ExpertsFMAmerica Votes Against Data Centers, Implications for Big Tech and Beyond, Were Google Bears Right All Along?
Sharp Tech with Ben Thompson · November 6, 2026
5 Selected Lessons:
reasoning_pattern · 3 case · 1 priority · 1
Shrink-to-fit strategy for platforms losing mass distribution (Reddit)
Source
Sharp Tech with Ben Thompson — America Votes Against Data Centers, Implications for Big Tech and Beyond, Were Google Bears Right All Along? (November 6, 2026) · at 1:13:20 ▸ · at 1:18:04 ▸ · at 1:24:20 ▸
When this applies
when a business built on free mass-scale distribution from a gatekeeper loses that distribution and must restructure around a smaller, known audience
The takeaway
When a business loses its scalable, anonymous-traffic funnel, prioritize converting the actual core value-generating users (often a small %) into known, monetizable relationships rather than mourning the loss of total volume.
Why it works
When volume-driven traffic disappears, revenue-per-user must rise to compensate, which requires knowing who your users are (login/identity) to target ads better — and recognizing that most of the genuine value (content creation) already comes from a small logged-in core, not the anonymous long tail.
One level up
Businesses built on someone else's free distribution must have a transition plan for when that distribution is withdrawn, since the gatekeeper's incentives can change.
Where it stops holding
This stops working if the core value-generating minority of users is too small or unmonetizable even when fully identified, or if the platform has no unique data/content asset (e.g., for resale to LLMs) to fall back on.
Detail
hierarchy:
- Recognize the core value-generating users (the ~10% who post/respond) are already logged in and most valuable
- Abandon reliance on anonymous, Google-driven traffic as the growth engine
- Require login to build targetable profiles and raise monetization per remaining user
- Preserve the separate revenue stream of selling continually fresh data access to LLMs
rationale: The 90% of lurkers who came via Google search were not actually core to Reddit's most valuable function (content generation), so losing them matters less than it appears on the surface.
In their words
where you maximize your revenue per user knowing you're not gonna have a maximum
Sentiment over facts in public opposition
Source
Sharp Tech with Ben Thompson — America Votes Against Data Centers, Implications for Big Tech and Beyond, Were Google Bears Right All Along? (November 6, 2026) · at 5:17 ▸
When this applies
when an institution assumes a backlash is caused by bad information and tries to win by refuting facts
The takeaway
When facing public backlash, diagnose whether you're dealing with facts or sentiment; refuting facts doesn't move sentiment-driven opposition.
Why it works
Opposition narratives function as confirmation-bias ammunition for a pre-existing sentiment (distrust/resentment), so disproving one fact just causes people to find another reason — the underlying sentiment is untouched.
One level up
Treat public opinion as driven by trust and perceived fairness first, information content second.
Where it stops holding
This stops working when the opposition truly is purely informational (no underlying resentment or distrust), in which case correcting facts can work.
Detail
approach: Zoom out from the specific false claim to ask why it's gaining traction; look for underlying resentment or distrust rather than engaging on factual details.
signals:
- people discard refuted claims and find new reasons to oppose
- opposition persists despite correction
- broader resentment toward the institution exists
rationale: Fake news/misinformation spreads because it confirms what people already want to believe, not because they lack correct information.
In their words
You're trying to refute facts when what you're actually dealing with is sentiment. Right. And it's the sentiment that is attracting stories, and that's why people will discard, okay, fine, that's not true, and they'll find another reason to oppose the data centers.
Scarcity vs. abundance flips strategic analysis of a commodity chain
Source
Sharp Tech with Ben Thompson — America Votes Against Data Centers, Implications for Big Tech and Beyond, Were Google Bears Right All Along? (November 6, 2026) · at 51:34 ▸ · at 55:36 ▸
When this applies
when analyzing any supply chain (e.g., a key input) where the endpoint assumption of 'eventually becomes a commodity' is used to discount a scarce intermediate player
The takeaway
Before applying 'eventually this becomes a commodity' logic to value a scarce resource provider, explicitly determine which regime (scarcity or abundance) currently governs, since the two produce opposite conclusions about where value accrues.
Why it works
In a scarcity regime, the entity delivering the most output-per-constrained-input (e.g. tokens per watt) captures outsized value because buyers will pay a premium for efficiency when the limiting factor (power) can't be relaxed; in an abundance regime that premium disappears and margins compress toward commodity levels.
One level up
Long-run end-state analysis is only useful once you've correctly estimated how long the current regime will persist — and misjudging that duration inverts your conclusions.
