Dado Van Peteghem

The through-line

Across his recent writing, Van Peteghem keeps returning to one question: who is actually positioned to compete at scale once you look past the headline story. He applies this to nations, asking not "whether China can compete in AI" but "how far they already are" [7], and to infrastructure, insisting that "not all compute is created equal" and that most conversations wrongly "treat 'data center' as one thing" [11]. The same instinct shows up when he turns the lens on himself, treating a website rebuild with Claude as a live experiment in what AI can now do in three hours rather than months [4]. What has shifted is the register: his talks framed transformation in organizational terms, entrepreneurial units, ecosystems, culture, while his current weekly output is a running, almost real-time tracking of AI and robotics developments, especially the US-China race and the physical AI wave [2][5][7][8][10][11][14][15].

On the AI and robotics race

Van Peteghem treats China's AI progress as the central story of the moment, arguing that Chinese labs and manufacturers now "form a serious alternative to the US ecosystem" with "many of the leading Chinese models" being "open weight," meaning "anyone can download them, run them on their own infrastructure and build on top without asking permission" [7]. He pairs this software story with hardware, noting that "Unitree Robotics, the leading humanoid player from China, listed in Shanghai" with its stock opening "629% above its IPO price" [7], and separately tracking that the leading Asian humanoid players "all sit between $2.9bn and $6.4bn" in value, with the shift being "who sets that number" now that several are "listed or mid-IPO" and "priced by a market instead of annually by a funding round" [15]. On infrastructure specifically, he breaks Europe's roughly "3,360" data centers into four workload types and warns that "on paper that reads like infrastructure strength" until you look at "what is actually inside them" [11]. He extends the same physical-AI curiosity to robotics demonstrations, flagging that Galbot's humanoids "reportedly managed 100+ consecutive autonomous rallies" against human tennis players and calling the evolution of "Physical AI... definitely worth following as the next wave of development" [2].

On building with AI tools himself

Beyond tracking the industry, Van Peteghem treats his own workflow as a testing ground. Rebuilding his website with Claude, he says it "almost one-shotted the website in 10 minutes" once given his "existing website," "branding," and reference sites, and that "the first version was already remarkably close to what I wanted" [4]. He describes the iteration process as feeling "more like giving feedback to a colleague than building a website," citing prompts like "Make this section smaller" or "Use more of my purple," with the tool "adapting on the fly" [4]. He also notes it produced fully responsive layouts across "desktop, tablet, mobile" without being asked, and that it "pulled the right pictures from my photo folders" on its own judgment [4].

From the stage

In talks, Van Peteghem makes arguments about organizational structure that his written posts do not cover. On corporate venturing, he uses the image of "a speedboat operating alongside a mothership rather than being absorbed into it," arguing that entrepreneurial teams should stay separate from a larger company's structure while still benefiting from its scale [16]. On transformation itself, he frames it as "fundamentally about organizational culture, process speed, and people rather than just technology itself," illustrating the point through his family's publishing business, which had to introduce VHS and DVD products after the internet and gaming devices disrupted its colorbook and comic album market in the 1980s [17]. And on competitive strategy, he argues that future competition happens "at the ecosystem level rather than among individual companies," tracing this from internet connectivity in the 1990s through social media in the 2000s to today's wave of organizational collaboration, and advising audiences to "leave their buildings and step outside their industry bubbles to meet unusual people" [18].

Takeaways

• Don't just ask if a competitor can catch up, ask how far ahead they already are; Van Peteghem applies this to China's AI ecosystem, where many leading models are open weight and downloadable without permission [7].
• Treat "data center" as four distinct workload categories, general-purpose cloud, frontier training, post-training/AI gyms, and inference, before drawing conclusions about a region's AI infrastructure strength [11].
• When evaluating AI coding tools, judge them on how closely the first draft matches intent and how naturally they take iterative feedback, not just raw speed [4].
• In corporate innovation, keep entrepreneurial units structurally separate from the core business, a speedboat alongside a mothership, rather than integrating them [16].
• Frame digital transformation projects around culture, process speed, and people, using concrete precedent (like a legacy product line forced to adapt to new formats) rather than leading with the technology itself [17].
• Push teams to physically leave their building and industry bubble to find ecosystem partners, since competition is increasingly won or lost at the ecosystem level [18].