The Data Center Veto (This Week in…

1. Key Themes

Theme 1: Physical Infrastructure as AI's Achilles Heel

AI's real-world dependency on data centers creates a structural vulnerability that didn't exist with previous tech waves like globalization. Local communities now hold genuine veto power over AI development through permitting and zoning processes.

"AI, however, depends on data centers in the physical world, and building data centers needs permission. This gives normal people the sort of veto power over AI they didn't have in the face of globalization."

Theme 2: The Collapse of Ad-Supported Internet Economics

The rise of agentic web traffic is fundamentally threatening the advertising model that has underwritten the open internet. As AI agents increasingly mediate content consumption, the human-eyeball assumption baked into ad economics breaks down.

"What will the internet look like when ad-supported models are rendered obsolete by shifting user behavior and the rise of agentic web traffic?... ads make sense for humans... incentivizing content for agents will be different."

Theme 3: Google's Distinct AGI Path via DeepMind

While the AI narrative is dominated by OpenAI and Anthropic, DeepMind is quietly pursuing a meaningfully different technical approach to AGI — one worth tracking separately, especially given Google's foundational role in the current AI era via the transformer architecture.

"When Ben highlights a DeepMind approach to building AGI that's distinct from the approaches at OpenAI and Anthropic, I'm compelled to both pay attention, and remember: for all of Google's faults and misses, they do in fact have plenty of historic hits."

Theme 4: The Inference Shift as a Structural Market Transition

A dedicated video update on "The Inference Shift" signals that Thompson views the move from training-dominant to inference-dominant AI economics as a major structural inflection — with significant downstream implications for chips, cloud infrastructure, and AI product design.

"This week's Stratechery video is on The Inference Shift."

2. Contrarian Perspectives

Perspective 1: Fighting Data Center Misinformation Is a Losing Strategy

The tech industry's instinct to combat opposition to data centers by correcting misinformation misdiagnoses the problem. The opposition is rational and structural — rooted in a genuine power dynamic — not simply a result of ignorance. Correcting misinformation treats the symptom, not the cause.

"Understanding this dynamic is more important than trying to correct misinformation, which is a symptom, not a cause, of data center opposition."

The implication: the only workable solution is economic compensation to affected communities — effectively buying consent rather than winning arguments.

"There are understandable reasons for people to oppose data centers; the only solution that will work is simply paying them off."

Perspective 2: Google's Apparent Chaos May Still Produce AI Leadership

Conventional wisdom frames Google's scattered AI strategy as a competitive liability. Thompson pushes back, noting that Google — despite being "ungovernable and poorly coordinated" — invented the transformer that made the entire AI era possible and is now a ~$5 trillion company.

"Google is now a nearly $5 trillion company and its transformer architecture supercharged the AI era... for all of Google's faults and misses, they do in fact have plenty of historic hits."

Perspective 3: Content Monetization Must Be Rebuilt From Scratch for the Agentic Web

Rather than adapting existing ad or subscription models, the agentic web may require an entirely new economic architecture for content creation and compensation — a much more radical claim than most current discourse acknowledges.

"Incentivizing content for agents will be different, and how Agarwal and Parallel are trying to solve them."

3. Companies Identified

Google / DeepMind

Description: Alphabet's AI research lab and parent company
Why Mentioned: Taking a distinct technical approach to AGI compared to OpenAI and Anthropic; Google I/O showcased a broad but scattered AI product push
Quote: "Ben highlights a DeepMind approach to building AGI that's distinct from the approaches at OpenAI and Anthropic."

Parallel (formerly referenced in context of Parag Agarwal)

Description: New startup focused on content valuation and monetization infrastructure for the agentic web
Why Mentioned: Directly building solutions for the collapse of ad-supported internet economics in an AI agent-driven world
Quote: "Parag Agarwal, former CEO of Twitter, is now focused on devising solutions for exactly this reality... how Agarwal and Parallel are trying to solve them."

OpenAI & Anthropic

Description: Leading frontier AI labs
Why Mentioned: Cited as the standard-bearers of AGI development against which DeepMind's distinct approach is contrasted
Quote: "A DeepMind approach to building AGI that's distinct from the approaches at OpenAI and Anthropic."

4. People Identified

Parag Agarwal

Description: Former CEO of Twitter; now founder of Parallel
Why Mentioned: Building new economic infrastructure to value and incentivize content creation in a world where AI agents, not humans, are the primary content consumers
Quote: "Parag Agarwal, former CEO of Twitter, is now focused on devising solutions for exactly this reality... the economics of content on the Internet, why ads make sense for humans, and why incentivizing content for agents will be different."

Ben Thompson

Description: Founder and primary author of Stratechery
Why Mentioned: Authored the data center discontent analysis and the Google I/O / DeepMind piece; produced the Inference Shift video
Quote: "When Ben highlights a DeepMind approach to building AGI that's distinct from the approaches at OpenAI and Anthropic, I'm compelled to both pay attention."

Andrew Sharp

Description: Co-host of Sharp Tech and contributing editor at Stratechery
Why Mentioned: Curated and contextualized the agent economics and Google I/O themes for this week's roundup
Quote: "I learned a ton from this interview, and I bet you will, too."

John Gruber

Description: Author of Daring Fireball
Why Mentioned: Co-hosts Dithering with Ben Thompson; discussed data center unpopularity and Google
Quote: Referenced via episode titles "Data Center Unpopularity" and "Google Being Google."

Bill Bishop

Description: Author of Sinocism newsletter
Why Mentioned: Co-hosts Sharp China; covered US-China stability dynamics and Trump's Taiwan comments
Quote: Referenced in episode: "Constructing US-China Stability; Trump's Taiwan Comments and More Summit Takeaways."

5. Operating Insights

Insight 1: Pay, Don't Persuade — On Navigating Community Opposition to Physical Infrastructure

For any operator or investor deploying physical AI infrastructure (data centers, energy assets, compute facilities), the playbook is economic, not communicative. Attempting to win public debate over local opposition is a distraction from the only lever that reliably works.

"The only solution that will work is simply paying them off."

Insight 2: Build for Agents, Not Just Humans

Entrepreneurs building content, media, or data products need to design monetization and distribution architectures that account for AI agents as first-class consumers — not just as a future consideration, but now. The Parallel venture signals this is already an investable category.

"Incentivizing content for agents will be different... Agarwal and Parallel are trying to solve them."

6. Overlooked Insights

Overlooked Insight 1: The Inference Shift as an Investment Signal

The dedicated video on "The Inference Shift" is buried at the bottom of the newsletter but may be the most consequential investment theme of the week. A shift from training to inference dominance would reshape demand for chips (favoring different GPU/accelerator profiles), cloud pricing models, and AI product economics. The article provides no detail, making it easy to skip — but Thompson rarely dedicates standalone video content to topics without high conviction.

"This week's Stratechery video is on The Inference Shift."

Overlooked Insight 2: Semiconductor Supply Chain Depth via Asianometry

Two Asianometry episodes — on vertical-cavity surface-emitting lasers (VCSELs) and Intel's three-decade manufacturing presence in Costa Rica — are listed without commentary but touch on critical, under-discussed nodes in the AI hardware supply chain. VCSELs are essential to fiber-optic data center interconnects; Intel's Costa Rica operations represent a meaningful non-Taiwan manufacturing footprint.

"The Little Vertical Laser That Everyone Uses" and "Intel's 30 Years in Costa Rica" — listed under Asianometry with Jon Yu.