Google is the Consumer AI Company Everyone is Waitig for. ARD #80
At Google I/O 2026 this week, the theme became clear: Google Gemini is becoming Google Search. And AI Search is becoming Google Search.
The scale frame matters. Over 3 billion people use Google Search worldwide every day — vs. roughly 900 million weekly users of ChatGPT, and a comparable weekly number on Google Gemini today as well. The daily-usage gap is enormous — and Google is now wiring Gemini into the same Search box those 3 billion already open every day.
Three Takes today, all on Google.
(1) Google I/O 2026 Highlights — “Google AI Is Becoming Google Search”
The big story out of Google I/O 2026 this week — full YouTube keynote here — is that Google AI is becoming Google Search.
The single most-reported moment is the Search box itself. The New York Times ran a piece titled “Google changes its Search box for the first time in 25 years” — and The Verge framed it the same way in “The future of Google is a Search Box that does everything”. The Verge also has a useful rundown of the 13 biggest announcements for a fast scan. The Information’s angle in “Google’s AI Search Leap Forward” is the most analyst-grade read of the three.
What did they actually ship.
Gemini 3.5 — Flash today, Pro soon — Google’s strongest AI Agentic and Coding model yet. Especially coupled with Antigravity 2.0, their AI Coding platform. Going up directly against Anthropic 4.7 and OpenAI 5.5.
Gemini Omni — any-to-any multimodal — text in, image out, audio in, video out, in any combination. The enterprise version of where consumer-facing multimodality is heading.
Gemini Spark — a 24/7 personal AI agent — Google’s answer to OpenAI’s OpenClaw and Anthropic’s Claude Code / Codex / Cowork lineup. Spark is the agentic-AI surface that runs in the background and acts on the user’s behalf. Available in beta to paid tier subscribers next week.
For longtime readers — I framed the underlying Google personal-intelligence rollout in AI-RTZ #1020, “Google Gemini AI Rolls Out Personal Intelligence” — that piece is the through-line connecting where Google was a few months ago and where I/O 2026 just took it.
MP Take: “Google continues to execute wide and deep on its Gemini AI and TPU infrastructure opportunities. At I/O 2026 they drew up a product road map that covers almost everything offered by competitors — with ‘good enough’ or better versions of each. Then integrated with core Google apps and services used by billions of people daily. From Search to Gmail to Docs to YouTube and many more. All of it is leveraged by the immense TPU-powered Google Cloud AI data centers — that Google is investing almost $200 billion in this year alone. This is very competitive for Google Cloud versus Amazon AWS and Microsoft Azure and others. I’ll have more detailed Substack AI-RTZ posts soon after Google I/O 2026 for Developers wraps up later this week.”
(2) Google Overhauls Its AI Pricing Plans
Take 2 — pricing.
ZDNET ran a useful walkthrough titled “Google’s new AI pricing plans — which should you choose” covering the new tier structure Google rolled out alongside I/O. The headline shift is more a la carte, more competitive entry pricing — and the free ad-supported tier just keeps absorbing more of Gemini’s capabilities for that 3-billion-user-a-day Search base.
I covered the broader industry move to a la carte in AI-RTZ #1064, “Anthropic and OpenAI’s Velvet Rope” — which framed why all three labs are converging on tier-mix experimentation. The Apple-Gemini distribution layer I covered in AI-RTZ #965, “Apple Goes With Google Gemini” — that distribution channel makes Google’s free-tier wedge even bigger going forward.
MP Take: “Google is leveraging its vertical AI Compute stack to offer more competitive subscription and a la carte pricing models than their competitors like OpenAI and Anthropic especially can afford. And leveraging its infrastructure to incorporate many versions of its latest models into the ‘free’ ad-based Google Search that serves over 3 billion users daily worldwide. This gets more powerful with its upcoming Apple distribution partnership with Siri and Gemini. Google in many ways is steering away from its peers in its ability to leverage pricing from the free to a la carte tiers. Expect this advantage to widen going forward.”
