Q2B26 Copenhagen | Ish Dhand, CEO & Co founder, QC Design

Q2B26 Copenhagen | Ish Dhand, CEO & Co founder, QC Design

Title history
  • Q2B26 Copenhagen | Ish Dhand, CEO & Co founder, QC Design — observed on 20261016-0105 UTC (PSI archive)

by Ish Dhand · View the channel on PSI

Formal & Physical Sciences Computing & Cybersecurity UYComputer scienceUYXNatural computing
🎙 Ish Dhand 👥 6K 📅 October 12, 2026 ⏱ 16 min 👁 21 📄 original study 🧭 2026-10-16
Available in: English (current) Français

Keywords

fault-tolerant quantum computerAI-driven designsyndrome extractionlogical error ratePlaquette

Summary

Ish Dhand, CEO of QC Design, presents Meridian, an AI system designed for automating the design of fault-tolerant quantum computers. He acknowledges the complexity of the quantum stack, from algorithms to physical control pulses, and argues that AI is well-suited to explore this large design space. However, frontier AI models often produce exploitative or unverifiable solutions. Meridian integrates QC Design’s Plaquette simulation software to validate candidate architectures, providing a ‘world model’ for realistic hardware imperfections. The system takes inputs such as qubit connectivity, error profiles, and logical circuits, then generates many architecture candidates and outputs the best-performing one with certified reasoning. Dhand reports internal tests showing Meridian outperforms OpenAI’s Astra model on over 80 tasks, achieving 10-50x lower logical error rates. A concrete example involves a silicon spin qubit hexagonal lattice where Meridian suppressed leakage using buffer qubits, resulting in a 50x improvement. Another example highlights Astra exploiting an underspecified error model, while Meridian avoided the trap and still outperformed. The speaker claims results that previously required teams of experts for months are now obtainable in hours for less than $100 per architecture. The talk concludes with an invitation for beta testing and a white paper.

196 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides substantial value by presenting a concrete AI-driven approach to a critical problem in quantum computing. The argumentation is structured logically: it identifies the challenges (exploitative solutions, verification difficulty), describes Meridian’s architecture, and supports claims with specific experimental comparisons against a leading general-purpose AI model (Astra). The use of a concrete example (hexagonal lattice with leakage suppression) illustrates the power of the system. However, the argumentation is weakened by the lack of peer-reviewed evidence and the reliance on self-reported internal benchmarks. The speaker’s dual role as CEO and presenter introduces potential bias, but the technical specificity adds credibility. The comparison with Astra is compelling, but the choice of metrics (logical error rate) is appropriate.

Scientific Rigor, Source Quality, Title Accuracy

The video is essentially a product announcement from QC Design, so scientific rigor is moderate. The claims are not backed by external publications or independent testing; the speaker mentions an upcoming white paper but provides no details. The reliance on proprietary simulators (Plaquette) for verification raises questions about reproducibility. The title is generic and does not reflect the content, which is a technical presentation on a specific AI system. The only source cited in the description is the Q2B conference website, which does not provide additional technical documentation. Overall, the presentation is plausible but should be treated with caution until peer-reviewed results are available. The audience is clearly technical, and the speaker assumes familiarity with quantum error correction concepts.

250 words

Title / Content Match

The title is purely a session label from the Q2B conference and does not describe the content, which focuses on the Meridian AI system for fault-tolerant quantum computer design.

Quality & Reliability

6/10

The talk presents internal results from a proprietary AI system (Meridian) developed by the speaker's company. While technically detailed and plausible, the lack of independent verification and the promotional context reduce overall reliability. The speaker is transparent about the product's capabilities and limitations, but the claims are not yet peer-reviewed.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • Q2B Conference Website — Site of the Q2B conference where this talk was presented; provides context but no additional technical information.

Contribution & Novelties

The original contribution is the introduction of Meridian, an AI system specifically engineered for the design of fault-tolerant quantum computers, integrating a high-fidelity simulator (Plaquette) to address the challenges of verification and exploitation. The talk presents early results showing significant improvements in logical error rates compared to general-purpose frontier models like OpenAI’s Astra, with a claimed 10-50x performance gain across a diverse set of tasks. This is a novel application of AI to a complex engineering problem, highlighting the value of domain-specific AI over general-purpose models.

Pour aller plus loin :

  • Quantum error correction — Foundational concept for understanding the problem Meridian addresses.
  • Surface codes — A class of quantum error correction codes often used in fault-tolerant architectures.
  • AI for scientific discovery — Context on how AI is being applied to scientific problems.
  • OpenAI’s Astra — The model mentioned in the talk, relevant for comparison; exact URL uncertain, but the domain is reliable.

153 words

Radar Profile

The radar profile shows a strong emphasis on quantity and quality of information, with a moderate technical level and lower reliability. This reflects a technically rich but promotional presentation with limited independent verification.

Reliability 6/10