Everyone Hates AI. Everyone's Using It. Here's What That Gap Means for Your Portfolio.
- Enterprise AI adoption surged from 55% to 78% in one year while public anti-AI sentiment hit all-time highs
- The four largest hyperscalers are spending $700 billion on AI infrastructure in 2026 — more than the entire US defense budget
- This "say one thing, do another" pattern has occurred with every major technology for 600 years — printing press, electricity, automobiles, internet
- Public hatred doesn't stop adoption. It delays consumer-facing deployment by 1-3 years while having near-zero effect on B2B infrastructure
- The Walmart parallel is exact: communities protested, then everyone shopped there. Hatred is a timing variable, not a directional one
- Consumer-facing companies (Meta, Google) carry more sentiment risk. B2B infrastructure (Nvidia, enterprise software) is nearly immune
- Countries that regulate from fear — Britain's Red Flag Act, the EU AI Act — cede the industry to those that don't
The Walmart Effect
In the 1990s and 2000s, when Walmart announced a new store location, communities lost their minds.
Town hall meetings. Newspaper editorials. Protest signs. "Save Main Street" campaigns. City council members running on anti-Walmart platforms. Documentaries. Lawsuits. Academic papers on the death of small-town America.
The outrage was real. The concerns were legitimate — local businesses did close. Downtown corridors did change. The cultural disruption wasn't imaginary.
Then everyone shopped there.
Not some people. Everyone. The same residents who signed the petition bought their groceries at Walmart. The same city council members who voted against the rezoning filled their prescriptions at the Walmart pharmacy. By 2023, Walmart had 4,700 US locations and 90% of Americans lived within 10 miles of one.
Public outrage existed in parallel with accelerating adoption. Not before it. Not instead of it. Alongside it. People can hate something and use it at the same time. This isn't hypocrisy — it's how humans process disruptive change. And it's happening right now with AI.
The Numbers Don't Lie (But the Sentiment Does)
Here's the situation as of April 2026. Two sets of numbers that should not be able to coexist, but do:
Read those numbers again. Enterprise adoption is accelerating while public fear is also accelerating. Those two lines are moving in opposite directions on the same chart.
That gap has a name in finance: mispricing.
When emotional sentiment diverges from economic behavior, asset prices don't reflect reality. They reflect feelings. And feelings, historically, are terrible investment signals.
What $700 billion actually looks like
The four hyperscalers — Amazon, Microsoft, Google, and Meta — are spending roughly $700 billion on AI infrastructure in 2026. That number is hard to get your head around, so let's put it in context:
- It's more than 1% of US GDP
- It exceeds half of the entire US defense budget
- It's larger than all EU defense spending combined
- It dwarfs total dot-com era telecom and internet spending
These aren't companies making exploratory bets. This is government-scale capital allocation by organizations with access to better data than anyone else on the planet. They can see their own adoption curves. They can see enterprise usage metrics. They can see the revenue trajectory.
They're not spending $700 billion because YouTube commentators are excited about AI. They're spending it because their customers are using it.
We've Literally Seen This Movie Nine Times
This is the part that people miss because they don't read history. What's happening with AI right now — the simultaneous hatred and adoption — has happened with every major technology for the last 600 years. Not approximately. Not loosely. Exactly.
| Technology | The Hatred | What Actually Happened | Who Won |
|---|---|---|---|
| Printing Press (1440s) | Venetian scribe called it "a whore." Catholic Church created the Index of Forbidden Books (maintained until 1966). | Literacy rates: 10% → 50%+. Scientific Revolution. Enlightenment. Protestant Reformation. | Luther (300K+ copies by 1520). Nations that didn't ban it (Netherlands, England). Protestant economies. |
| Electricity (1880s) | Gas companies funded fear campaigns. Edison electrocuted elephants publicly to discredit AC current. | Cities that adopted early: economic boom. Cities that delayed: fell behind within a decade. | Westinghouse, GE. Cities that wired up fast. Literally every industry that existed after 1920. |
| Automobiles (1900s) | UK Red Flag Act: person must walk ahead of every car waving a red flag. Judge called cars "houses of prostitution on wheels." | UK had early auto pioneers. Lost entire industry to US and Germany. GDP per capita tripled 1900-1930. | Henry Ford (richest person alive). Standard Oil. Suburban developers. Gas stations. Motels. Fast food. |
| Radio (1920s) | "Will destroy newspapers." "Will rot brains." "Will enable mass propaganda." | Created entirely new industries. Didn't kill newspapers — forced them to specialize. | RCA/NBC. CBS. P&G (world's largest advertiser via soap operas). FDR (4 elections). |