(3) Google’s Overall Tokenmaxxing as a Company
Take 3 — Impressive growth in tokens processed, across all of GOogle.
On the Google I/O keynote stage (Axios coverage), CEO Sundar Pichai noted that Google is now processing 3.2 quadrillion tokens per month — a key measure of total AI workload across the company’s products.
The growth curve is the story:
-
Two years ago — 9.7 trillion tokens per month.
-
One year ago — 480 trillion tokens per month.
-
Today — 3.2 quadrillion tokens per month.
That’s a roughly 6x+ year-over-year acceleration on a base that itself was already up a lot a year ago.
I framed the broader industry move to use this number as a justifying metric in AI-RTZ #1033, “The Latest Engineer Financial — Silicon Valley’s Tokenmaxxing Obsession”. The whole sector is now leaning on tokens-per-month as the headline KPI for justifying the hundreds of billions in AI data center capex.
MP Take: “Expect more of this from the big tech companies — highlighting these token metrics and related numbers as another justification for the hundreds of billions in AI Data Center spend. More useful will be more granular analysis on these tokens down the road — assuming the companies provide that level of granular detail. This is a ‘put a pin in this metric’ type of moment — for Google and for the AI industry writ large.”
Overall MP Take
Google continues to execute wide and deep on its Gemini AI and TPU infrastructure opportunities.
At Google I/O 2026, Google drew up a product road map that covers almost everything offered by competitors — with “good enough” or better versions of each. Then integrated with core Google apps and services used by billions of people daily. From Search to Gmail to Docs to YouTube and many more.
All of it is leveraged by the immense TPU-powered Google Cloud AI data centers — that Google is investing roughly $200 billion in this year alone.
This is very competitive for Google Cloud versus Amazon AWS and Microsoft Azure and others.
I’ll have more detailed Substack AI-RTZ posts soon after Google I/O 2026 for Developers wraps up later this week — there’s still material I want to dig into once the developer-track sessions land.
The single-line read: of the consumer-AI giants, Google is the one with both the daily-distribution surface (3 billion+ Search users) AND the model-tier vertical stack (Gemini 3.5, Omni, Spark, on TPUs). Nobody else has both. OpenAI has model momentum but no distribution surface anywhere near 3 billion. Anthropic has model quality and enterprise traction but a smaller consumer footprint. Apple has the device distribution but is using Google Gemini for the AI layer. Google is the company everyone in the consumer-AI race is now structurally measured against.
And they’re step-for-step with OpenAI and Anthropic on the developer / enterprise side with their Antigravity 2.0 AI Coding products and the latest versions of LLM AI models Google Gemini 3.5 Pro and beyond.
Google has the opportunity to match everything the peers offer — and build a lot of distance from them with its vertical stack, global distribution, and meaningful financial resources. It’s the opportunity Microsoft had vs. Netscape in the 1990s Internet wave — and far more. And without the regulatory issues that constrained Microsoft then.
And a market cap approaching Nvidia’s leading number of five trillion dollars.
Gadget AI — Google Gemini App Gets Redesigned with “Neural Expressive” UI/UX and “AI Cursor”
The Verge covered the redesign in “Google Gemini App gets updated with ‘Neural Expressive’ design” — the new Gemini app pairs the Gemini 3.5 Flash model with a UI/UX system Google is calling “Neural Expressive.”
The naming alone signals where Google is going — “Neural Expressive” frames the AI surface as emotionally responsive, conversationally fluid, visually dynamic — rather than the chat-box-plus-buttons paradigm that’s defined the LLM app category since ChatGPT.
MP Take: “The Gemini app redesign is the consumer-facing complement to the Search-box overhaul Take 1 covered. Two product surfaces moving in lockstep: the everyday Search box for the 3 billion daily Google users, and the dedicated Gemini app for the AI-native power users who want Gemini 3.5 directly.
