• The GPU
  • Posts
  • Issue #17: AMD, G42, DataOne, Fluidstack, & Sesterce Bet Big in France

Issue #17: AMD, G42, DataOne, Fluidstack, & Sesterce Bet Big in France

Feat. France, Crusoe, AMD, G42, DataOne, HydraHost, Fluidstack, Sesterce, Mistral, Positron, Nvidia, EDF, and Tencent

AI literally went nuclear this week.

Paris hosted world leaders and tech CEOs debating AI regulation, but the real action was in AI compute investments. France locked in major AI data centre deals, with Fluidstack, Sesterce, AMD and the UAE's G42 pouring billions into nuclear-powered AI hubs. Meanwhile, Mistral doubled down on enterprise AI, Positron raised $23.5M to take on Nvidia, and Tencent Cloud planted its flag in Saudi Arabia as Chinese cloud influence expands in the Middle East.

I’m Ben Baldieri, and every week I break down the moves shaping GPU compute, AI infrastructure, and the data centres that power it all.

Here’s what’s inside this week:

  • Paris AI Summit: Policy debate or just a power grab?

  • Investments ramp up with Crusoe, AMD, G42, and HydraHost.

  • Fluidstack & Sesterce bet big on France.

  • Mistral overtakes ChatGPT and starts playing the enterprise game.

  • Positron lands $23.5M to challenge Nvidia with energy-efficient AI chips.

  • France’s EDF unlocks 2GW of nuclear-backed AI compute.

  • Tencent Cloud moves into the Middle East at LEAP.

Let’s get into it.

The GPU Audio Companion Issue #17

Want the GPU breakdown without the reading? The Audio Companion does it for you—but only if you’re subscribed. Fix that here.

Paris AI Summit: Global Policy or Power Play?

World leaders and tech CEOs gathered in Paris this week to discuss AI’s future.

The tone at the AI Action Summit was geopolitically mixed. With Trump gutting AI regulations in the US, pressure is mounting on Europe to ease restrictions or risk losing its AI players to more lenient markets. JD Vance, US Vice President, pulled no punches in his speech (above). This comes off the back of China’s DeepSeek shaking the AI world with open-weight models, forcing the West to rethink its approach to AI openness, and the entire industry to react.

Why this matters:

  • The US, China, and Europe are taking wildly different AI approaches, splintering global AI policy.

  • The deals announced at the summit clearly demonstrate that France is Europe’s AI hub, outpacing Germany and the UK.

  • Companies are pushing back against heavy AI regulation, warning that overreach could kill European competitiveness.

Investments Ramp Up with Crusoe, AMD, G42, and HydraHost

Major players from top to bottom of the AI stack are making big bets on infrastructure.

Crusoe is setting up a low-emission AI compute hub in Ireland, adding 100 jobs and using stranded energy for GPU workloads. AMD & G42 are partnering to build AI infrastructure in France with DataOne, accelerating AI model development in a market increasingly dominated by Nvidia-powered data centres. And Hydra Host closed a Seed+ funding round led by Flume Ventures.

Why this matters:

  • Crusoe are taking their stranded-energy-use model global.

  • G42 are a serious player. They’re already working with Microsoft across a number of projects, so this partnership could mean AMD is finally starting to wake up and give a meaningful challenge to Nvidia.

  • A funding round for Hydra Host in the crowded brokerage/marketplace GPUaaS sub-vertical is a pretty strong vote of confidence in their business model.

Read the full stories below:

Fluidstack & Sesterce Bet Big On France

Geopolitics and regulation weren’t the only things on the agenda in Paris this week.

Fluidstack is building a 1GW AI supercomputer in France, powered by low-carbon energy to fuel next-gen AI workloads. Phase 1 will host close to 500,000 next-generation GPUs. Future phases plan to scale beyond the initial 1GW of dedicated compute by 2028.

Meanwhile, Sesterce is committing to launch 1.5GW of capacity across multiple AI data centres with a €52B investment. The first phase of the project will be focused in France near Valencia with a proposed 40,000 GPU deployment and an investment of €1.8B.

Why this matters:

  • France is clearly becoming an attractive destination outside of Scandinavia for neoclouds.

  • Stable and competitive power pricing is essential for GPU providers, given the amounts of power large-scale cluster deployments require.

  • With Fluidstack and Sesterce joining Nscale in large-scale deployment plans, a European counterweight to American neocloud dominance from CoreWeave, Crusoe, and Lambda et al. could be emerging.

Read the full stories below:

Mistral’s AI Assistant Tops iOS & Goes All-In on Enterprise

Mistral just became the No.1 AI app on iOS. But that’s just the headline.

