AI that matters, from the architect's desk. Curated and engineered by Saaket Varma, PhD — no hype, just signal.
HOW TO READ THIS Read top to bottom: Sandia certifies, then the 98-qubit chip, then the gate test, then the fidelity result.
Sandia National Laboratories independently certified Quantinuum's 98-qubit Helios trapped-ion system at 99.92% two-qubit and 99.9975% single-qubit gate fidelities, with the results peer-reviewed in Nature. That matters because these are the error rates fault-tolerant quantum computing actually requires — and for the first time the numbers carry a U.S. national lab's signature, not just a vendor's press release. If your tech roadmap still files quantum under 'a decade out,' this quarter is the time to revisit that assumption. And until a result is independently and peer-reviewed validated, treat any quantum performance claim as marketing.
HOW TO READ THIS Read top to bottom: xAI ships Grok 4.3, it joins Bedrock's model catalog, it plugs in directly instead of through the third-party adapter layer, and the payoff is a 1M-token context window.
xAI becomes a first-party Bedrock provider — a 1M-token frontier model under your existing AWS account, billing, and controls, with no new contract to sign.
HOW TO READ THIS Read top to bottom: D-Wave ships a simulator through Leap, it renders a 21-qubit gate circuit in software, letting developers test before real gate hardware exists.
The annealing pioneer moves into gate-model development, letting builders design around real processor noise now — ahead of its September arrival on the Leap cloud.
Stanford's LLM knowledge-curation system that researches a topic and writes a full, cited report end to end.
754 structured cybersecurity skills for AI agents, mapped to MITRE ATT&CK, NIST CSF 2.0, and MITRE ATLAS.
An agent development environment for running a fleet of parallel coding agents on your own subscription.
Garry Tan's exact Claude Code setup — 23 opinionated tools acting as CEO, designer, eng manager, and release manager.
A review-first terminal diff viewer built to keep a human in the loop on agent-written code.
Interconnects Nathan Lambert's case for why banning open-source AI would be a policy mistake — timely reading as regulation debates heat up.
The Sequence Last week in AI: a reported $60B Cursor deal, Google talent attrition, and Midjourney's body scanner.