The ‘Neural Expressive’ UI/UX on the App is functionally differentiated. It’s a deliberate departure from the LLM-as-chatbox UI paradigm — and a signal that Google sees the next consumer-AI competitive layer as interface design, not just model quality. Whoever ships the most expressive, most fluid, most natural-feeling AI surface wins the daily-habit war — and Google has the design talent, the model integration, and the distribution to play that game at scale.
I especially like the AI Cursor features Google showed off earlier — around its Android / Googlebook announcements. The AI Cursor is the on-screen agentic pointer that complements the Neural Expressive surface — a small but telling product detail that signals Google is thinking about how the AI actually navigates a user’s screen, not just how it talks to them.
Pair all this with the Search-box redesign and Gemini Spark and you have the three legs of Google’s consumer-AI stool: ambient AI in Search · explicit AI in the app · agentic AI through Spark. Each one is good enough on its own. Together, they’re the closest thing any company has to a coherent consumer-AI platform in 2026.”
Questions
Q1 — What’s the one Google I/O 2026 product MP is most looking forward to trying?
Answer: Google Gemini Spark — Google’s new AI agent product that goes head-to-head with Claude Code / Cowork.
The Wall Street Journal has a useful frame in “Google unveils new Gemini AI agent for personal tasks”. Spark is positioned as a 24/7 personal agent — running in the background, acting on user intent, integrated across Google’s app surfaces.
If Spark delivers even part of what the demo promised, it’s the closest a hyperscaler has come to matching Anthropic’s Cowork experience with lots of security, guardrails, and hand-rails for a mainstream user base — except Spark ships with billions of users one click away through the existing Google account graph. The agentic-AI category is hot, and Spark just became its most-distributed entrant overnight.
Q2 — What’s the second one?
Answer: Google Docs Live.
The Wall Street Journal covered the rollout in “Google Docs Live test”. The pitch is real-time AI-augmented co-authoring in Google Docs — Gemini riding alongside the user as they write, suggesting, editing, completing, restructuring.
For a daily Google Docs user — and that’s most of the global writing-knowledge-worker class — this is the most concrete day-one productivity upgrade Google I/O 2026 shipped. Cowork-style AI assistance, but native to the doc you’re already in. Quietly the most usable announcement of the keynote.
Clips from today
Clip 1 — Google’s 3.2 Quadrillion Token Math
Sundar Pichai on the Google I/O 2026 keynote stage: Google is now processing 3.2 quadrillion tokens per month. A year ago: 480 trillion. Two years ago: 9.7 trillion. Roughly 6-to-7x year-over-year acceleration on an already-enormous base.
MP Take: “Expect more of this from the big tech companies — highlighting these token metrics and related numbers as another justification for the hundreds of billions in AI Data Center spend. More useful will be more granular analysis on these tokens down the road — assuming the companies provide that level of granular detail. This is a ‘put a pin in this metric’ type of moment — for Google and for the AI industry writ large.”
Clip 2 — Tesla’s $1.5T AI Piggy Bank
On Google’s $190B AI data center spend this year, MP frames the Elon Musk capital muster: Tesla as Elon’s $1.5 trillion piggy bank, with SpaceX/XAI as a separate $1.75 trillion piggy bank number two. A lot of capital chasing AI — but Google has the assets and the actual distribution to deploy that capital across 3 billion daily Search users.
MP Take: “Google has the opportunity to match everything the peers offer — and build a lot of distance from them with its vertical stack, global distribution, and meaningful financial resources. It’s the opportunity Microsoft had vs Netscape in the 1990s Internet wave — and far more. And without the regulatory issues that constrained Microsoft then.”
Clip 3 — Google’s $5T Market Cap & Berkshire Bet
Google’s market cap is now approaching $5 trillion — close to Nvidia’s leading Mag Seven number. And Berkshire Hathaway under the new Warren Buffett successor Greg Abel is adding Google shares. None of this is stock recommendations, but it underlines that investors are starting to recognize the Google AI differentiation that other coverage has missed.
