The grand opening of Europe’s premier manufacturing facility for supercomputers, artificial intelligence factories, and high-performance enterprise systems in Angers, France, marks a remarkable chapter in the continent’s industrial history. Spearheaded by Bull, the historic technology brand now operating under the Atos umbrella, this new production line represents the single largest modern industrial investment in sovereign European high-performance computing (HPC). Ironically, the company’s foundational roots trace back nearly a century to a desperate bid by a French paper manufacturer locked out of a foreign technology monopoly. Today, that enduring quest for digital autonomy has come full circle, highlighted by a landmark €387.8 million contract to install the LUMI AI factory in a repurposed paper mill in Kajaani, Finland.
As global markets experience an unprecedented surge in demand for generative AI infrastructure, policymakers, engineers, and economists are confronting familiar challenges: energy constraints, fierce geopolitical competition, and the delicate balance between domestic industrial policy and international market dynamics. To better understand these recurring themes, computing historian Pierre Mounier-Kuhn of the French National Center for Scientific Research (CNRS) and the University of Paris-Sorbonne offers a comprehensive historical lens on how Europe’s computing landscape evolved from the Great Depression to the current era of exascale computing and artificial intelligence.
The Paper Trail: Origins of European Computing Sovereignty
The genesis of Bull lies not in Silicon Valley garage culture, but in the economic desperation of the interwar period. In the 1930s, Papeteries Aussedat, a French paper mill company, found itself severely restricted by IBM’s ironclad commercial practices. IBM required its customer base to purchase proprietary punch cards, effectively shutting out independent suppliers of consumable paper products. To secure a recurring revenue stream for its mills and bypass the American monopoly, the paper company made a calculated financial investment in a fledgling enterprise founded around the patents of Norwegian engineer Fredrik Rosing Bull.
Fredrik Bull, who served as a technical inspector at the Storebrand insurance company in Oslo, had designed a more cost-effective tabulating machine to handle actuarial statistics. Though initially skeptical due to the high risks associated with new office machinery, Scandinavian insurers backed the engineer’s two-year development cycle. Following Bull’s untimely death from cancer, his collaborator, Knut Andreas Knutsen, sought a broader continental market. Finding Scandinavia too small to scale production, Knutsen partnered with Swiss calculating machine manufacturer Egli, which ultimately established an industrial workshop in Paris to capture the French market while Egli retained Switzerland.
This early industrial marriage was deeply intertwined with national security and sovereignty. Following the hard lessons of World War I—where France discovered critical dependencies on imported German machine tools for manufacturing armaments—government administrators and military officials viewed domestic technological capacity as a matter of vital national defense. Much like the U.S. Census Bureau and the subsequent creation of the Social Security Administration provided foundational, guaranteed markets for IBM, European public administrations and railway networks offered the early, stable demand required for Bull to survive successive commercial near-collapse events.
Expansion, Consolidation, and the American Challenge
By the early 1960s, Bull had outgrown its Parisian footprint. In 1961, the company elected to consolidate its computer manufacturing operations in Angers, strategically capitalizing on a ready pool of skilled labor migrating away from the region’s declining textile sector. When the Angers plant officially opened in 1963, it stood as Europe’s largest computer manufacturing facility, eventually employing over 3,000 workers.
However, the 1960s also ushered in intense structural turbulence. The rapid commercialization of second- and third-generation American computers, combined with the exorbitant research and development expenditures required for the Gamma 60 computer series, severely strained European balance sheets. According to Mounier-Kuhn, several European manufacturers attempted to develop domestic super-calculators during this period, only to face commercial disappointment and unprofitability. Consequently, major European firms abandoned the supercomputer market for the next four decades, leaving the high-end computational domain largely to American and Japanese competitors.
Sensing a lucrative growth frontier outside a saturated U.S. market, American industrial conglomerates made aggressive moves into Europe. In 1964, General Electric (GE) acquired Olivetti’s computer division in Italy and secured a controlling stake in Bull. While the acquisition injected vital capital and advanced American managerial methods into the French firm, it provoked acute political anxiety within the French government. Fearing a permanent loss of technological independence, Paris launched the ambitious Plan Calcul in 1966—a foundational state-backed industrial policy aimed at fostering a sovereign French and European computer manufacturing base independent of U.S. influence.
