Google’s AI comeback: Gemini 3 wakes the sleeping giant

Lead

As of late November 2025, Alphabet’s Google has staged a marked comeback in the global artificial‑intelligence race after unveiling Gemini 3 and making strategic chip and cloud deals, including a major tie‑up with Anthropic. Investors rewarded the moves: Alphabet’s market value has climbed nearly $1 trillion since mid‑October and shares jumped as much as 3.22% in New York on Tuesday, putting the company on track for a $4 trillion valuation. Gemini 3 drew immediate praise for stronger reasoning, coding and niche‑task performance, and reports that Meta is in talks to use Google’s TPUs helped lift sentiment. Together with steady cloud growth and renewed demand for Google’s custom chips, the company’s position looks materially strengthened versus earlier doubts that it had fallen behind OpenAI.

Key takeaways

• Gemini 3: Google’s newest multi‑purpose model has been lauded for advanced reasoning and coding capabilities, improving on prior model shortcomings in image and overlay text tasks.
• Market reaction: Alphabet’s market capitalization has increased by nearly $1 trillion since mid‑October; shares rose up to 3.22% on Tuesday and the company is on pace for a $4 trillion valuation.
• Chip momentum: Anthropic agreed to use as many as 1 million Google TPUs in a multi‑billion‑dollar deal; a report said Meta may use Google chips in 2027, boosting demand signals.
• Cloud performance: Google Cloud posted $15.2 billion in third‑quarter revenue, up 34% year‑over‑year, though it remains behind Microsoft and AWS in absolute scale.
• User metrics: Google says 650 million people use the Gemini app; OpenAI has reported 800 million weekly users for ChatGPT; monthly app downloads in October were 73 million for Gemini and 93 million for ChatGPT (Sensor Tower).
• Market spillover: Nvidia shares tumbled as much as 5.51% on Tuesday, erasing roughly $243 billion in market value, while SoftBank fell to a two‑month low amid concern about competitive pressure.
• Strategic depth: Google controls a broad “full stack” — apps, models, cloud infrastructure and custom TPUs — and retains large proprietary data sources such as Search, Android and YouTube for model training.

Background

When ChatGPT emerged in late 2022 it shifted perceptions about the pace of consumer and enterprise adoption of large language models and highlighted gaps in incumbent firms’ commercial AI offerings. Several analysts and even some current and former Google staff publicly argued that Google had ceded ground in a market it had helped invent. That narrative pressured Google to consolidate and accelerate its AI work, centralizing leadership and refocusing resources on foundation models that can address both consumer and enterprise needs.

Google’s strengths are structural as well as financial. The company operates profitable advertising and services businesses that generate cash to fund expensive research, and it runs extensive global data centers and software platforms. Historically Google built its own tensor processing units (TPUs) for internal use; those chips have long accelerated Google’s internal AI workloads and are now being commercialized to select customers. The combination of proprietary data, engineering talent and cloud scale has long been the company’s defensive moat — one it is now leaning on more visibly.

Main event

In recent weeks Google released Gemini 3, a model positioned as a step change in reasoning and code generation. Industry leaderboards and several experts ranked Gemini 3 near the top of evaluated models, with specific praise for solving complex science and math problems and for improving accuracy on visual tasks. Google framed the release as part of a “full‑stack” AI strategy that spans apps, models, cloud services and silicon.

At the same time Google has turned its TPU business into a potential external revenue stream. Anthropic announced an agreement to use up to 1 million TPUs in a deal worth tens of billions, and reporting from The Information said Meta is in talks to use Google’s chips in its data centers beginning in 2027. Google declined to confirm specific customer plans but told investors its cloud business is accelerating demand for both its TPUs and Nvidia GPUs.

Market responses were immediate. Alphabet’s shares rose sharply, aided by broader enthusiasm that included Warren Buffett’s Berkshire Hathaway taking a roughly $4.9 billion stake in the third quarter. Conversely, Nvidia shares fell as investors weighed the prospect of alternative chip suppliers and renewed competition, wiping out about $243 billion in market value on the day of the report.

