There are far fewer global search engines than there are global search brands, and the gap between those two numbers is the whole subject of this page.
To qualify as a genuinely global general engine, a service has to do three hard things at once. It has to crawl the open web continuously in every major language, because an index that stops being refreshed is out of date within weeks. It has to store and serve that index at a scale measured in hundreds of billions of pages. And it has to rank it well enough that ordinary people keep using it, in markets where the incumbent is already installed as the default in the browser and on the phone. Each of those is a permanent operating cost rather than a one-off build. Very few organisations on earth have ever paid all three bills at the same time, and today, for a general index with worldwide coverage, the number is two: Google and Microsoft.
What the market measurement actually says
StatCounter Global Stats put Google at 91.31% of worldwide search across all devices in July 2026. That figure needs its methodology attached every time it is quoted. StatCounter counts page views on websites carrying its tracking code and attributes each referral to whichever engine sent it. It is a proxy, not a census; it cannot see searches that never produce an outbound click, and its panel is thin in some countries. It is nevertheless the standard public number, and no serious measurement disagrees about the shape of the picture: one engine holds the overwhelming majority of general web search outside China, Russia, South Korea and a small number of other markets.
The second position is more interesting than the percentage suggests, because Bing's importance is not its direct audience. Microsoft's crawler, bingbot, feeds the only other general web index with worldwide coverage, and Microsoft sells access to it. That makes Bing the wholesale supplier underneath a large part of what people think of as "alternative" search.
The supplier relationships underneath
Yahoo is the clearest illustration and the reason it sits in this category at all. Yahoo Search began in 1994 as the search facility over a hand-built human directory; it did not operate its own web crawler until 2004, when it assembled one from the acquisitions of Inktomi, AltaVista and AllTheWeb. It then gave that up. The Microsoft–Yahoo Search Alliance was announced on 29 July 2009, migration ran through the early 2010s, and Yahoo has not run a general web crawler since. The deal was renegotiated in April 2015 so that Bing was required on only the majority of Yahoo's desktop traffic, and between October 2015 and 2018 Yahoo ran a separate agreement with Google for some results — which is why "Yahoo is powered by Bing" was, for a period, only partly true. That Google agreement has expired.
Yahoo has not stopped investing in search; it has stopped indexing. Its AI answer engine Yahoo Scout launched on 27 January 2026, and Yahoo's description of it is unusually candid about the division of labour: the language model is Anthropic's Claude, the web grounding comes from Microsoft's Bing grounding API, and Yahoo supplies the ranking, the interface and its own consumer data. As Yahoo put it in January 2026, "the underlining search index is from Bing, but the responses, ranking, and experience is all Yahoo." That sentence is a good summary of what a major search brand can be in 2026 without owning an index.
The same dependency runs well beyond Yahoo. Bing has supplied web results to DuckDuckGo since roughly the start of the 2010s, and has at various times supplied Ecosia, Qwant and Swisscows. When a Bing outage occurred in 2024, several of those engines stopped returning web results at the same moment — the clearest public demonstration available of where the links actually come from.
Why the supplier layer changed in 2025
Microsoft retired all of its public Bing Search APIs on 11 August 2025, decommissioning existing instances and closing new signups, and pointed customers instead at "Grounding with Bing Search" inside its Azure AI products. This matters more than a product-line tidy-up. The public API had been the cheap, self-service route by which small engines, research tools and AI products obtained web results. Its replacement is a grounding service designed to feed a language model, not a raw results interface. Longstanding direct syndication partners with their own contracts — DuckDuckGo among them — continued to return results afterwards, but the terms of those private contracts are not published by either side. The practical consequence is that any claim that an engine "uses Bing" after August 2025 should be read with a follow-up question: under what arrangement, and for how long?
What "global" excludes, and why
Several engines with their own genuine indexes are deliberately not in this category. Yandex and Baidu run large independent crawls but are overwhelmingly concentrated in one language market each. Naver, Seznam and Cốc Cốc are national engines by design and by business model. Brave Search, Mojeek and Marginalia crawl the open web independently and globally, but at index sizes one to two orders of magnitude below Google's, which is a real limit on coverage rather than a quibble — Mojeek's own published milestones reached about nine billion pages in 2025, against a Google index generally understood to be in the hundreds of billions. Coverage of that kind is enough for a great many queries and visibly insufficient for others, and the honest way to describe it is with the number attached.
What follows below, then, is not a ranking. It is three brands sitting on two indexes: one company that built the dominant index of the open web, one company that built the only other index with comparable global reach and turned it into a wholesale business, and one company that once had an index, sold the idea of search without it, and is now running an AI answer engine on top of a competitor's grounding service.