Tensions escalated further when the United States government denied an export permit for a Control Data 6600 supercomputer to the French Atomic Energy Commission, a move widely interpreted in European political circles as an attempt to impede independent nuclear research capabilities. Rather than directly nationalizing Bull, the French government heavily subsidized a domestic rival, Compagnie Internationale pour l’Informatique (CII). CII later attempted a pan-European alliance known as Unidata alongside West Germany’s Siemens and the Netherlands’ Philips.
When Unidata collapsed in 1975 due to shifting national priorities, CII merged with Honeywell-Bull, following GE’s decision to divest its computer operations to Honeywell. This merger briefly ushered in a golden era for Bull engineers, who collaborated on equal footing with their American counterparts in designing a new generation of enterprise architectures.
The Physics Lesson of Material Substitution
Industrial innovation is rarely a linear path, and historical archives often harbor cautionary tales regarding material substitution and engineering assumptions. In the late 1970s and early 1980s, coinciding with severe global economic disruptions triggered by oil shocks, the Soviet invasion of Afghanistan, and skyrocketing gold prices, Bull engineers at the Angers facility confronted soaring manufacturing costs. The gold used on the substrates of mainframe computer circuits represented a significant expense.
Laboratory tests suggested that copper could successfully replace gold as a cheaper conductive substrate. However, the engineers failed to account for complex solid-state physics phenomena. In extremely fine copper micro-circuits, high electric current densities induce electromigration—the gradual displacement of metal ions caused by the momentum transfer of conducting electrons. Because the design team lacked specialized nuclear chemists or solid-state physicists to anticipate this atomic-level degradation, an entire production batch of advanced mainframes suffered catastrophic reliability failures and had to be scrapped.
The technical crisis necessitated a costly redesign and a swift return to established metallurgical practices, a remediation effort completed just as Bull was nationalized by the French government in 1982. This state takeover occurred amid broader market shifts, as the dominance of large mainframe manufacturers like IBM, Univac, and Bull was aggressively challenged by the democratization of mid-range minicomputers and personal microcomputers. It was not until the early 2000s, under new executive leadership, that Bull executed a successful corporate turnaround by abandoning general-purpose computing to specialize entirely in high-performance servers and supercomputers—effectively breaking the glass ceiling that had constrained European hardware manufacturers for decades.
Full Circle: Energy, AI Factories, and the Kajaani Transformation
Today, the explosive growth of artificial intelligence and large language models has transformed computing once again into a capital- and energy-intensive heavy industry. The modern race to construct massive AI training clusters has elevated power availability to a primary strategic constraint, breathing new life into industrial sites equipped with robust, green energy infrastructure.
This dynamic is exemplified by Bull’s involvement in the LUMI supercomputer and AI factory project in Finland. Valued at €387.8 million, the initiative is hosted at a former paper mill in Kajaani. Originally established in 1907 as a bustling pulp and paper production hub, the mill suffered an economic decline with the digital transition of print media and closed its doors in 2008. The municipal authorities successfully transformed the industrial brownfield into a modern business park that now surpasses the mill’s peak employment figures.
Crucially, the site inherited a heavy-duty 230-megawatt electrical connection tied directly to local renewable hydroelectric generation capacity. This abundant, green power supply made the location an ideal home for what the Central European Supercomputing Consortium (CSC) projects will be one of Europe’s largest and most energy-efficient artificial intelligence factories upon its full completion in 2027.
Implications and Outlook
The historical trajectory of European high-performance computing illustrates a persistent tension between market forces and sovereign industrial strategy. While globalization and cross-border acquisitions have frequently disrupted domestic manufacturing ecosystems, current geopolitical realities and supply chain vulnerabilities have re-elevated digital sovereignty to the top of the European Union’s policy agenda.
Observers note that contemporary European initiatives—such as the EuroHPC Joint Undertaking and substantial investments in domestic AI infrastructure—reflect a concerted effort to prevent historical dependencies from repeating themselves in the quantum and artificial intelligence eras. Whether Europe can successfully maintain a sustainable equilibrium between stringent regulatory frameworks, competitive private-sector innovation, and indigenous hardware production remains a defining question for policymakers and technologists alike. As Europe inaugurates its advanced manufacturing lines in Angers and powers up exascale systems in repurposed industrial hubs, the lessons of the past century underscore that technological leadership requires not only capital and engineering talent, but also long-term strategic resilience and secure control over foundational resources.
