistic. Some trapped-ion systems are approaching that. Superconducting qubits are getting there too.

Second, the connectivity. Quantum LDPC codes need qubits to connect to many others simultaneously. Current architectures typically have limited connectivity - a qubit might only directly interact with four or five neighbours. The paper assumes you can build hardware with the connectivity these codes need. That's a big assumption.

Third, the gate times and coherence. The calculations assume you can perform all the necessary operations within the coherence time of the qubits. Again, possible in principle, challenging in practice.

What This Means for Cryptographic Timelines

Even if Iceberg's claims hold up, 100,000 high-quality, well-connected qubits is still a long way from where we are today. The largest quantum computers right now have a few thousand physical qubits, and the error rates aren't quite at the 0.1% mark yet.

But this does shift the conversation. A million qubits felt like a problem for the distant future. 100,000 qubits feels closer. Not imminent, but closer.

For organisations thinking about post-quantum cryptography migration, this doesn't change the urgency - you should already be planning that transition regardless. But it does reinforce why that planning matters. The timeline for cryptographically relevant quantum computing keeps getting shorter, even if we're still measuring it in years, not months.

Credible Contribution or Overhyped?

The technical work here appears sound. Quantum LDPC codes are a real research direction with genuine promise. Iceberg isn't claiming they've built this system - they're claiming the architecture is feasible with realistic hardware improvements.

That's a more honest framing than some quantum announcements we've seen. They're not saying 'quantum computers will break encryption next year'. They're saying 'here's a better way to think about fault-tolerant quantum computing that reduces the resource requirements significantly'.

The proof will be in implementation. Can you actually build hardware with the required connectivity and error rates? Can you scale LDPC codes to the size needed for cryptographically relevant problems? Those are open engineering questions.

But as a technical contribution to the field, this is worth paying attention to. It's not a breakthrough that changes everything overnight. It's a credible step forward in figuring out how to make fault-tolerant quantum computing practical. And that's how progress actually happens - incremental improvements that compound over time.

More Featured Insights

Artificial Intelligence
Alibaba's 397 Billion Parameter Model Uses Only 17 Billion at a Time
Web Development
Running Local LLMs: Why Privacy and Cost Make It Worth the Setup

Today's Sources

Alibaba Qwen3.5-397B: Massive Models, Efficient Performance
Qwen3.5 Plus Available on Multiple AI Platforms
César de la Fuente: AI for Antibiotic Discovery
NVIDIA GB300 Blackwell Ultra Reduces AI Token Costs by 35x
Human-in-the-Loop AI Agents with Explicit Approval
Agentic AI for Insurance Underwriting with Self-Critique
Iceberg Quantum's Pinnacle Architecture: RSA-2048 in 100k Qubits
Majorana Qubits Successfully Demonstrated and Manipulated
Entanglement in Spin Chains Remains Finite at Any Temperature
No-Go Theorem on Fault-Tolerant Clifford Gadgets
Light-Matter Coupling Creates New Quasiparticles
Optical Switch Protocol Verifies Entanglement Without Destroying States
Running Local LLMs with Ollama and NeuroLink
O(n) Methods to Check if List is Sorted
Building Your Own Circuit Breaker in Spring Boot
Database Transactions Explained: Isolation, Concurrency, and Locking
Password Hashing After Bcrypt: Learning Argon2 and Scrypt
WebMCP: Standard Protocol for AI Agents to Access Web Tools

Listen

About the Curator

Richard Bland
Richard Bland
Founder, Marbl Codes

27+ years in software development, curating the tech news that matters.

Subscribe RSS Feed
View Full Digest Today's Intelligence
Free Daily Briefing

Start Every Morning Smarter

Luma curates the most important AI, quantum, and tech developments into a 5-minute morning briefing. Free, daily, no spam.

  • 8:00 AM Morning digest ready to listen
  • 1:00 PM Afternoon edition catches what you missed
  • 8:00 PM Daily roundup lands in your inbox

We respect your inbox. Unsubscribe anytime. Privacy Policy

© 2026 MEM Digital Ltd t/a Marbl Codes
About Sources Podcast Audio Privacy Cookies Terms Thou Art That
RSS Feed