Capital & Industrial Strategy

Mistral AI has confirmed a €3 billion ($3.5 billion) Series D funding round led by Samsung Electronics, with participation from existing investors including ASML Holding and new undisclosed backers, according to Bloomberg and the Wall Street Journal. The valuation is reported as €21 billion by Bloomberg and above $24 billion by the WSJ — a discrepancy likely reflecting euro-dollar conversion timing and whether certain tranches are included. This is a closed deal with confirmed capital committed; the valuation range should be treated as a live reporting disagreement rather than ambiguity about the transaction itself.

Samsung's strategic rationale is clear: the Korean conglomerate needs a credible AI model partner outside the US hyperscaler ecosystem for its device and semiconductor businesses, and Mistral's open-weight positioning offers optionality that a dependency on OpenAI or Google does not. CEO Arthur Mensch has publicly argued that macroeconomic pressures will accelerate enterprise adoption of open-source models — a bet that positions Mistral not as a consumer AI brand but as infrastructure for sovereign and enterprise deployments. The proceeds are earmarked for model development and compute infrastructure, suggesting Mistral is moving from pure model provider toward vertical integration into its own training and inference stack, directly competing with the hyperscalers its customers are trying to avoid. The Financial Times frames this explicitly as Europe straining to keep pace, but the Samsung anchor also imports Asian industrial capital into European AI in a way that complicates the 'European AI sovereignty' narrative.

Anthropic has decided against acquiring Decart AI, a generative video and world-model startup reportedly valued at $6 billion, according to Bloomberg. No further terms or reasons were confirmed. The decision is notable for its timing: Anthropic's bankers are simultaneously pushing credit rating agencies for investment-grade designations ahead of an IPO, as reported by the Financial Times. These two data points in combination suggest Anthropic is managing its balance sheet conservatively ahead of public markets scrutiny — a $6 billion all-cash or equity deal would have complicated the clean capital structure narrative required for investment-grade debt.

The investment-grade push by both Anthropic and OpenAI is strategically significant beyond the IPO context. An IG rating would allow these labs to issue corporate bonds at rates comparable to large-cap tech, enabling them to finance data centre capacity and chip procurement on far more favourable terms than the venture-backed structures they currently rely on. This is the same financing arbitrage that allowed Amazon and Microsoft to scale cloud infrastructure in the 2010s. The labs are not just seeking public listings for liquidity — they are engineering access to the cheapest possible long-duration capital to fund what is structurally an infrastructure buildout business.

Two distinct infrastructure deals this week confirm that the AI compute buildout has materially shifted beyond US and European geographies. Nvidia-backed Firmus has signed a data centre capacity agreement with OpenAI in Malaysia, per Reuters via Google News, giving OpenAI sovereign-adjacent compute capacity in ASEAN at a moment when US export controls make direct chip sales to certain markets increasingly constrained. Separately, Semafor reports that Saudi Arabia's DataVolt is racing to build a $1 billion data centre at NEOM with an explicit strategy of exporting compute capacity globally — framing Saudi Arabia not as an AI consumer but as a compute exporter.

The strategic logic of both moves converges: nations and companies with access to energy, capital, and (via Nvidia partnerships) chips are positioning themselves as neutral compute providers to AI labs that need capacity outside politically sensitive US-controlled infrastructure. For labs like OpenAI, sovereign compute in Malaysia or the Gulf reduces dependency on a small number of US hyperscalers and provides geographic redundancy. For Nvidia, backing Firmus and analogous vehicles creates downstream demand for its chips in markets it cannot sell to directly.