But how to get to that European cloud?

Original Article ↗ Hacker News Discussion ↗

Calls to build a truly sovereign European cloud highlight tensions between digital autonomy and the dominance of US hyperscalers like AWS, Azure and Google Cloud. Commenters debate why previous initiatives such as Gaia‑X faltered, pointing to design‑by‑committee bureaucracy, weak incentives, comparatively low tech salaries, and fragmented capital markets, even as smaller providers like OVH and Hetzner already exist. Many argue that only a mix of targeted regulation (e.g., “EU‑law‑only” procurement rules), protectionist measures, and serious investment in engineering talent can create viable large‑scale alternatives, while others question whether centralised “Euroclouds” are even the right goal versus more decentralised, locally controlled infrastructure.

Geopolitics, sovereignty, and who should build it

  • Some argue Europe (and possibly allies like Australia) needs its own cloud for legal and strategic independence, especially post‑CLOUD Act and as US defense guarantees look shakier.
  • Others say Australia should orient toward Asia or stay neutral, and that a truly “sovereign cloud” owned by a US firm (e.g. AWS “EU sovereign cloud”) is contradictory unless tech and control are fully localized.
  • A minority questions whether “sovereign cloud” is even coherent: renting “other people’s computers” is inherently non‑sovereign, so owning hardware and decentralizing might be safer.

Industrial policy, protectionism, and R&D

  • Proposals include tariffs on foreign clouds, “China-style” requirements for local partners owning local tech, and tying public spending to EU providers.
  • Counterpoint: tariffs are distortive and similar effects could come from subsidies or R&D support, but switching from incumbent US clouds still needs strong policy push.
  • Several want a “European DARPA” focused on long‑term, high‑risk tech, arguing current EU programs (Horizon, Gaia‑X) are too bureaucratic and consultant‑driven.
  • Disagreement over whether EU regulators and political culture (risk‑averse, lawyer‑led, fragmented financial markets) are capable of designing the right incentives.

Gaia‑X and previous “sovereign cloud” efforts

  • Gaia‑X is widely cited as a failure: design‑by‑committee, money to bureaucrats and large incumbents, little going to engineers.
  • Some describe it as covert subsidy for the usual big industrial players, with poor technical outcomes and even weak hiring signals.
  • Example from France: “sovereign” clouds that turned out to be Huawei‑run, later shut down, leaving customers repeatedly forced to migrate.

Existing European providers and product gaps

  • OVH, Hetzner, Scaleway, Infomaniak, Outscale and telecom‑run datacenters are mentioned as real, working European infrastructure.
  • Critique: many are essentially VPS/“lumber” sellers, lacking polished, self‑service, integrated services (autoscaling, managed DBs, S3‑like storage, CI/CD‑friendly patterns) that make hyperscalers attractive.
  • Others argue government workloads are often small and don’t need hyperscaler‑level scale; but reliability and operational tooling (rollouts, HA) still benefit from “cloudy” features.

Market dynamics, regulation, and free‑market debate

  • One camp says “if there’s demand, the market will provide”; government should cut regulation and taxes.
  • Another camp says cloud is a sticky, scale‑and‑network‑effects market where free entry fails; active regulation is needed to create demand for “subject only to EU law” providers.
  • Broader dispute over tariffs vs subsidies, and whether cloud is (or tends toward) oligopoly‑style concentration that justifies antitrust and industrial policy.

Talent, pay, and capital

  • Many see low European tech pay (e.g. ~€70k) as a core blocker: hard to “bend over backwards” to build massive infrastructure when US remote roles offer far more.
  • There’s debate over how big the EU–US gap really is once taxes, social benefits, and COL are included, but consensus that median EU tech salaries are far below US hyperscaler levels.
  • Some advocate aggressively poaching engineers who built AWS/Azure/GCP, combined with tax incentives, high‑speed rail connectivity (London–Paris–Randstad), and cross‑border startup funding.
  • Others emphasize Europe’s problem is risk‑averse capital: lots of pensions and industrial giants, very little true high‑risk VC.

Government procurement and practical experience

  • Several describe disastrous experiences with big “framework” hosting contracts for EU governments: slow, incompetent providers chosen on price, lots of red tape, poor K8s/HA designs, and attempts by dev teams to escape to AWS/Azure/Hetzner.
  • Lesson drawn: a European cloud for the public sector only works if it matches hyperscalers’ self‑service and developer experience.

Alternatives to a single Euro‑hyperscaler

  • Some think chasing an AWS clone is “fighting the last battle”; better to:
    • Standardize around simpler clouds (Hetzner‑like),
    • Encourage many smaller “drones” (local or sectoral providers) instead of one “aircraft carrier,”
    • Or push on‑prem / microcloud solutions for resilience against cyber or connectivity disruptions.
  • Cloud itself is questioned: for many workloads, traditional hosting with good ops may beat complex cloud constructs; cloud’s main value is operational convenience and managed services, not theoretical elastic scaling.

Soft power, ecosystems, and Microsoft/AWS dominance

  • US clouds benefit from entrenched ecosystems: universities teaching proprietary stacks, free credits for students, consulting firms tied to Microsoft, and “never fired for choosing X” brand equity.
  • Some suggest universities should stop teaching vendor‑specific stacks (e.g. ASP.NET/Azure) to weaken this lock‑in and create space for neutral or European alternatives.