Where it stops holding
This stops working once the specific bottleneck (e.g., power) is actually resolved, at which point the abundance-regime logic (commoditization, margin compression) takes over.
Detail
approach: Identify whether the market is currently scarcity-bound or heading to abundance; map how value capture (premium for efficiency vs. commoditized margins) differs in each; estimate how long the scarcity regime will likely persist (Ben believes it's being elongated by the data center backlash).
signals:
- buyers willing to pay a margin premium for the most efficient unit under a hard constraint (power)
- a market with abundant power (e.g., China) not needing efficiency premium, defaulting to 'more/cheaper but less efficient' chips
rationale: NVIDIA's push into China made sense partly because China's power abundance meant CUDA/chip lock-in was the only way to keep margin, versus the US scarcity regime where efficiency itself commands a premium.
In their words
The analysis in a market defined by scarcity versus a market defined by abundance is fundamentally different
Market disruption favors new entrants precisely because incumbents are over-optimized
Source
Sharp Tech with Ben Thompson — America Votes Against Data Centers, Implications for Big Tech and Beyond, Were Google Bears Right All Along? (November 6, 2026) · at 1:35:32 ▸ · at 1:38:26 ▸
When this applies
when deciding whether a period of technological upheaval is a threat or an opportunity for a newcomer
The takeaway
Periods when 'everything is being shaken up' are opportunities for newcomers precisely because incumbents' deep optimization for the old paradigm becomes a liability, not an advantage, under the new one.
Why it works
Dominant incumbents become extremely optimized for their existing paradigm (e.g., the app store model) but that very optimization makes them structurally incapable of adapting to a new paradigm, opening space for new entrants who aren't tied to the old model.
One level up
Disruption redistributes opportunity away from the most optimized incumbents of the prior regime toward unencumbered newcomers of the new one.
Where it stops holding
This stops working when incumbents have genuine slack capital and willingness to cannibalize themselves (cf. the Google Capital Company case), in which case their scale advantage can persist into the new paradigm too.
Detail
approach: Evaluate how 'sclerotic' incumbents are to their core business — their very excellence there signals inability to adapt to a new paradigm, which is the newcomer's opening.
signals:
- a previously 'closed' market (dominant players everywhere) suddenly reopens due to a new paradigm shift
- incumbents are unbelievably optimized for their old core business but incapable of doing anything else
rationale: By the late 2010s web, the app-store paradigm was fully 'done' with dominant players locked in; AI reopens the field by changing the paradigm entirely.
In their words
It's easier to differentiate yourself in part because the dominant players have wiped out everyone else. They've become so tuned to that paradigm, the app store paradigm, the scale sclerotic
Loudoun County's 30-year virtuous cycle from data centers
Source
Sharp Tech with Ben Thompson — America Votes Against Data Centers, Implications for Big Tech and Beyond, Were Google Bears Right All Along? (November 6, 2026) · at 20:52 ▸
When this applies
when judging any large, concentrated infrastructure investment by its end-state outcome without accounting for the time and path needed to get there
The takeaway
Judging a long-cycle infrastructure investment by its mature end-state, while ignoring the multi-decade path and near-term local costs, misleads both advocates and critics.
Why it works
Early-mover data center infrastructure (interconnects, power, ecosystem) compounds over decades into a self-reinforcing cluster that eventually delivers lower taxes, lower power prices, and better services — but only after enduring an early period of strain (grid demand, price increases) before new power generation catches up.
One level up
Big-at-scale analysis must separate 'the eventual equilibrium' from 'the transition path,' since stakeholders who can vote/exit care most about the transition.
Where it stops holding
This stops working when a community cannot tolerate the near/medium-term downside (e.g., rising electricity prices) long enough to reach the eventual payoff, or has no historical first-mover advantage like Loudoun's early internet infrastructure.
Detail
situation: Loudoun County became a historic internet/data-center hub starting ~30 years ago due to interconnects, proximity to DC, power availability, and population.
decision: The ecosystem compounded: more data centers attracted more power infrastructure, more connectivity, reinforcing dominance.
outcome: Loudoun became the richest county in the country with falling property taxes and (per Ben's belief) falling power prices, plus better services — but only after decades, and new builds elsewhere will see price increases first.
lesson: You can't sell communities on a 30-year payoff while they bear short/medium-term costs (e.g., electricity price hikes) without mitigating or compensating them.
In their words
So, yes, you're seeing the outcome of a thirty year virtuous cycle. In the meantime, if you build a data center