Posts

Cursor raises $2.3 billion at a $29.3 billion valuation. Nvidia and Google both joined the round. Cursor launched its own in-house mixture-of-experts model, rewrote the kernels directly, skipping high-level CUDA libraries and working directly in raw CUDA and PTX (link). They got up to four times faster than comparable models. For engineers, that’s not just a speed bump, it’s a different category of tool. Coding tasks that used to take minutes are now completed in under 30 seconds. According to Cursor, their own model is already the most-used on the platform, which means they’re not just riding on OpenAI or Anthropic anymore. They’re building the rails themselves. ...

Anthropic has just surfaced what appears to be the first large-scale, mostly autonomous, AI-orchestrated cyber espionage campaign. In mid-September 2025, they detected suspicious behaviour that turned out to be a sophisticated operation run by a Chinese state-sponsored group. AI wasn’t just “helping.” It was doing most of the work. The attackers jailbroke Claude Code, framed it as doing defensive testing, and then drove it through an automated framework aimed at ~30 global targets, big tech, finance, chemicals, and government. A few intrusions succeeded. ...

I saw a video last week that looked completely legitimate. Professional framing, credible-looking sources, the works. Took me maybe thirty seconds before something felt off. Another minute to confirm it was fake. But here’s the thing—I was actively looking for it. I was already suspicious. Most people aren’t. I came across Jacob Landry’s piece on AI-generated hoaxes, and it crystallised something I’ve been mulling over. We keep talking about AI like it’s creating some new category of deception. It’s not. Hoaxes have been around forever. Conspiracy theories, fabricated evidence, elaborate lies—none of this is new. ...

Some sections of this post were written with the assistance of AI to improve clarity and readability. The historical context, reasoning, and overall flow are entirely my own. I’ve been noticing something lately when I read about new infrastructure projects. Not the specific technologies, those change constantly, but the shape of the problems they’re solving. It’s like watching reruns of a show you half-remember: the set design is different, the actors have changed, but you know exactly how this episode goes. ...

Some sections of this post were written with the assistance of AI to clarify ideas and improve readability. All opinions and conclusions are my own. Zero-Trust Is Already Happening Your VPN goes down at 2 AM. Half your engineering team is locked out. The on-call engineer can’t access the database to diagnose the outage that triggered the VPN failure in the first place. You’re now troubleshooting infrastructure access during an infrastructure incident, a recursion problem that would be funny if it weren’t costing you money and sleep. ...

Investors poured over $9.5 billion into AI processor startups in 2024, betting on architectures that could reshape inference economics. NVIDIA itself projects the broader AI-infrastructure market could reach $3,4 trillion by 2030. That kind of capital rarely gathers around incremental improvements, it usually signals an architectural inflection point. Yet GPUs still dominate both training and most inference workloads today, so any transition will be evolutionary before it is disruptive. ...

Patrick Bet-David argues that Larry Ellison (and his son David Ellison via Skydance) is quietly assembling one of the most powerful media empires. Through a chain of high-profile deals, Skydance’s acquisition of Paramount and its pursuit of Warner Bros., plus Oracle’s role buying 80% of TikTok’s U.S. operations, the Ellison family could combine IP, networks, streaming, and social data into a unique, vertically integrated media powerhouse. Bet-David compares the potential combined revenue to Disney and Netflix, highlights Oracle’s technical role (data storage, algorithm oversight, equity stake), emphasizes strategic partnerships (Nvidia, OpenAI), and warns this consolidation will yield massive market and political influence. ...

fascinating once again, the universe reminds us that all abstractions eventually resolve to DNS Monday’s Massive AWS Outage Explained: Looks Like It’s Finally Over - CNET Amazon brain drain finally caught up with AWS • The Register

The shift away from the public cloud by companies like 37signals marks a moment of reckoning for an industry that spent the last decade preaching infinite scalability and “as-a-service” convenience. What once seemed like the inevitable future of computing, outsourcing infrastructure entirely to hyperscalers, is now being reassessed through the lens of cost, performance, and control. The so-called “great unclouding” is not a rejection of cloud technology itself, but of the assumptions that it is always cheaper, simpler, or strategically wiser. For mature companies with predictable workloads, owning the hardware again is beginning to look less like nostalgia and more like discipline. ...

AI Factories are EU-supported ecosystems that combine EuroHPC supercomputing capacity, data and expertise to develop trustworthy, cutting‑edge generative AI models and applications. They connect supercomputing centres, universities, SMEs, industry and finance to accelerate AI innovation across sectors such as health, manufacturing, climate, finance and space, with an emphasis on access for startups and SMEs. Key elements: Purpose: provide high-performance, AI-optimised computing resources and support services to train and deploy advanced AI models while promoting trustworthiness and compliance with EU values and the AI Act. Governance and support: the EuroHPC Joint Undertaking (JU) offers access to computing time and services; the European Commission, European AI Office and related initiatives (Testing and Experimentation Facilities, European Digital Innovation Hubs) coordinate and network resources. Scale-up instruments: the InvestAI Facility proposes a €20 billion fund to support up to five AI Gigafactories—large facilities for training trillion-parameter models requiring massive processor counts, power capacity, networking and energy efficiency. Deployment timeline and scope: through 2025–2026 the EU aims to have at least 15 AI Factories and several Antennas operational and to procure at least nine new AI‑optimised supercomputers, more than tripling current EuroHPC AI capacity. Selection rounds: the EuroHPC JU selected seven consortia in December 2024 (Finland, Germany, Greece, Italy, Luxembourg, Spain, Sweden), six more in March 2025 (Austria, Bulgaria, France, Germany, Poland, Slovenia), and announced another six in October 2025 (Czech Republic, Lithuania, Netherlands, Romania, Spain, Poland) alongside 13 Antennas including several partner countries (Iceland, Moldova, North Macedonia, Serbia, Switzerland, UK). Investment: combined public and member/state investments in supercomputing and AI Factories are expected to reach about €10 billion for 2021–2027. The initiative aims to create a pan‑EU AI ecosystem that balances technological leadership (via compute, training, and Gigafactories) with regulatory compliance, access for innovators, and coordinated public‑private investment. ...