Executive Summary
The distribution of France 2030’s €54 billion between startups and established industrial groups is one of the plan’s most contested political economy questions. The data reveals an unsurprising pattern: by number of projects, startups and SMEs dominate France 2030’s portfolio — accounting for over 70% of funded entities. By value of funding, large groups and their industrial consortia account for the majority of capital committed. This distribution is not a design failure — it reflects the reality that building battery gigafactories, semiconductor fabs, and bioproduction facilities requires the balance sheets, supply chains, and regulatory capabilities that only large organizations possess. But it does raise legitimate questions about whether France 2030 is primarily creating tomorrow’s industrial leaders or primarily subsidising today’s.
The Distribution in Numbers
Bpifrance’s portfolio data provides the clearest picture. As of early 2025, France 2030 had funded approximately:
- 4,500+ startups and SMEs (companies with under 250 employees), accounting for approximately 70% of projects by count
- 500+ medium-sized companies (ETIs, 250-5,000 employees), approximately 15% of projects
- 200+ large groups and their subsidiaries (5,000+ employees), approximately 10% of projects
- 200+ research organisations, universities, and consortia, approximately 5% of projects
By value of funding committed:
- Large groups and industrial consortia: approximately 60-65% of committed capital
- ETIs: approximately 15%
- Startups and SMEs: approximately 15-18%
- Research organizations: approximately 5-7%
This distribution is consistent with France 2030’s structure. The plan’s flagship investments — the STMicro-GlobalFoundries Crolles expansion (€2.9B public), ACC gigafactories (€1B+ public), ArcelorMittal Dunkirk DRI (€850M public) — are necessarily large-group investments. Removing these structural megaprojects from the analysis, the startup-SME share of non-megaproject funding is approximately 40-45%.
The Case for Prioritising Startups: Innovation at Speed
The startup case for France 2030 funding priority rests on three empirical observations:
Startups create disproportionate innovation value. The canonical examples — Mistral AI, Pasqal, Alice & Bob, Verkor — demonstrate that France’s most globally recognized France 2030-adjacent companies are predominantly recent startups rather than established groups. Mistral AI, founded April 2023, achieved a higher global profile in 18 months than legacy AI research programmes running for decades. Pasqal’s neutral atom quantum computing approach represents genuine innovation that no incumbent would have developed.
Startups respond faster to market signals. A startup pivots in weeks; a large group changes direction in years. In sectors where technology and competitive dynamics are moving rapidly — AI, quantum computing, green hydrogen, new battery chemistries — startup agility is a genuine competitive advantage. France 2030’s flagship startup competition instruments (I-Nov, Deeptech Seed, French Tech Acceleration) are appropriately targeted at this speed premium.
The valuation multiplier on startup investments is higher. €5 million in Bpifrance early-stage equity in a company that becomes a €1 billion company generates a 200x return to public investment. €500 million in grants to a large group generating 0.5% efficiency improvements at existing facilities generates minimal public return. The financial logic of innovation economics strongly favours early-stage startup investment on expected value grounds, even accounting for high failure rates.
The counter-argument: startup failure rates in deeptech are high — typically 70-80% of funded companies fail to achieve commercial viability. The 20-30% that succeed create enormous value, but public investors cannot easily identify ex ante which companies will succeed. Competition mechanisms that evaluate technical merit are imperfect selection tools in domains where success depends as much on commercial execution as scientific validity.
The Case for Prioritising Incumbents: Scale, Supply Chains, and Market Access
The incumbent case is equally compelling at the level of France 2030’s most ambitious industrial objectives:
Large-scale industrial production requires large organizations. A battery gigafactory producing 16 GWh annually requires: supply chain agreements with lithium, cobalt, and manganese suppliers; customers (automakers) who will commit to multi-year offtake contracts; manufacturing engineering experience building complex production facilities; access to hundreds of millions in project finance; and regulatory relationships with environmental and safety agencies. Startups can conceptually develop these capabilities, but incumbents have them now. Verkor, the closest to a startup among the major battery producers, needed €2 billion+ in capital and partnerships with Renault (offtake) and various industrial suppliers before breaking ground.
