You are the IT lead at a German Stadt or Landkreis whose digital-sovereignty roadmap includes changing the default search engine on managed devices. The European Parliament has just made Qwant the default on Microsoft Edge and Mozilla Firefox across 720 MEP offices and several thousand staff devices — the change went live on Thursday, 4 June 2026. Your CIO has forwarded the Politico story with two words: we too?

The answer depends on which layer of search-engine sovereignty you are buying. The Parliament’s announcement, the trade-press coverage and most public commentary have treated the change as a single decision. It is not. It is two decisions stacked on top of each other, and only the visible one is what the Parliament has actually made.

Qwant runs a frontend (the search interface) and depends on a backend (the search index). The Parliament’s decision changes the frontend default for its 720 MEPs. The backend question — where the index that produces the search results actually comes from — is partly answered by Qwant’s own crawler and partly answered by a long-standing Bing API contract that Microsoft can throttle, raise prices on, or terminate. A frontend default-switch is a sovereignty improvement at one layer and not at the other. Your decision about following the Parliament should specify which.

This article is the audit of what the default-switch actually changes, what the EUSP joint venture between Qwant and Ecosia will change in the next two years, and the three-question test your IT office should run on every default-tool decision before citing the Parliament’s example.

What the default-switch actually changes

The change applies automatically to default search settings in Edge and Firefox on Parliament-administered devices. Users can still manually select any other search engine, including Google. There is no policy prohibition on Google use. The change is administrative and reversible at the user level.

Defaults have measurable behavioural effects. The literature on choice architecture documents that defaults drive most usage even when changing the default is trivial. Making Qwant the default will materially shift Parliament search traffic to a European provider, even though Google remains accessible. Symbolic acts in legislative bodies have downstream effects: a Parliament that has formally moved its own infrastructure off Google is in a stronger rhetorical position to legislate about other institutions’ infrastructure choices.

Qwant is a meaningfully different ownership and operational structure than Google. Synfonium, Qwant’s owner, is French. The operational stack is in EU jurisdiction. The privacy commitments are stronger than Google’s commercial defaults. These differences are real, not cosmetic. They produce sovereignty improvement at the frontend layer — the layer where data leaves the user’s device and reaches the search provider’s logs.

That is what the Parliament has bought. The backend layer is a different audit.

The Bing dependency in the index

Qwant runs its own crawler — by the company’s own description, the index contains around 20 billion pages and adds roughly one billion crawler updates per day. It also uses Bing to supplement its own results, particularly for long-tail queries and image search. The arrangement has been in place since 2016, deepened in 2019, and survived a crisis in 2023 when Microsoft sharply raised Bing API pricing and Qwant restructured its business model around it.

The arrangement has consequences. If Microsoft were to terminate the Bing API agreement — under US sanctions pressure, on commercial grounds, in response to a US Congressional subpoena, or for any of the reasons documented in Microsoft’s 2025–2026 compliance pattern — Qwant’s search quality would degrade in the long-tail segment. Microsoft sees the aggregated query patterns from Qwant traffic, even if it does not see user-identifying metadata. Qwant’s privacy claims hold at the frontend layer; the crawler-and-index sovereignty claim depends on partner relationships at the backend.

Qwant and Ecosia have responded. In November 2024, the two companies announced the European Search Perspective (EUSP), a 50–50 joint venture to build an independent European search index at scale. Qwant transferred its existing indexing infrastructure and a portion of its engineers and data scientists to EUSP. The full migration of long-tail queries from Bing to EUSP is in progress and not yet complete. The Parliament’s announcement, on the day it was made, was sovereign at the frontend, mixed at the backend, and trending toward sovereign at the backend on a multi-year horizon.

None of this is in the Parliament’s communication. The institutional framing is “Qwant is a privacy-focused European search engine.” That is true. It is not the whole architecture.

The three-question test for your default-tool decisions

The Parliament’s decision is an instructive case study for IT offices considering similar default-switches. Every default-tool decision your office makes should pass three architectural questions before being cited as a sovereignty improvement.

One: which layer does the default actually affect? A search engine has a frontend (user interface, where queries enter), a query-routing layer (where the frontend decides which backend to ask), and a backend (the index that produces results). An office-suite default has a UI layer, a sync layer and a storage layer. An email-client default has a UI, a transport layer (SMTP/IMAP/Microsoft Graph) and a hosting layer. Naming which layer the default actually changes makes the sovereignty claim defensible. Failing to name it makes the claim performative.

Two: what is the underlying dependency at the layer below? For Qwant, the question is Bing. For Nextcloud, the question is whether the hosting provider is European-headquartered. For LibreOffice deployed in your office, the question is the source-distribution endpoint of the upstream packages — most of which are on GitHub. Naming the underlying dependency tells you what would still be at risk if the visible vendor were perfect.

Three: is the underlying dependency trending sovereign, neutral or worsening? EUSP is the trending-sovereign answer for Qwant — a two-year programme to migrate the Bing-dependent long-tail queries to an EU-controlled index. For Nextcloud, the trending-sovereign answer depends on whether your hosting provider is committing to EU-resident operational control. For LibreOffice, the answer depends on whether the project commits to a primary source-distribution endpoint outside US jurisdiction. A trending-sovereign answer is a defensible decision today. A trending-worsening answer is a decision that will be cited against you in 2028.

These three questions, run on the Parliament’s decision, produce a clear answer: frontend sovereign, backend partially Bing-dependent, trending sovereign via EUSP on a 2027–2028 horizon. That is a defensible sovereignty improvement, named carefully. Your IT office can make the same change for the same reasons, with the same architectural caveat in writing.

What this article is not

It is not a claim that the Parliament’s decision is empty. Defaults matter, and a default change at scale is non-trivial. It is not a claim that Qwant is the wrong choice — within the privacy threat model, it is a meaningfully better default than Google. Within the European industrial-policy frame, the EUSP joint venture with Ecosia is exactly the kind of multi-year work that could close the backend gap. It is not a claim that the Parliament has gone sovereign — the Parliament has changed a default; the backend dependency on Microsoft is incomplete, in motion, and not yet resolved.

What to write in your tender

If your IT office writes a default-tool decision into a procurement file citing the Parliament’s example, the file should specify which layer the change addresses, name the underlying dependency at the layer below, and document whether that dependency is trending sovereign. A three-paragraph note covering these three points reads as serious to any auditor. A one-line “we are following the Parliament’s example” reads as performative.

The signal worth watching, twelve months from now, is the EUSP rollout. If the Ecosia–Qwant joint venture delivers a European long-tail index in 2026–2027, the Bing dependency closes meaningfully and Parliament’s default-switch becomes sovereign at both layers. If EUSP slips into 2028 or beyond, the gap persists, and the Parliament’s institutional adoption of Qwant continues to rest on infrastructure it does not fully control — a state worth knowing about before your CIO commits to we too.

Sources

Topic overview: Digital Sovereignty in Europe Related articles: CADA before it lands, Sovereignty on Microsoft’s servers