Google started with search in 1998 and restructured into the holding company Alphabet in 2015. The Search, YouTube, Android, and Gmail you use every day mostly come from here. In the AI race, it has one trait no rival shares: as of October 2026, it is the only publicly listed company among the AI big three. OpenAI and Anthropic are not listed yet, so Google’s audited financials are open for everyone to read.
Its playbook is “own everything.” For models it has Gemini. For chips it has the in-house TPU (Google’s own AI accelerator). For distribution it has Search, Android, and YouTube, which reach billions of people. For the cloud it has Google Cloud. Rivals rent compute wherever they can; Google covers nearly the whole chain itself, from chip to channel.
Core data snapshot
| Item | Data |
|---|---|
| Founded | Google 1998; Alphabet restructured 2015 |
| Company type | Public company (Nasdaq: GOOG/GOOGL); headquartered in Mountain View, California, USA |
| 2025 total revenue | About $402.8 billion (up about 15% year over year) |
| Google Cloud (Q2 2026) | Quarterly revenue $24.8 billion, up 82% year over year; operating margin 35.6%; signed contracts not yet recognized as revenue (backlog) of $514 billion |
| AI product usage | Gemini App over 1 billion monthly actives (August 2026); Search AI Overviews over 2.5 billion monthly actives (I/O 2026); AI Mode over 1 billion monthly actives (Q2 2026) |
| 2026 CapEx guidance | $195–205 billion (raised in July 2026) |
| Headcount | About 190,820 (as of end of 2025) |
How to read these numbers. The table comes from Google’s formal financial reports and official announcements, which are more reliable than figures from a private startup. But Google does not report an “AI business” on its own. AI’s contribution is buried inside Search and Cloud and can only be estimated from the totals.
A quick tour of seven dimensions
Each dimension links to a deeper standalone piece.
① Technology and product roadmap: The main model line is the Gemini family, with Flash, Pro, Flash-Lite, and others each playing a role. At the end of September 2026 it announced its next-generation frontier model, Gemini 4 Argon, opened first only to a small number of cyber-defense organizations. Underneath sits the in-house TPU; the eighth generation is split into the TPU 8t for training and the TPU 8i for inference. For how to choose within the family, see How to choose across the whole Gemini family.
② Customer base and market positioning: On the consumer side it has Search, YouTube, and Android, which reach billions of people. On the enterprise side it sells Gemini and compute through Google Cloud, and since Q2 2026 it has also installed TPU systems directly in some customers’ own data centers. People often ask whether Gemini or ChatGPT has more users, but that question hides a trap.
③ Ecosystem and partnership strategy: Gemini is built directly into Search, Android, and Workspace, a distribution advantage others find hard to copy. While building its own models, it also invests in its rival Anthropic: up to $40 billion, according to multiple media reports and confirmed by both sides, plus TPU compute.
④ Valuation and financial model: As a public company, it has a market cap, a P/E ratio, and financial reports to study. Google Cloud is its most watched growth engine in recent years. But its AI capital expenditure is enormous, and free cash flow in Q2 2026 was negative $5.9 billion. Whether that money brings matching returns is a bet the market keeps debating.
⑤ Commercialization risk and regulation: The biggest risk sits on its home turf: whether AI search dilutes how much money search ads make. On regulation, Google has appealed the search antitrust case, and oral arguments have not been scheduled. In the ad tech case, the judge in September 2026 did not order a breakup of its ad exchange AdX and imposed behavioral restrictions instead; the final judgment is not settled yet. The EU fined it €890 million under the Digital Markets Act (DMA) in July 2026.
⑥ Geopolitics and supply chain: In-house TPUs let Google rely on Nvidia less than its peers, but TPUs still depend on TSMC’s advanced processes, CoWoS (advanced packaging), and HBM (high-bandwidth memory). According to CNBC, starting in October 2026 it is also renting about 110,000 Nvidia GPUs from SpaceX as bridge compute. How much TPUs actually save it and its data-center energy plans are long-term variables. For how the whole chain works, see The AI hardware supply chain, end to end.
⑦ Leadership and governance: Sundar Pichai is CEO of both Google and Alphabet. After the August 2026 reshuffle, Demis Hassabis became chairman of Google DeepMind and chief scientist of Alphabet, and Koray Kavukcuoglu took over the Gemini models and app. Founders Page and Brin have no day-to-day roles, yet hold a majority of the voting power through Class B shares. For details, see Who runs Google’s AI. The co-founder memo that reportedly leaked in April 2026, admitting Gemini was behind, gave a rare look at this layer of power.
Key milestones
| Time | Milestone |
|---|---|
| 1998 | Google is founded and launches its search engine |
| 2014 | Acquires the UK AI lab DeepMind, which becomes its AI research core |
| 2015 | Restructures into the holding parent company Alphabet |
| 2023 | Launches the Gemini model, going all in on generative AI |
| 2024 | A U.S. federal court rules that Google holds an illegal monopoly in search |
| 2025 | The search case remedies ruling comes out |
| H1 2026 | Appeals the search case ruling; announces an investment of up to $40 billion in Anthropic; unveils Gemini 3.5 Flash and Omni at I/O 2026 |
| H2 2026 | Raises CapEx guidance to $195–205 billion; EU DMA fine of €890 million; AI leadership reshuffle; Gemini App passes 1 billion monthly actives; no breakup ordered in the ad tech case; limited release of Gemini 4 Argon |
This table is current as of October 2026.
Penchan’s summary
Google is the most “all-in-one” of the AI giants: chips, models, cloud, and channels that reach billions of people are all in its own hands, and its financials are the most transparent.
Two question marks to watch: whether AI search, by giving answers directly, eats into its most profitable business, search ads; and whether its huge capital expenditure brings matching returns. How the antitrust cases end will decide how fully it can use these advantages.
Further reading and the standalone pieces to come
- How to choose across the whole Gemini family: Flash, Pro, Omni, Flash-Lite, what’s the difference
- Does Gemini or ChatGPT have more users? Why you can’t compare MAU head to head
- Why Google is putting up to $40 billion into its rival Anthropic
- Where Google’s AI capital expenditure goes, and what the market is worried about
- Where Google’s antitrust cases stand: the real impact on Gemini and search
- In-house TPUs: how deep is Google’s chip-cost moat
- Who runs Google’s AI: Hassabis, Pichai, and the two founders
- Even Google itself says it’s behind Claude: the co-founder’s leaked memo
- The full deep-dive report (2026 Q1 edition)