Technology transfer from incumbent to startup is often the innovation chain. Many of France 2030’s most valuable startup successes trace their origin to incumbent spinoffs or university partnerships: Pasqal was founded by researchers from the CEA and Laboratoire Charles Fabry; Genvia is a CEA-Schlumberger joint venture; Nuward is an EDF-CEA-Naval Group consortium. The clean separation of “startups” and “incumbents” misrepresents the innovation chain in deeptech, where large research organisations (CEA, CNRS, INRIA) generate fundamental research that startups and incumbents commercialize jointly.
Foreign investment is more reliably attracted by incumbent-scale projects. The Choose France summits’ €50+ billion in cumulative FDI commitments are dominated by large-scale investments by global corporations (Microsoft’s €4B cloud/AI commitment, GlobalFoundries’ semiconductor expansion, ProLogium’s battery gigafactory). These decisions are made by global corporate boards responding to investment environments that small startups cannot single-handedly create. France 2030’s largest-value investments in incumbents are also France’s most effective FDI attraction tools.
Sector-by-Sector Distribution: Where France 2030 Bets on Which Player Type
France 2030’s sector allocations reveal implicit choices about startup vs. incumbent balance:
AI and Quantum Computing: Strongly startup-oriented. The AI sector’s France 2030 support is dominated by startups — Mistral AI, Kyutai, Alice & Bob, Pasqal, Quandela — rather than incumbents. This is appropriate: the frontier AI and quantum computing sectors do not have established French incumbents with relevant capabilities. The research infrastructure (Jean Zay, INRIA, CEA) supports both, but the commercial players are predominantly startups.
Batteries: Startup-incumbent hybrid. Verkor is a startup (founded 2020), but at €2 billion+ it requires large-organization capabilities. ACC is a joint venture of Stellantis and TotalEnergies — definitionally incumbent. The battery sector requires both: startups pushing new chemistries and manufacturing techniques, incumbents executing gigafactory scale.
Semiconductors: Strongly incumbent-oriented. STMicro, GlobalFoundries, Soitec — all large established companies. Semiconductor manufacturing at scale has no startup-appropriate entry point due to fab construction costs exceeding €1 billion. French semiconductor startups (Lynred in infrared, several photonics companies) receive France 2030 support but are niche players relative to the sector’s capital-intensive core.
Nuclear: Incumbent-dominated, with startup components. EDF, Framatome, CEA, Naval Group — all large organizations — anchor the EPR2 and Nuward programmes. But NAAREA (molten salt micro-reactor), Jimmy Energy, and Newcleo represent a genuine nuclear startup layer that France 2030 has supported through dedicated competitions. This is a novel phenomenon: competitive nuclear startup ecosystems did not exist before France 2030 created the market signal.
Health and Biotherapies: Mixed. Sanofi’s bioproduction investments (incumbent) anchor the sector alongside DNA Script, Ose Immunotherapeutics, and GENFIT (startups). The health innovation competitions — particularly the Biotherapies and Bioproduction track — have funded both with reasonable balance.
The Hidden Beneficiary: Research Organizations
A third category often overlooked in the startup-incumbent debate is research organizations: CEA, CNRS, INRIA, INSERM, Inrae, and their network of 17 Instituts de Recherche Technologique. These organizations are neither startups nor commercial incumbents, but they receive substantial France 2030 funding — both directly through ANR research competitions and indirectly through participation in industrial consortia.
Research organization participation is structurally important because it ensures that France 2030 investments maintain connection to the scientific frontier. The alternative — funding only commercial companies — risks funding incremental optimization of existing approaches rather than fundamental innovation. The risk of funding primarily research organizations is the opposite: valuable research that never becomes commercially deployed technology.
