Seventy-five million euros of public money, doubled by the same amount poured into a direct competitor, only to end ten years later with two empty shells sold for next to nothing. This is the blunt summary of the Andromède project, launched in 2011 under Nicolas Sarkozy to equip France with a sovereign cloud in the face of Amazon, Google, and Microsoft. The original fault? A disagreement among French industrial players that the State resolved by funding two rival projects instead of one champion, dooming the ambition before its first server was even switched on.
Key takeaways
- Why did the State fund two rival clouds instead of backing a single champion?
- How did 150 million euros disappear in seven years without a trace?
- Fourteen years on, Orange returns to the heart of the new French sovereign bet: a tale of starting over?
A single project that becomes a two-front war
At the outset, Andromède was to be a joint venture. Orange and Thales on one side, Bull and SFR on the other, were set to embark on the French cloud adventure. But industrial egos and commercial rivalry among telecom operators undermined the façade of unity. From the start, the chosen French tech groups could not agree; Orange and SFR refused to walk hand in hand. As a result: rather than arbitrate and appoint a single champion, the State chose the most expensive and least rational path. The project split into two, and with it the 150 million euros of aid: Orange teamed with Thales to create Cloudwatt, and SFR formed a partnership with Bull to create Numergy.
The financial wiring of the two structures was nearly a perfect copy. The companies shared similar frameworks: a major- shareholder operator, Orange with 44.4% of Cloudwatt and SFR with 47% of Numergy, the Caisse des Dépôts as the second shareholder at 33.3% and 33%, and a security specialist, Thales for Cloudwatt and Bull for Numergy. Two almost identical organizational charts, financed by the same taxpayer, to sell exactly the same type of service to the same potential customers. An economic absurdity that even some experts had anticipated from the outset, denouncing “unfair competition,” a project deemed “without a future,” a “waste of public money.”
Two years to lose everything
The market spoke quickly and decisively. After two years in operation, the two services’ records were catastrophic: despite public and private money invested, neither cloud found clients, with revenue peaking at 6 million euros for Numergy and 2 million euros for Cloudwatt. These figures were meager next to the initial ambitions: Cloudwatt had targeted 500 million euros in revenue by 2017, and Numergy 400 million. The gap between promise and reality was dizzying.
Facing this fiasco, the State folded in 2015. The State was compelled to withdraw, forcing Orange to buy back 100% of Cloudwatt and SFR to buy back Numergy. The exit happened discreetly and at a loss: in 2015, Numergy was placed under safeguard proceedings and bought back 100% by SFR, while Cloudwatt was folded into Orange Business Services. Numergy disappeared entirely two years later, the platform vanishing during SFR’s cloud activities overhaul in 2017. Orange did attempt a revival of Cloudwatt in 2018, but the reprieve was short: in August 2019 Orange announced Cloudwatt would permanently shut its doors on February 1, 2020. Seven years after its grand launch, France’s sovereign cloud no longer exists. Not a single client, not a single infrastructure, not a tangible technological heritage. Just a line in a Court of Audit report and 75 million euros gone.
The sovereignty mirage, ten years on
The paradox is painful: never has the issue of sovereign data hosting been so urgent as it is today, at a moment when generative artificial intelligence swallows vast volumes of sensitive data like never before. And France is almost back to square one. The current landscape has rearranged itself around three families of actors. On one side, real French and independent companies, qualified as SecNumCloud by the ANSSI: by March 2026, three major players hold SecNumCloud 3.2 status, namely OVHcloud, Outscale (Dassault Systèmes), and NumSpot. On the other side, hybrid “clouds of trust” offerings, which rely on the technology of American giants under French operation: offerings such as Bleu (Orange and Capgemini, based on Microsoft Azure) and S3NS (Thales and Google Cloud) provide hyperscaler technologies operated under French law.
This compromises model fuels a debate nearly as lively as Andromède’s was in its day, about what sovereignty really means. The digital sovereignty debate asks whether the SecNumCloud qualification granted by the ANSSI to S3NS is not excessive, given the practical dependence on Google or Microsoft remains intact. Meanwhile, public money continues to flow toward these solutions: purchases of SecNumCloud cloud services from European suppliers reached 22 million euros in 2025, up from 20 million euros in 2024. These amounts are not on par with the 75 million swallowed by Cloudwatt alone, but they nonetheless highlight the slow restart of sovereign cloud in France.
The most telling detail perhaps lies in a name that has not changed: Orange, the majority shareholder of Cloudwatt in 2012, is back at the heart of France’s new sovereignty bet today, via Bleu, alongside Capgemini. The same actor, the same wager, fourteen years apart. France has never lacked ambition for cloud computing. It has merely failed to pick a side, once and for all, before signing the paycheck.
Sources : clubic.com | senat.fr
