Déjà Vu over La Défense

Psychedelically speaking, a lot of things that have never happened before are presently happening again, all at once.

As the United States and China lock horns in the 2,500-year-old Thucydides Trap, the brewing animosity between various Western nations hearkens back to the European Reformation and even the Wars of Religion. European native population decline, economic constipation, and an escapist indifference toward national defense, coupled with unchecked Third World immigration, draw direct lines to the nightmarish “Fall of Rome.” The unraveling of the world’s unified economic fabric is strikingly similar to how World War I destroyed the Western Garden of Eden circa 1913. The Russia-Ukraine killing fields are littered with dead humans and robots; the increasingly net-centric and unmanned contests there mix 20th-century industrial warfare with a brand-new emphasis on consumer electronics, a reliance on open-source, and artificial intelligence. Finally, the current open-source situation is a proverbial Tragedy of the Commons: everyone wants to consume free, high-quality code, but few are ready to contribute in any meaningful sense.

For many engineering and manufacturing corporations, life starts to look like simultaneously fighting off a white elephant, a gray rhino, and a black swan on the fungal-networked frontier of modern industry. They face the inevitable, irreplaceable “Great Retirement” of their most experienced employees; their enterprise software is often less an empowerment and more an expensive burden; and their hardware and software supply chains feel fragile or have already started crumbling. Paradoxically, the mitigation strategies rooted in artificial intelligence and its associated technologies are arriving at the very same time; one need only embrace them properly.

In that context, while everyone is talking about the USA, China, and occasionally Russia, France has always been a source of visionary concepts. Richelieu pioneered raison d’état, Napoleon unleashed a revolution in military affairs, and de Gaulle and his successors promoted a solid nuclear energy foundation for the economy alongside a comprehensive, integrated military-industrial base. Bernard Charlès’ Dassault Systèmes has become a uniquely innovative software powerhouse that so many champions around the world have come to depend upon.

Many Europeans are upset about the latest American discourse toward Ukraine, NATO, and Greenland. Yet, it was France that abandoned its then-ally Israel in 1967, supported Canada’s separatists, and expelled American forces while openly flirting with the Soviets at the height of the Cold War. Nowadays, the flirtation is with China, and de Gaulle’s “strategic autonomy” means “Sovereign AI” in terms of European-controlled chips, hyperscalers, and their own LLM factories.

Countries dominating “Physical AI” will outperform those that do not in terms of scientific and engineering research by an order of magnitude, essentially creating a divide comparable to that between the 19th-century European colonial powers and their prey. The United States has established “Project Genesis,” China has responded with “AI for Science,” and a collection of Euro-centric “middle powers” apparently bet on a Dassault-led, Mistral-centered “Sovereign AI” push.

While Siemens automates the factory floor with an “Industrial AI Operating System,” PTC modestly pursues the optimization of specific, high-friction points in the lifecycle, and smaller vendors’ AI initiatives promise better parts, improved BOMs, and ultra-streamlined design reviews, Dassault Systèmes firmly positions itself at the epicenter of the new global arms race. The recently announced strategic partnership between DS and NVIDIA explicitly targets science-grounded “Industry World Models” combined with the more conventional Virtual Companions. Here, safe, sovereign, and deterministic DS Companions will be trained on the deep physics and proprietary know-how that the company has curated for forty years.

There is a lot of promise in that drive for Dassault Systèmes customers, yet there is also a lot of danger. First, to put it mildly, it is not clear what kind of retaliation European attempts to challenge American digital juggernauts may invite. Second, the traditional, walled-off DS ecosystem faces an OpenClaw tsunami. The latter is open-source, modular, and integrates, Pandora-like, with everything from Slack to local Python scripts. If engineers find OpenClaw’s ClawHub skills (already featuring thousands of community-built engineering shortcuts) more useful than the official DS agents, the Dassault platform risks becoming a “silent backend” while the actual engineering decision-making happens in the OpenClaw interface.

On a separate note, properly orchestrated OpenClaw access to still-ubiquitous PDF-based information may become particularly powerful for many companies. This will create a bridge over the notorious PDF Moat, where it is seriously beneficial to produce data in PDF format, yet it is quite torturous to consume that same data later. The struggle includes reconstructing technical publications flowing from engineering to manufacturing into either actionable work plans or machine-readable formats, as well as navigating the materials certification documentation that American aerospace companies must provide to satisfy “friendly” European ESG regulations.

Marveling at OpenClaw’s power brings us back to the unfolding open-source debacle. Software developers are still human for the time being, and that means they occasionally need to eat. Companies that create an open-source core plus a more functional Enterprise Edition are bleeding money, to the applause of quasi-consulting shops that skip the hard part and live happily off providing expertise for that very same core. Consequently, more open-source projects are either flipping to commercial-only models or just shutting down abruptly, causing great anguish to their innocent, totally unsuspecting users. Against this backdrop, an organization like the MariaDB Foundation is a gold standard of integrity and tenacity: it preserves independent guardianship of one of the most-used databases worldwide; handles the dirty work of negotiating the best terms for the community with the main corporate contributor, MariaDB plc; maintains a diplomatic dialogue with Oracle regarding the fate of MySQL; and proves reasonably adept at begging for money from sponsors, big and small. Mr. Bean would be proud of them.

In that optimistic spirit, if you are an engineering and manufacturing company dealing with a splintering universe, it is a perfect moment to check your configuration management strategy for the inbound flocks of swans and rhinos. Now is the time to re-evaluate your engineering data and relevant vendors in terms of a long-term archival strategy while reflecting on AP242 benefits for both parts and assemblies. If you are worried about the use of cloud LLM overlords, there are safe options for on-prem hosting for every budget that will prevent your company from falling behind in the AI race. And, if your company depends on an open-source project such as MariaDB, think about contributing code, money, or simply a kind word on social media. Karma goes a long way.

Au revoir!