| Television (1950s) | Darryl Zanuck (Fox): "TV won't hold any market." "Will rot children's brains." | NBC/CBS/ABC dominated media for 40 years. Disney's 1954 TV pivot was mocked, proved visionary. | Madison Avenue. JFK. Ted Turner/HBO. Anyone who built for the medium instead of against it. |
| Video Games (1980-2020) | 30+ years of bipartisan outrage. Congressional hearings. "Causes mass shootings." Jack Thompson (later disbarred). | Zero successful US legislation. Supreme Court: games = protected speech (7-2, 2011). | Nintendo. Rockstar (GTA V: $8B+). NVIDIA (gaming GPUs became AI foundation — now $2T+ company). |
| Internet (1990s) | Krugman 1998: "Impact no greater than fax machine." "Will enable predators." Dot-com crash "proved" it was a bubble. | Amazon: $107 → $7 → $3,000+. Anyone who sold during the crash made the worst financial decision of their lives. | Bezos. Page & Brin. Thiel ($500K → $1B on Facebook). Everyone who held through the crash. |
| Social Media (2010s) | Congressional hearings. Documentaries. "Destroying democracy." "Destroying teen mental health." | 3.9 billion users. Meta stock recovered to ATH despite constant criticism. TikTok: fastest-growing platform in history during maximum backlash. | Zuckerberg. ByteDance. Creator economy ($100B+ annually). Influencer marketing industry. |
| Walmart (1990-2000s) | Protests. Documentaries. Lawsuits. "Save Main Street." Academic papers on small-town devastation. | 4,700 US locations. 90% of Americans within 10 miles. $611B annual revenue (2023). | Walmart. Its suppliers. Logistics companies. Consumers who saved money. |
Phase 1: Early adopters experiment. Phase 2: Incumbents warn of civilizational threat — identical claims every time ("corrupts youth," "destroys jobs," "undermines authority," "enables immorality"). Phase 3: Regulatory scramble, usually too late and misguided. Phase 4: Society adapts; new industries emerge nobody predicted. Phase 5: Technology becomes invisible infrastructure; previous panic seems absurd.
AI in April 2026: We are in late Phase 2, entering Phase 3. The panic is at peak volume. The adoption is accelerating underneath it. This is, historically, the optimal entry point for investors who can separate emotion from economics.
The Say/Do Gap: What Companies Tell the Public vs. What They Actually Spend
Here's something the YouTube AI-haters don't know, because they've never looked at an earnings call transcript: companies are talking out of both sides of their mouths.
Publicly, many companies are cautious about AI. They publish "responsible AI" frameworks. They hire ethics boards. They issue press releases about "human-centered" approaches. They're very careful not to look like they're replacing workers.
Privately, the numbers tell a different story. Here it is, side by side:
| What America Says | What America Does |
|---|---|
| "AI is just a chatbot that makes stuff up" | 95% of OpenAI engineers use AI coding tools daily. Anthropic's models write the majority of their internal code. |
| People still sharing Will Smith AI spaghetti video from 2023 | Darren Aronofsky is making an entire feature film with AI in 2026. The mockery is 3 years out of date. |
| "Nobody's actually using this stuff" | Enterprise AI adoption: 55% → 78% in one year. Gemini: 450M monthly users, 50% quarterly growth. |
| "AI companies are burning cash with no revenue" | Anthropic: $0 → $100M → $1B → $4B+ in three years. Google Cloud: $13.6B, up 32% YoY, margins doubled. |
| "It'll blow over like crypto" | $700B hyperscaler AI capex in 2026. More than the entire US defense budget. BCG, McKinsey, Accenture deploying "AI co-workers at scale." |
| "AI can't really code" | SWE-bench coding scores: 4.4% → 71.7% in one year. AI task completion doubling every 7 months. |
| "It's too expensive to actually use" | GPT-4-equivalent capability costs 90% less than at launch. Inference costs down 50x in 3 years. |
| Half of YouTube feed: "AI is evil" | AI overviews: 2 billion monthly active users. Quietly becoming the default way people get answers. |
That table is the entire investment thesis in one glance. The left column is what drives sentiment. The right column is what drives stock prices. They're living in different realities.
For any company, look at two numbers: (1) what they say about AI in press releases and public statements, and (2) what they spend on AI in their capital expenditure. The wider the gap, the bigger the opportunity. Companies being publicly cautious while privately deploying aggressively are the ones most likely to surprise on earnings. Companies that are loud about AI but aren't spending are the ones most likely to disappoint.
The data extraction gap
Want to see the say/do gap at its most extreme? Look at content scraping. According to Cloudflare's data, Google scrapes 15 web pages for every 1 visitor it sends back to the content creator. Ten years ago that ratio was 1:2.
OpenAI's scrape-to-visitor ratio has worsened from 250:1 to 1,500:1 in six months.
These companies publicly advocate for content partnerships and responsible AI while privately extracting orders of magnitude more data than they return in traffic. The "responsible AI" framing is brand management. The capital allocation tells you what's actually happening.