France 2030’s design attempts to bridge this gap through “collaborative research” competitions that require consortium membership combining research organisations and commercial companies. The evidence suggests this approach succeeds when the commercial partner is genuinely invested in commercialization timelines — and fails when the commercial partner participates for grant access rather than genuine innovation partnership.
The Political Economy: Why Incumbents Will Always Win on Capital Volume
The political economy of France 2030 structurally favours incumbents on capital allocation, regardless of the plan’s stated intentions. Large industrial groups:
- Have dedicated teams to monitor and respond to France 2030 competitions
- Employ lobbyists who participate in competition design consultation phases
- Have relationships with Bpifrance investment officers developed over decades
- Can absorb the application cost (100+ page applications, months of senior team time) more easily than startups
- Are more likely to be included in the “national champion” frameworks that receive the largest direct commitments
This is not corruption — it is institutional gravity. The companies best equipped to navigate a complex competition system are those with the most resources. France 2030’s explicit startup-specific tracks (I-Nov, Deeptech Seed, the French Tech Accélération programme) partially compensate by creating ring-fenced competitions where large groups are ineligible, but the largest capital flows remain accessible primarily to organizations with large administrative capabilities.
The Honest Assessment: Both Are Necessary
The startup vs. incumbent framing is a false dichotomy. France 2030’s most successful investments are neither pure startups nor pure incumbents but hybrid structures that combine startup dynamism with incumbent resources:
- Verkor: founded 2020 (startup origin), but backed by Renault (incumbent), Schneider Electric (incumbent), and EIT InnoEnergy
- Genvia: CEA-Schlumberger joint venture — research institution + energy incumbent
- Nuward: EDF-CEA-TechnicAtome-Naval Group — four incumbents creating a new entity with startup-like mandate
- Mistral AI: founded by researchers from Meta and DeepMind (tech incumbent alumni), backed by Andreessen Horowitz (US VC), Xavier Niel (French tech incumbent), and Iliad (French telecom)
The pattern is consistent: the most effective France 2030 investments combine cutting-edge scientific capability (startup or research origin) with market access, supply chain relationships, and capital access (incumbent or large investor network). France 2030’s design implicitly recognises this through consortium requirements in major competitions — but the implementation often defaults to large-group-led consortia that include research organisations as subcontractors rather than genuine innovation partnerships.
The Bottom Line
By project count, France 2030 serves startups well — the deeptech pipeline is larger, better-funded, and more globally recognised than at any previous point in French innovation history. By capital value, large groups dominate France 2030’s balance sheet — as they must, given the capital intensity of gigafactories, semiconductor fabs, and bioproduction facilities.
The distribution is defensible but not optimal. The most effective use of France 2030 capital would shift somewhat toward: earlier-stage startup investment where public capital is most catalytic; simpler, faster competition instruments for small companies; and performance-conditioned grants that incentivize commercialization outcomes rather than project completion milestones. These shifts would not fundamentally alter the startup-incumbent distribution — the structural logic of capital-intensive sectors ensures incumbents will always dominate by value — but they would improve the marginal return on France 2030’s startup-targeted investment.
Key Data Points
- France 2030 project count distribution: 70% startups/SMEs, 15% ETIs, 10% large groups, 5% research
- France 2030 capital value distribution: 60-65% large groups, 15% ETIs, 15-18% SMEs/startups
- Application cost for major I-Démo competition: typically 2-4 weeks of senior team time, 100+ page application
- Startup deeptech failure rate: 70-80% fail to achieve commercial viability — high but consistent with expected innovation economics
- Verkor founding to gigafactory groundbreaking: under 3 years (2020-2023)
- Mistral AI founding to unicorn valuation: under 8 months (April 2023 - January 2024)
- Research organization France 2030 participation: ANR-managed competitions plus industrial consortium roles across all sectors
- France deeptech startup count: 500+ active deeptech companies (TRL 2-7) in France, Europe’s highest density per capita