Behind the scenes, Mistral is shifting from pure research to business execution. It launched Le Chat, an AI assistant for both consumers and enterprises, and secured major deals with France’s national employment agency, Stellantis, and Veolia. Even French President Emmanuel Macron promotes Mistral over ChatGPT, calling it a "French champion."

The bigger move?

Mistral is building its own AI data centre in France to train next-gen models. With OpenAI and Anthropic raising billions, Mistral needs scale, and it’s betting that European AI sovereignty will bring in government and corporate customers.

Why this matters:

  • Mistral is positioning itself as Europe’s AI leader.

  • Enterprise AI is the real battleground, and Mistral is already locking in strategic deals.

  • Owning its own compute gives Mistral independence, reducing reliance on US infrastructure.

Positron Raises $23.5M to Take on Nvidia

Positron just secured $23.5M to scale its US-manufactured AI chips.

Led by former Lambda executive Mitesh Agrawal, their hardware offers 3.5x better performance per dollar and 70% faster inference than Nvidia’s H100. Positron’s FPGA-powered architecture hits >93% memory bandwidth utilisation, slashing power consumption and cutting data centre CapEx by 50%. The company is already shipping plug-and-play AI accelerators to NeoClouds and data centres, with next-gen chips launching in 2026.

And with the Nvidia Rubin Ultra’s supposedly hitting 819kW peak rack density, the appetite for lower-powered hardware will likely grow.

Why this matters:

  • Very few data centres can support such high rack densities.

  • Power-efficient AI chips will likely become increasingly important as inference workloads scale.

  • Positron’s US supply chain offers a domestic alternative to foreign chip reliance.

France Unlocks 2GW of Nuclear-Powered AI Compute

France continues rolling out the red carpet for nuclear-powered AI data centres.

EDF is offering 2GW of pre-connected, ready-to-build land to fast-track AI infrastructure, cutting deployment times by years. The move ties into Project Giga, EDF’s initiative to supply AI data centres with dedicated energy infrastructure. Microsoft and Google are already circling, and with AI power demand exploding, France is positioning itself as Europe’s top AI hub.

Why this matters:

  • Energy-efficient AI hubs are becoming a competitive advantage, with power constraints limiting expansion elsewhere.

  • EDF’s direct involvement means energy suppliers are now shaping AI infrastructure decisions, not just data centre operators.

  • AI clusters near nuclear plants could create a new blueprint for large-scale AI deployments, reducing transmission losses and stabilising costs.

Tencent Cloud Expands into Saudi Arabia

Tencent Cloud just launched its first Middle East cloud region in Saudi Arabia, with two availability zones and a $150M investment to fuel the Kingdom’s digital ambitions.

Announced at LEAP 2025, the new region will support AI, cloud gaming, e-commerce, and financial services, strengthening Saudi Arabia’s push to become a global tech hub. But with China’s deepening ties in the region and US restrictions on AI chip exports, Tencent’s move also raises questions about the emerging geopolitical fault lines in both AI infrastructure and the GCC.

Why this matters:

  • Saudi Arabia is fast becoming a global AI and cloud hotspot, drawing hyperscaler investment.

  • Tencent’s expansion signals growing Chinese cloud influence, challenging US tech dominance.

  • Middle East AI infrastructure is increasingly caught in US-China tech tensions.

The Rundown

AI went nuclear this week.

Paris hosted world leaders debating AI policy while the real players were locking down AI infrastructure. Crusoe is expanding in Ireland, AMD and G42 are building in France, and HydraHost just raised fresh capital to scale its GPU cloud business. Fluidstack and Sesterce committed Billions to French AI data centres. Mistral overtook ChatGPT on iOS, went all-in on enterprise AI, and even got a Macron endorsement. Positron raised $23.5M to challenge Nvidia’s power-hungry chips, just as EDF unlocked 2GW of nuclear-backed development land.

And in Saudi Arabia? Tencent Cloud landed at LEAP 2025, marking China’s latest play in the AI cloud wars.

Tl;dr for this week? Less regulation talk, more AI land grabs, and a whole lot more money moving into compute.

Keep The GPU Sharp and Independent

Good analysis isn’t free. And bad analysis? That’s usually paid for.

I want The GPU to stay sharp, independent, and free from corporate fluff. That means digging deeper, asking harder questions, and breaking down the world of GPU compute without a filter.

If you’ve found value in The GPU, consider upgrading to a paid subscription or supporting it below:

☕ Buy Me A Coffee → https://buymeacoffee.com/bbaldieri

It helps keep this newsletter unfiltered and worth reading.

Reply

or to participate.