MP Take: “If there’s anyone who can make some of these AI services more liked by consumers and more widely used by consumers, it’s probably Google — just like they did with search. People already know search. They’re going to use AI from Google like they do search, and it’s a relatively straightforward habit. There’s over 3 billion people who search on Google every day. That’s a great number to start from.”
We’re not here to deliver the AI news — we’re here to help you think through it.
ARD 81 on AI-RTZ #1093 next. Stay tuned.
NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here.
Links
Theme — Google I/O 2026
-
Google I/O 2026 event site: https://io.google/2026/
-
YouTube Keynote:
-
NY Times — Google changes its Search box for the first time in 25 years: https://www.nytimes.com/2026/05/19/business/google-seach-bar-ai-gemini.html?unlocked_article_code=1.jlA.95yh.ptfBUHf-rBtB&smid=url-share
-
The Information — Google’s AI Search Leap Forward: https://www.theinformation.com/newsletters/the-briefing/googles-ai-search-leap-forward
-
The Verge — The future of Google is a Search Box that does everything: https://www.theverge.com/tech/934217/google-search-box-does-everything-ai-io-2026
-
The Verge — 13 biggest announcements at Google I/O 2026: https://www.theverge.com/tech/933415/google-io-2026-biggest-announcements-ai-gemini
-
Google Blog — Gemini 3.5 (Flash today, Pro soon): https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/
-
VentureBeat — Google unveils Gemini Omni: https://venturebeat.com/ai/google-unveils-gemini-omni-any-to-any-ai-model-what-enterprises-should-know
-
Engadget — Google’s Gemini Spark is an agentic AI assistant: https://www.engadget.com/2176556/googles-gemini-spark-is-an-agentic-ai-assistant/
-
AI-RTZ #1020 — Google Gemini AI Rolls Out Personal Intelligence (backcat):
Pricing
-
ZDNET — Google’s new AI pricing plans: https://www.zdnet.com/article/google-new-ai-plans-which-should-you-choose-2026/
-
AI-RTZ #1064 — Anthropic and OpenAI’s Velvet Rope:
-
AI-RTZ #965 — Apple Goes With Google Gemini:
Tokenmaxxing
-
Axios — AI tokens everywhere (Pichai keynote): https://www.axios.com/2026/05/19/google-ai-youtube-gemini
-
AI-RTZ #1033 — Silicon Valley’s Tokenmaxxing Obsession:
Gadget AI — Gemini App “Neural Expressive” + “AI Cursor”
-
The Verge — Google Gemini App “Neural Expressive” redesign: https://www.theverge.com/tech/933699/google-gemini-redesign-ai-3-5-flash-io-2026
-
DeepMind Blog — AI Pointer / “AI Cursor”: https://deepmind.google/blog/ai-pointer/
Questions — Spark + Docs Live
-
WSJ — Google unveils new Gemini AI agent (Spark): https://www.wsj.com/tech/ai/google-unveils-new-gemini-ai-agent-for-personal-tasks-b8093197?st=EkSGBz&reflink=article_imessage_share
-
WSJ — Google Docs Live test: https://www.wsj.com/tech/personal-tech/google-docs-live-test-e4473e07?st=agufZS&reflink=article_imessage_share
Today’s companion post + episode + clips
-
AI-RTZ #1092 — Regular Folks Want to Slow Down AI in the US (today’s companion):
-
ARD 80 — Main on YouTube:
-
Short 1 — Google’s 3.2 Quadrillion Token Math:
-
Short 2 — Tesla’s $1.5T AI Piggy Bank:
-
Short 3 — Google’s $5T Market Cap & Berkshire Bet:
Subscribe to AI: Reset to Zero for daily AI Ramblings + Sunday Bigger Picture posts.
