THE 2028 GLOBAL INTELLIGENCE BOOM

Preface

What if our AI bullishness continues to be right...and what if that’s actually bullish?

What follows is a scenario, not a prediction. This isn’t techno-utopian fan fiction. The sole intent of this piece is modeling a scenario that’s been relatively underexplored amid the growing chorus of AI labor displacement doomers.

The bear case for AI abundance has been well-articulated. Intelligent machines replace workers, workers stop spending, the economy contracts, and the financial system buckles under the weight of assumptions that no longer hold. It’s internally consistent, intellectually rigorous, and worth taking seriously.

But it depends on a series of assumptions about how economies, labor markets, and human behavior respond to technological deflation, assumptions that have been wrong every single time they’ve been tested over the past two centuries.

Hopefully, reading this leaves you more prepared for the possibility that abundant intelligence is exactly what it sounds like: abundant.

Macro Memo

The Consequences of Abundant Intelligence

Michael Bloch

February 22nd, 2026 June 30th, 2028

The unemployment rate printed 3.1% this morning. The market barely reacted. Six months ago, a number like this would have triggered a face-ripping rally. Now it’s just the baseline.

The S&P 500 crossed 12,000 last week. The Nasdaq is above 40,000. But the numbers that matter most aren’t on the ticker. Real median household purchasing power is up 18% since 2025, the largest three-year gain since the postwar boom. And the most remarkable part: nominal wage growth has been modest. The gains are almost entirely from deflation.

Two years. That’s all it took to get from “AI will destroy the economy” to an economy that no longer resembles the one any of us grew up in. This quarter’s macro memo is our attempt to reconstruct the sequence — a post-mortem on the pre-boom pessimism.

How It Started

In late 2025, agentic coding tools took a step function jump in capability.

A competent developer working with Claude Code or Codex could now replicate the core functionality of a mid-market SaaS product in weeks. Not perfectly or with every edge case handled, but well enough that the CIO reviewing a $500k annual renewal started asking the question: “what if we just built this ourselves?”

The bears saw this and modeled economic destruction. They were right about the mechanism — software margins compressed, enterprise spending was renegotiated, and the long tail of SaaS got repriced. Where they went wrong was in confusing the repricing of one sector with the collapse of the economy.

That summer, we spoke with a procurement manager at a Fortune 500. He told us about one of his budget negotiations. The SaaS vendor had expected to run the same playbook as last year: a 5% annual price increase, the standard “your team depends on us” pitch. The procurement manager negotiated a 30% discount. Then he told us what happened to the savings. His division didn’t send it to the bottom line. They hired three people for a new market they’d been wanting to enter for two years but could never justify the overhead. AI had cut their software costs enough to fund the expansion.

This was the pattern the bears missed. They tracked the dollars leaving software budgets but didn’t follow where those dollars went.

The Deflation Nobody Was Prepared For

By early 2027, LLM usage had become default. People were using AI agents who didn’t even know what an AI agent was, in the same way people who never learned what “cloud computing” was used streaming services. They thought of it the same way they thought of autocomplete or spell-check — a thing their phone just did now.

Consumer agents began handling purchasing decisions, subscription management, price comparison, and service negotiations. The bears modeled this as the destruction of the intermediation layer. They were right about the mechanism. Where they were wrong was in thinking that trillions of dollars in eliminated friction would simply vanish from the economy.

It didn’t vanish. It was returned to consumers.

The numbers were staggering. The average American household was spending roughly $8,000–12,000 per year on services whose primary value proposition was navigating complexity on the consumer’s behalf. Tax preparation. Insurance brokerage. Travel booking. Financial advice. Real estate commissions. Subscription fees for products with cheaper alternatives.

Agents compressed these costs by 40-70%. For a median household, this was the equivalent of a $4,000–7,000 annual raise — tax-free, requiring no legislation, and compounding as agents got better.

The real estate example was instructive. The bears flagged the compression of buy-side commissions from 2.5-3% to under 1% as evidence of economic destruction. But the median home price in 2027 was $420,000. A 2-point commission reduction saved the buyer $8,400. Multiply that across 5 million annual home transactions, and that’s $42 billion per year flowing to homebuyers instead of to intermediaries. Those homebuyers spent the money on furniture, renovations, moving costs, and settling into their new communities.