The Gap Between AI Hype and AI Reality Is Where Money Hides
28 industries scored across 8 analytical dimensions. We separate the hatred from the data — and tell you which sectors benefit from the mispricing.
Free bi-weekly report. No spam. Unsubscribe anytime.
Where AI Hatred Actually Affects Stock Prices (And Where It Doesn't)
OK. We've established that public hatred doesn't stop adoption. But that doesn't mean it's irrelevant. Hatred creates regulatory drag, and regulatory drag has real, measurable effects on specific sectors and timelines.
Here's the breakdown:
High sentiment exposure (hatred affects these sectors)
| Sector | Why Hatred Matters Here | Effect |
|---|---|---|
| Consumer Tech (Meta, Google) | Consumer-facing AI products get the most backlash. AI overviews replacing search results = visible disruption. Social media AI = visible creepiness. | Stock price volatility tied to sentiment cycles. 1-3 year delay on aggressive consumer AI rollouts. Brand risk is real. |
| Media & Entertainment | SAG-AFTRA/WGA strikes set precedent. "AI replacing creative work" is the loudest backlash category. | Slows adoption in visible creative industries. Doesn't slow adoption in invisible backend (editing, VFX, script analysis). |
| Education | "Students using ChatGPT" panic. School districts banning AI tools. | Delayed institutional adoption. EdTech companies face political headwinds. But students use it anyway (Walmart pattern). |
Low sentiment exposure (hatred barely touches these)
| Sector | Why Hatred Doesn't Matter Here | Effect |
|---|---|---|
| Semiconductors (Nvidia, TSMC) | Nobody protests chip manufacturing. Infrastructure is invisible. The public doesn't know what an H100 is. | Near-zero sentiment drag. Demand driven entirely by enterprise capex, which is accelerating. |
| Enterprise Software | B2B AI deployment happens behind corporate firewalls. No consumer backlash vector. CFOs don't care about YouTube opinions. | Adoption driven by ROI math, not public sentiment. Fastest-growing segment. |
| Pharma & Biotech | "AI discovers new drugs" is one narrative the public actually likes. AlphaFold mapped virtually every known protein structure. | Positive sentiment tailwind. AI drug discovery timelines compressing from decades to years. |
| Defense & Cybersecurity | National security framing overrides AI backlash. "AI protects us" beats "AI threatens us" in this context. | Government AI spending insulated from public opinion cycles. |
| Financial Services | Banks have used "AI" (machine learning, quant models) for decades. JPMorgan, Goldman already deep. No novelty = no backlash. | Quiet AI deployment. Risk analysis, fraud detection, algorithmic trading all accelerating without public friction. |
The Regulation Question: Red Flags and Red Flag Acts
Here's where the hatred turns from noise into signal: when it drives regulation.
Public sentiment feeds political incentive. Politicians see AI hatred trending and smell votes. The regulatory scramble — Phase 3 of our universal pattern — is now fully underway in 2026.
The question for investors isn't whether regulation is coming. It is. The question is: does regulation kill the industry, or does it consolidate it?
History answers this unambiguously: it consolidates.
The Red Flag Act lesson
In 1865, Britain passed the Locomotive Act, requiring a person to walk 60 yards ahead of every automobile waving a red flag. This wasn't repealed until 1896. During those 31 years, the US and Germany built their automotive industries, and Britain — which had early pioneers in motor vehicles — lost the entire sector. Forever.
That's not an analogy. That's the single most-cited example in regulatory economics of fear-driven policy destroying domestic innovation.
What's happening right now
| Region | Regulatory Approach | Result |
|---|---|---|
| European Union | EU AI Act: most aggressive AI regulation globally. Classification tiers, compliance requirements, fines up to 7% of global revenue. | European AI companies relocating to US and UK. Brain drain accelerating. EU creating compliance moat that only mega-caps can afford. |
| United States | Executive orders, chip export controls, TSMC/Samsung reshoring incentives. Sector-specific rather than blanket regulation. | AI development concentrated in US. Export controls on China accelerating domestic semiconductor investment. $400B+ CHIPS Act pipeline. |
| China | National AI strategy to dominate by 2030. DeepSeek R1 (95% cheaper than OpenAI). No culture war about AI. | Chinese firms rushed to order $16B of Nvidia chips before tariffs. Government-backed AI labs racing on all fronts. |
The ESRB (video game ratings) was created to prevent government regulation of games. Result: it became a moat that prevented new entrants while established studios thrived. The FCC's broadcast regulations didn't kill TV — they created a three-network oligopoly that dominated for 40 years. Financial regulation after 2008 didn't kill banks — it made small banks uncompetitive and mega-banks more powerful.