Operationally, Google has also shown incremental progress diversifying beyond search and ads. Waymo advanced its commercial driverless taxi service into additional cities and added freeway driving capability, demonstrating how AI investments can feed into distinct business lines. Still, Google’s cloud revenue of $15.2 billion in Q3 keeps it in third place behind Microsoft and AWS, underlining both growth and persistent gaps.

Analysis & implications

Strategically, Gemini 3 and commercial TPU deals change the market calculus: Google is no longer a passive supplier of models and infrastructure but an active platform contender. Having the “full stack” gives Google leverage to optimize performance end‑to‑end and to bundle services for enterprise customers, potentially lowering total cost and integration friction for buyers who choose Google Cloud.

However, the TPU advantage comes with trade‑offs. TPUs are accessed primarily through Google Cloud, which can lock customers into the Google ecosystem; by contrast, Nvidia GPUs are widely available across clouds and on‑premises, offering greater flexibility. That trade‑off means some large customers may adopt TPUs for price or performance but retain multi‑vendor strategies to avoid dependency and preserve bargaining power.

Regulatory risk — long a specter for Google — appears less immediate after recent legal developments in the U.S., where the most extreme breakup scenarios receded partly because lawmakers and courts weighed AI competition as a countervailing force. Nonetheless, antitrust scrutiny and privacy rules remain possible headwinds as Google expands AI integration across search, ads and other products.

Comparison & data

The table summarizes key public figures referenced in reporting. Market‑cap and share‑price moves are short‑term market reactions; user counts and downloads reflect app‑level metrics that do not fully capture active engagement or revenue. Cloud revenue shows growth momentum but also the gap to Microsoft and Amazon in absolute scale.

Reactions & quotes

Analysts and industry participants framed Google’s moves as confirmation that the company can convert research scale into commercially relevant products and hardware relationships. Below are representative remarks with context.

“A sleeping giant that is now fully awake.”

Neil Shah, Counterpoint Research (analyst)

This characterization was offered to capture Google’s long‑standing technical depth and the sudden, visible commercial traction after Gemini 3 and multiple chip and cloud signals. Shah’s view reflects market sentiment that Google’s investments are beginning to yield competitive advantages.

“We’ve taken a full, deep, full‑stack approach to AI.”

Sundar Pichai, CEO of Google and Alphabet (investor comment)

Pichai has repeatedly framed Google’s strategy around integrated stacks — models, infrastructure and services — arguing that vertical integration helps sustain competitiveness. Investors have treated the statement as a summary of Google’s rationale for owning chips, cloud and apps.

“They’ve made great advances in AI… Nvidia is a generation ahead of the industry — it’s the only platform that runs every AI model.”

Nvidia spokesperson (company statement)

Nvidia emphasized its continuing leadership even as it acknowledged Google’s progress. The comment underscores that large‑scale model support, software ecosystem and broad availability remain Nvidia strengths despite competitors’ gains in certain niches.

Unconfirmed

• The Information’s report that Meta will use Google’s TPUs in 2027 has not been publicly confirmed by Meta or Google with a detailed timetable or contract terms.
• The long‑term scale at which non‑Google firms will commit to TPUs versus multi‑vendor GPU strategies remains uncertain and will depend on price, performance and portability metrics still being negotiated.
• Public user and download counts (Gemini and ChatGPT) are snapshots and may not reflect active engagement, monetization, or enterprise adoption rates over time.

Bottom line

Google’s recent product launches, strategic chip deals and improving cloud metrics have materially altered how markets and competitors view its role in AI. Gemini 3’s technical gains plus the commercialization of TPUs give Google credible options to compete with OpenAI and to offer differentiated cloud‑AI packages to enterprise customers.

That said, structural challenges persist: cloud scale gaps versus Microsoft and AWS, the flexibility advantages of Nvidia GPUs, and the open question of customer lock‑in versus multi‑vendor strategies. Investors and enterprise buyers should watch actual TPU adoption rates, enterprise contract wins, and whether Gemini’s user engagement translates into durable revenue growth beyond short‑term market enthusiasm.

Sources

• Fortune — reporting and analysis (news)
• The Information — report on Meta talks (news)
• Google / Alphabet — corporate statements and investor communications (official)
• Nvidia — company statement (official)
• Sensor Tower — app download data (market analytics)
• Berkshire Hathaway — stake disclosure context (SEC/corporate filings)