Notes RSS feed

The Wire — short items from the day's overflow, verdict follow-ups on earlier calls, and changelogs for the paper's maintained data assets. When a story earns a full argument, it runs as an article instead.

Wire

Pathway reaches a $500M AI valuation
08/14/26, 07:34 AM

AI lab Pathway raised fresh funding at a $500 million valuation, bringing total seed financing to $30 million, with new money earmarked partly for NVIDIA GB300 compute. The funding report arrives alongside Pathway’s claim that its 150-million-parameter BDH-CQ model scored 29.5% on ARC-AGI-1 at $0.0007 per task; its technical announcement says the comparison is computed hardware cost, not a general production price. Builders following the archive’s reproducibility gate for AI mathematics should treat the valuation as a bet on architecture, not proof that latent reasoning scales beyond one benchmark.
Wire

OpenAI hires Wiz's operator as revenue chief
08/14/26, 07:31 AM

OpenAI named former Wiz President and COO Dali Rajic its chief revenue officer as the company says its products reach more than one billion weekly active users and more than two million businesses—both counts are 2× their levels a year ago. In its announcement of Rajic’s appointment, OpenAI says both business counts are twice what they were a year ago; the hiring signal is a shift from proving demand to building a repeatable revenue operating system. Teams tracking ChatGPT’s ads-and-intent market should file the move as evidence that enterprise distribution, not just model capability, is becoming the next constraint.
Wire

Anthropic eyes Decart for a $6B buyout
08/14/26, 06:36 AM

Anthropic is reportedly in talks to acquire Israeli AI startup Decart for about $6 billion, potentially its largest acquisition and first major foothold in Israel. The Times of Israel reports that the talks are not final; Decart has raised more than $450 million, employs 106 people, and builds chip-optimization and real-time video technology. For teams tracking Anthropic’s push to turn compute scale into an enterprise moat, the new buyer disclosure matters more than the headline valuation: Anthropic may be buying inference efficiency and a regional performance team, not just another model.
Wire

Sterling raises $3.8M for AI accounting
08/14/26, 05:36 AM

New Zealand accounting startup Sterling raised a $3.8 million seed round after saying its software had already completed about 1,200 hours of finance work for customers. RNZ reports that the company was founded 10 months ago and plans to expand into Australia and the United States, positioning the product as an autopilot for repetitive finance work rather than a chatbot. Finance leaders should treat the early usage figure as a proof-of-demand signal—not an accuracy guarantee—alongside the enterprise agent controls needed to make automation accountable.
Wire

USC and ASCM launch a supply-chain AI benchmark
08/14/26, 05:33 AM

USC Marshall and the Association for Supply Chain Management launched an annual benchmark for agentic AI adoption, investment, and impact across global supply chains, with its first report due in December 2026. The ASCM launch announcement says the study will track use cases, organizational maturity, workforce effects, and returns rather than another one-off adoption poll. Supply-chain teams should file the December release as a possible denominator for comparing automation claims with the operational evidence behind them; the case for governed enterprise agents still depends on that missing measurement layer.
Wire

Vantage explores a $100B AI data-center IPO
08/14/26, 04:44 AM

Vantage Data Centers is holding early discussions about an IPO or sale at roughly $100 billion, potentially raising about $10 billion, as AI demand pushes data-center infrastructure toward public-market scale. The reported options say Vantage runs 35 campuses on five continents and has taken in roughly $11 billion since late 2023, while any deal could still disappear. AI infrastructure buyers should read this as a capital-market signal, not delivered capacity: Oracle’s $55.7 billion annual AI buildout is the better reminder to separate contracted power and revenue from valuation headlines.
Wire

Legora seeks $10B valuation for legal AI
08/14/26, 04:41 AM

Swedish legal-AI startup Legora is in early talks to raise at a valuation above $10 billion, only four months after a $5.6 billion mark, while annual recurring revenue reached $150 million in the second quarter. The reported financing discussion says Legora’s customer base grew 25% in three months to roughly 1,500 firms and in-house teams; a $10 billion price would be about 67× ARR. Legal-technology buyers should treat the figure as a demand signal for workflow-specific agents—not a reason to skip portability and liability checks—as Harvey’s $15.5 billion legal-AI valuation makes the competitive pricing pressure explicit.
Wire

Google trains SensorFM on a trillion wearable minutes
08/14/26, 04:38 AM

Google Research trained SensorFM on more than one trillion minutes of multimodal wearable signals from five million consented participants across over 100 countries. The SensorFM research release says the frozen representation outperformed feature-engineered baselines on 34 of 35 health tasks, while an agentic “classroom” explored more than 30,000 prediction-head candidates. Health-AI builders should file the result as a data-and-label advantage, not a clinical clearance: the archive’s ChatGPT Health shadow-clinic analysis is the reminder that a strong representation still needs a deployment evidence gate.
Wire

AMD offers £5M in compute to Oxford AI lab
08/14/26, 03:43 AM

AMD signed an MOU to provide up to £5 million in compute resources to the University of Oxford-led British Open-ended Learning and Discovery Lab (BOLD). The announcement says support would flow through AMD infrastructure, cloud providers, public-sector partnerships, and hardware contributions across a phased program; the MOU targets open research into multi-agent systems, adaptive physical AI, and alternatives to backpropagation. Researchers tracking AMD’s scientific-AI capacity should read the figure as access promised, not delivered training time—the milestone is whether the agreement becomes reproducible work on AMD’s stack.
Wire

Cascadia pools Intel PCs for private inference
08/14/26, 03:40 AM

Community Labs launched Cascadia, an open-source runtime with 3 serving modes that pool Intel CPUs, integrated GPUs, NPUs, and Arc GPUs into a private inference fleet. The launch report says teams can run one model on one device, replicate it across devices, or shard a larger model across the network; the Cascadia runtime is OpenAI API-compatible. Teams weighing local inference on long-context open models should treat idle PCs as a testable fourth path between cloud APIs and dedicated accelerators, then demand fleet-level latency, failure-recovery, and energy data before calling the pool production-ready.