AI regulation will follow the same pattern. Compliance costs create barriers to entry. Only well-funded companies can navigate complex regulatory frameworks. Incumbents with legal teams and lobbying power gain advantage. The EU AI Act isn't going to stop AI — it's going to ensure only Google, Microsoft, Meta, and a handful of others can afford to deploy it in Europe.
When evaluating AI regulation risk, ask three questions: (1) Does this regulation apply to consumer-facing or B2B products? Consumer regulation is more likely to stick because it has public support. (2) Does the company have the resources to comply? Compliance is a moat, not a threat, for large-cap AI companies. (3) Does the regulation apply domestically or does it create competitive asymmetry with China? Restrictions that only apply to US/EU companies while China operates freely are the most dangerous for investors — they don't slow AI, they shift where it happens.
The Dot-Com Crash: The One Precedent Everyone Gets Wrong
When AI skeptics want their strongest argument, they reach for the dot-com crash. "See? The internet was overhyped too. The bubble popped. AI is the same."
They're right that there are parallels. They draw exactly the wrong conclusion from them.
The dot-com crash destroyed companies that had no revenue, no business model, and no customers. Pets.com, Webvan, Kozmo.com — these were ideas funded on vibes. They deserved to die.
The crash did not destroy the internet. It did not slow adoption. It did not prove the technology was fake. What it did do was create the greatest buying opportunity in the history of public markets:
- Amazon: $107 → $7 during the crash → $3,000+ over the next two decades
- NASDAQ bottom (2002-2003): Anyone who bought the index and held has compounded at 13%+ annually for 23 years
- Peter Thiel: Invested $500K in Facebook in 2004 — during the post-crash hangover — returned $1B+
The pattern: technology is real, valuations may be premature, selling after a crash is historically the worst possible move.
If AI has a dot-com-style correction — and it might — the companies with real revenue, real adoption metrics, and real enterprise customers will be the Amazons. The companies burning cash on demos and press releases will be the Pets.coms. The distinction is knowable now, before the correction, if you look at the data instead of the sentiment.
The Track Record of AI Doom Predictions
We track specific, falsifiable AI predictions in our research database. Here's a sample of how the most prominent doom-sayers are doing:
- Eliezer Yudkowsky's "AGI by September 2025": Failed. We're past the date. He has since avoided setting new specific deadlines.
- "50% job losses" predictions: Goldman Sachs estimates 0.5% unemployment rise attributable to AI. BLS data shows no measurable AI-driven job losses at scale as of April 2026.
- Meta-pattern: Doom predictors systematically avoid specific deadlines because specificity enables falsification. Vague doom claims ("AI will destroy humanity eventually") are designed to resist disproof.
What This Means for Your Portfolio: Five Actionable Takeaways
1. Treat AI hatred as a timing variable, not a directional signal. Public backlash slows consumer-facing adoption by 1-3 years. It doesn't stop adoption. It doesn't reverse it. Every historical precedent shows the same pattern: hatred delays, then yields. Position for the delay, not the reversal.
2. Go where the hatred isn't looking. AI hatred focuses on visible consumer applications: chatbots, image generators, content creation. The bigger money — enterprise software, B2B infrastructure, chip manufacturing, pharma AI — operates with almost zero public attention. That's where the say/do gap is widest and the mispricing is deepest.
3. Watch the capex, not the commentary. When companies are spending $700 billion on AI infrastructure while the public debates whether AI is real, one side has better information. Capital expenditure decisions are made by people with access to internal adoption data, customer pipeline data, and revenue projections. YouTube commentators are making decisions based on vibes and confirmation bias.
4. Understand that regulation is a moat. The EU AI Act will not destroy AI in Europe. It will make AI in Europe affordable only for companies with massive legal and compliance budgets. That means Google, Microsoft, Meta, and a handful of others. For large-cap AI companies, regulation isn't a headwind — it's a barrier to entry that protects their market position.
5. If there's a crash, don't be the person who sold Amazon at $7. The dot-com crash destroyed bad companies and created generational wealth for anyone who bought the survivors. If AI sentiment drives a correction, the question isn't "is AI over?" (it isn't — 78% enterprise adoption doesn't reverse). The question is: "which companies have real revenue and real customers, and which ones are running on hype?" We track this distinction across 28 industries.
Stop Listening to the Hatred. Start Reading the Data.
28 industries. 8 analytical dimensions. 167 cross-industry cascade effects. 5 time horizons. 3 scenario modes. Updated quarterly. Built to cut through sentiment and show you where value is actually moving.
See the Full Industry AnalysisGet Free Industry Forecasts Every 2 Weeks
28 industries scored across 8 analytical dimensions. Boom, base, and doom scenarios. The signals that factor models can't touch.
No spam. Unsubscribe anytime.
Related reports: 600 Years of Technology Panic | The Ontological Mispricing | AI Fear vs Your Portfolio | Hype vs Reality