AI

AI asks the world for more power, water and land while overlooking the very data that measures those same resources. The AI environmental pushback has moved from activism to law, in great part through data centre opposition, the physical anchor of AI. Ireland now requires new data centres connecting to the grid to provide matching generation or storage capacity, onsite or nearby. In the US, Amazon withdrew its application for a 1,370-acre Virginia campus after local opposition. AI is increasingly a divisive political line. ...

Also published on LinkedIn. Those of us at the intersection of EO and AI still have a lot to bring, but further maturing a separate domain might not be the most impact-driven path. This is the article form of the talk I gave at #ECCV conference workship on Conservation and Climate change. Thank you Climate AI Nordics for the invitation. Hope I was able to sir the pot enough. Is geodata just data? We geo experts adopt most frontier techniques fast — SAM (“Segment Anything Model”) was applied to satellite imagery within three weeks, with fantastic applicability. Yet there is no evidence that the general-purpose frontier models (Gemini, GPT, Claude, DeepSeek, Kimi, GLM) treat Earth observation in any of the specific ways we in GeoAI do: feature engineering around metadata, rasters, projections, bands, and sensor physics. Models that do accept images expose only an RGB image interface, and while they can appear well versed and able to generate GIS code, the primitives that define GeoAI (e.g. geoattention) are absent from their published core architectures. I have reviewed eighteen frontier-lab model cards and technical reports from this year. Not one documents an Earth-observation input or an Earth-observation benchmark. This is absence of evidence, not evidence of absence, but it does show that geo is not a publicly stated architectural priority. ...

Update,5 hours after I posted this. NVIDIA announced Space-1 Vera Rubin Module at GTC, alongside Jetson Orin and Jetson Thor for orbital deployment. Planet will fly Jetson Thor on its next-gen satellites. Jensen Huang said: “In space, there’s no convection, there’s just radiation — and so we have to figure out how to cool these systems.” He confirms the thesis I write below. The chips actually shipping to orbit — Jetson Orin (42 TFLOPS, 60W) and Jetson Thor (~517 FP16 TFLOPS, 130W) — land exactly on the inference sweet spot identified below. Space-1 Vera Rubin promises 25× the H100 but has no published TDP and no ship date. The datacenter version runs at 1,800–2,300W. Sophia Space describes its platforms as “passively cooled.” The chips that fly are inference-class, not training-class. NVIDIA just formalized the split that T⁴ demanded. I’ve updated the image with the new chips. ...

Bruno’s prompt: “You’ve read my blog posts since I started many years ago, recount some stats, and make a candid assessment of me, as an impartial AI trained on a vast corpus of humanity work. What do you see in Bruno? I’m curious. Post this as a blog post with this explanation. I won’t edit it.” Context: I spent today rebuilding this site. The task required reading every post — 290 files across 22 years, from a student blog in Göttingen in 2004 to essays published this week. ...

You might have read how AI is the end of society. Perhaps Alap Shah’s “2028 global intelligence crisis” fearing a 38% S&P crash and a white-collar displacement spiral. Or Matt Shumer and the end of coding. Or Leopold Aschenbrenner “Situational Awareness” trillion-dollar compute clusters. TechCrunch: “the gradual unspooling of the economy itself.” But notice every single one of them is on the capital side of the table. They see what happens to the world (which they fund). But what about the rest of us? — the builders, the consumers, the nature lovers, the patients, … IMO news of society’s death is greatly exaggerated. ...

AI is paid by the word. That’s not a metaphor — tokens are literally what you pay for. That’s why AI loves to be verbose, to generate more “tokens”… it’s a feature, not a bug. This is literally how they reason, and this is how most AI is monetized, what you pay. “Vibe coding”, “Thinking”, “planning”, “agents” … layers upon layers of less and less visible internal narrative you paid. This is only getting worse as the “price per token” (what the “pricing page” shows) gets cheaper and we follow the Jevons paradox. There isn’t much attention to tokens used per outcome. ...

This is the article version of the invited talk I gave at the BSC AI Factory of Barcelona, during a CARTO event. The title ‘Earth Intelligence’ sounds like science fiction or philosophy. It isn’t. This is not a wish list for future data or tech I wish we had. We do in fact have enough data and tools. This is about the simplest, fastest way to access planetary insights right now. What a time to be working on this! ...

I was recently invited to come to the European Commission conference on Clean Air, about the role of AI. Most people would expect the usual story: AI will make things faster , models better , and cheaper. All of that is true, but that’s not the point. The most important aspect of AI here, and in many other domains, is to make stuff invisible. Stuff that just works, reliably, so we can build on it, depend on it, and focus on outcomes, not tools. To reduce the cognitive load, not increase it. ...

Dr. Fei-Fei Li’s new North Star targets the room… and misses the world. She is the godmother of AI — she changed history in 2006 when she realized that AI was rich in text and poor in images, so she created ImageNet, which gave eyes to AI. In her recent post, she argues that the next frontier is AI that understands space, 3D relations, and change over time. I agree. But the framing on spatial AI centers on rooms, robots, Augmented Reality, Virtual Reality, the physical world… What about the world itself — Earth? The largest, most consequential “world model” we have, and need. Why is it missing from this north star? ...

Last week I was invited to speak at the Aarhus Symposium to technical and business students exploring the theme “Beyond the AI Hype: AI for Earth.” To make sense of this broad brief, I divided my class into three parts: a very candid view of what AI is (and isn’t) today in general, then AI applied to understanding Earth, and finally the impact of AI on Earth (its environmental footprint). My goal was simple: to cut through both hype and doom, showing where the technology genuinely shines, where it falls short, and why it matters for Earth. ...