Sundar Pichai at Google I/O 2026: Keynote Recap and What It Means

Sundar Pichai's Google I/O 2026 keynote opened with a landmark figure: the Gemini app crossed 900 million monthly active users, more than double the 400 million reported a year earlier, while AI Mode in Search separately surpassed one billion monthly active users, according to Pichai's keynote blog post. The session also unveiled two new TPU chip architectures, a restructured AI Ultra pricing plan, and Google's formal declaration of the "agentic Gemini era."

StartupHub data

Companies working on this

Profiles of the companies named in this story, with funding and a one-liner from our database.

Google Search

$25M

The world's most dominant and influential web search engine, organizing information and making it accessible.

Google DeepMind

$677M

Pioneering AI research and development to solve intelligence and advance science for humanity.

Gemini Spark

gemini-spark.io es su primera y mejor fuente de información sobre gemini spark. Aquí también encontrará temas sobre cuestiones de interés general. ¡Deseamos que encuentre lo que está buscando!

Gemini 3.5 Flash and the pivot from chat to action

Pichai opened with a statement that functioned as a mission update: "It's clear we're firmly in our agentic Gemini era." The phrase reframes what Google is asking consumers to expect from its AI products. Gemini 3.5 Flash, the first model released that day across the Gemini app, Google Search, and the Gemini API, is described by Google as "our first in a series of models combining frontier intelligence with action," a deliberate shift away from question-answering toward task completion, per the official keynote post.

The model surpasses Gemini 3.1 Pro across coding, agentic, and multimodal benchmarks while running at four times the output-token speed of other frontier models, CNBC reported. Gemini 3.5 Pro, the heavier sibling, is in testing and due in June 2026. Two additional releases complete the I/O slate: Gemini Omni, an any-to-any model that accepts and emits image, audio, video, and text; and Gemini Spark, an agentic assistant initially exclusive to AI Ultra subscribers, per 9to5Google.

The user-growth data gives these releases urgency. Daily request volume on Gemini grew sevenfold year-on-year, per Pichai's keynote. AI Mode in Search, one year since launch, hit one billion MAU with queries "more than doubling every quarter." The scale of that distribution is what separates Google's position from every pure-play AI lab. For additional context on how Google builds agent systems, see our coverage of DeepMind's agent infrastructure.

Silicon as strategy: the TPU 8t and 8i split

The most structurally significant announcement at I/O was not a model but a chip architecture decision first disclosed at Google Cloud Next in April and reinforced on stage. Google's eighth-generation TPU family divides into two purpose-built designs: the TPU 8t for training and the TPU 8i for inference. The split is a public signal that Google now regards training compute and inference compute as different economic problems requiring different silicon, per the Google infrastructure blog.

TPU 8t scales to 9,600 chips in a single superpod with two petabytes of shared high-bandwidth memory, achieving three times the raw compute of the previous-generation chip at up to twice the performance per watt, Bloomberg reported in April. TPU 8i, the inference sibling, connects 1,152 chips per pod with three times the on-chip SRAM; Google notes the design prioritises latency because "interactions between agents at scale magnify even small inefficiencies." Both chips target general availability later in 2026.

This is Pichai's most direct structural answer to Nvidia's dominance in AI compute. Proprietary training and inference silicon lets Google price Gemini API access below the cost achievable by cloud providers reliant on third-party GPUs, a margin advantage that compounds as agent-call volumes grow. For context on the broader capital flows reshaping AI infrastructure, see our piece on the inference-layer funding wave of May 2026.

Products and pricing: Android Halo, Gemini Intelligence, and the restructured AI Ultra tiers

Pichai used I/O to show how Gemini Intelligence, an agentic layer embedded into Android, ChromeOS, Wear OS, and Android Auto, shifts the product question from "what can the chatbot answer" to "what can the agent complete while the phone is in a pocket." Android Halo, a persistent agent interface on Android phones, is due in summer 2026 and is designed to run Gemini Spark-class tasks in the background, including booking, research, and summarisation, without requiring the user to open an app, per 9to5Google. Android XR audio glasses with iPhone support follow in autumn 2026.

The pricing restructure signals confidence in the consumer AI subscription market. AI Ultra now costs $200 per month, down from $250, while a new $100-per-month tier was introduced below it; Google's stated aim is to align price points with the direct competitors in the professional-user segment, per CNBC. The $50 reduction at the top tier, alongside adding a new entry point, suggests Google views the AI subscription customer base as elastic enough to expand significantly with better pricing rather than to protect existing revenue at the margin.

What it means

Google I/O 2026 is the clearest public statement Pichai has made about where he believes the AI value chain settles. The company is betting that the combination of proprietary training silicon, proprietary inference silicon, a billion-user Search distribution layer, and a 900-million-user Gemini app creates a cost and reach position that pure-play AI labs cannot replicate. The word "agentic" ran throughout the keynote not as marketing but as a structural claim: that the next phase of AI revenue comes from background task completion, not from conversations that users must initiate. For Pichai, I/O 2026 was less a product launch and more a statement of the full-stack position Google intends to hold.

Sources

• Google Blog: Sundar Pichai's Google I/O 2026 opening keynote
• 9to5Google: Everything Google announced at I/O 2026
• CNBC: Google unveils AI model Gemini 3.5 and AI agent Gemini Spark
• TechCrunch: Google updates its Gemini app to take on ChatGPT and Claude at IO 2026
• Bloomberg: Google Cloud Unveils New TPU Chips to Accelerate AI Training and Inference
• Google Blog: Our eighth generation TPUs: two chips for the agentic era

Editorial standards: every claim is sourced. Tips: [email protected]