Gemini and Google DeepMind: The Full…

Gemini today: 900 million users and the fastest growth in AI history

By May 2026, Gemini's share of the global AI-assistant market reached 27.7%, up from 18.2% in December 2024, while its web-traffic share grew from 5.4% in January 2025 to 27.4% by April 2026 — the fastest sustained market-share gain of any major AI platform on record. At Google I/O 2026, the company announced 900 million monthly app users, First Page Sage notes.

One major driver of that growth: Gemini isn't confined to a standalone app. It's built directly into Google Search results, Gmail, Google Docs, and Android itself — meaning millions of people use the model every day without ever opening a dedicated chatbot app, and often without realizing they're interacting with Gemini at all.

Born from two labs and one disastrous launch

Gemini grew out of the merger of two separate research groups: Google Brain (founded in 2011) and DeepMind (founded in London in 2010, acquired by Google in 2014 for $400 million). In April 2023, the two were combined into a single lab, Google DeepMind.

But before that, on February 6, 2023, Google unveiled a chatbot called Bard, built on its LaMDA model — and in the very promo video, the bot incorrectly claimed the James Webb Space Telescope took the first image of a planet outside our solar system (it was actually the European Southern Observatory). Alphabet's shares fell 7-8%, wiping out more than $100 billion in market value in a single day, NPR reports.

Gemini itself was unveiled at Google I/O on May 10, 2023, and Gemini 1.0 launched that December — the first model trained natively multimodal from scratch, learning text, images, audio, video, and code together rather than separately, unlike GPT-4 at the time. Bard was formally rebranded as Gemini in February 2024.

The 2026 product: Gemini 3.1 Pro, Deep Think, and video generation

The 2026 flagship is Gemini 3.1 Pro, released February 19, with a 1-million-token context window and up to 65,000 tokens of output, ranking #1 on 12 of 18 tracked benchmarks. There's also a faster Gemini 3.5 Flash model, a Gemini 3 Deep Think mode for harder problems, and Gemini Omni, which generates and edits video from any combination of image, audio, video, and text inputs.

The difference between regular Gemini 3.1 Pro and Deep Think mode comes down to how much "thinking time" the model spends before answering. The standard mode responds quickly and suits everyday queries, while Deep Think deliberately slows down, working through several reasoning steps before producing a final answer — built specifically for complex, multi-step problems like scientific calculations or parsing large sets of documents, not simple questions.

Google DeepMind co-founder Demis Hassabis announced that Gemini would support MCP (Model Context Protocol), calling it "a good protocol" that's "rapidly becoming an open standard for the AI agentic era." And at the I/O 2026 keynote, Google CEO Sundar Pichai said: "AI is the most profound way to advance our mission and improve people's lives at scale," according to Google's own blog.

The 2024 controversy: overcorrecting for bias

In February 2024, Google paused Gemini's ability to generate images of people after a wave of criticism: the model produced historically inaccurate pictures — Black Vikings, an Asian woman in a German WWII-era military uniform, a racially diverse group of America's Founding Fathers. Google said its tuning to show a range of people "failed to account for cases that should clearly not show a range," acknowledging the generator had become "way more cautious" than intended. Sundar Pichai personally apologized, Forbes reports.

The 2026 controversy: a court holds Google liable for its AI's answers

In July 2026, a Munich court ruled that Google can be held directly liable for false claims in AI Overviews — the Gemini-powered search feature that generates text summaries right in the results page. The court held that such summaries are Google's own words, not a protected traditional search result, and ordered the company to stop repeating specific false claims about two Munich publishers that the summaries had wrongly linked to scam schemes, Search Engine Land reports.

In parallel, in February 2026, media company Penske Media (owner of Rolling Stone, Variety, Billboard, and other outlets) presented court evidence that AI Overviews is "cannibalizing" publisher traffic — an Ahrefs study found that AI Overviews correlate with a 58% drop in click-throughs from search results. The European Publishers Council has filed a formal complaint with the European Commission.

Tellingly, both German legal actions concern not the Gemini chatbot itself, but its embedding inside ordinary search — meaning the issue touches millions of people who have no idea they're reading an AI-generated answer rather than a publication's own editorial text.

Bottom line: outgrowing ChatGPT while inheriting its trust problems

Gemini may have the strangest trajectory of any major AI product: it started with a public, $100 billion flop, then went on to post the fastest market-share growth in the industry's history — while walking the same path its rivals already have, from bias controversies to lawsuits over inaccurate answers. We've already mapped the full competitive picture, including Gemini, in our comparison of 2026's top five AI services, and you can read about Anthropic, OpenAI, and xAI in our separate profiles of Anthropic and Claude, OpenAI and ChatGPT, and Grok and xAI.

This article is for informational purposes only and does not constitute investment advice.