The same math held everywhere. When DoorDash’s take rate collapsed, restaurants kept more revenue and consumers paid lower fees. The delivery drivers, liberated from the duopoly’s 40% rake, earned more per delivery on fragmented platforms competing for their labor. This was characterized as destruction, but it was straightforwardly a transfer from rent-extractors to the people doing the actual work and the people actually eating the food.

MASTERCARD Q1 2027: NET REVENUES +6% Y/Y; PURCHASE VOLUME GROWTH SLOWS TO +3.4% Y/Y; MANAGEMENT NOTES “AGENT-LED PRICE OPTIMIZATION” AND “PRESSURE IN DISCRETIONARY CATEGORIES” | Bloomberg, April 29, 2027

This print terrified the market. Mastercard fell 9%. The interpretation was that agentic commerce was destroying the payments infrastructure.

What actually happened over the following twelve months was more interesting. Total transaction volume accelerated even as per-transaction interchange compressed. It turned out that when every purchase is optimized, people don’t stop buying things. They buy more things, just at better prices, through more efficient channels. Visa and Mastercard’s revenue eventually re-accelerated on volume growth that more than offset the rate compression. The stock recovered and made new highs by Q2 2028.

The intermediation economy didn’t collapse. It got competed down to its actual value and the surplus went to everyone else.

The SaaS Repricing Was a Feature, Not a Bug

ServiceNow’s Q3 2026 report was the print that convinced people the software economy was in structural decline.

SERVICENOW NET NEW ACV GROWTH DECELERATES TO 14% FROM 23%; ANNOUNCES 15% WORKFORCE REDUCTION AND ‘STRUCTURAL EFFICIENCY PROGRAM’; SHARES FALL 18% | Bloomberg, October 2026

The narrative was that the SaaS model was broken. And for the SaaS model as priced in October 2025, it was. But the bears made a critical error in scope. They saw a sector repricing and projected an economic collapse.

Enterprise software spending in the United States was roughly $500 billion annually in 2025. It was bloated, overpriced, and riddled with shelfware. Companies were spending millions on tools that 30% of their employees actually used. The AI-driven renegotiation and consolidation cycle cut that bill significantly. Where the bears lost the thread was in their implicit assumption that reduced software spending was the same as reduced economic activity.

Software spending is an input. It is a cost that businesses pay in order to generate revenue. When the cost of the input drops, the business has more resources to deploy toward the output — expansion, R&D, new hires in revenue-generating roles, capex in new markets. This is, quite literally, how productivity growth has always translated into economic growth.

The disruption was real. The long tail of SaaS was decimated. Monday.com, Asana, Zapier — the tools that could be replicated by a competent developer with agentic coding tools in a weekend — saw their revenues cut in half. But the total economic activity supported by those same businesses increased, because they were now spending less on overhead and more on growth.

New software companies emerged, too. Built on AI-native architectures with radically lower cost structures, they offered capabilities that didn’t exist in the prior generation. Enterprise software spending didn’t shrink for long; it got reallocated. By Q3 2027, total enterprise tech spending had recovered to 2025 levels but the composition was unrecognizable.

Half the budget went to AI-native tools that were a quarter the price and four times as capable.

The Labor Market That Refused to Break

This was the crux of the bear case: white-collar workers would be structurally displaced, consumer spending would collapse, and a negative feedback loop would accelerate until the economy cratered.

Here is what actually happened.

U.S. JOLTS: JOB OPENINGS FALL BELOW 5.5M; UNEMPLOYED-TO-OPENINGS RATIO CLIMBS TO ~1.7, HIGHEST SINCE AUG 2020 | Bloomberg, Oct 2026

The October 2026 JOLTS print was genuinely concerning. White-collar job openings were collapsing. The bears were right about this data point, and we won’t pretend otherwise. Q4 2026 and Q1 2027 were frightening. The labor market did soften meaningfully in professional services, software, consulting, and financial services. Unemployment touched 5.8% in February 2027, the highest since the pandemic.

But the bears made an assumption that turned out to be wrong: that displaced white-collar workers would remain displaced.

What happened instead was faster and messier than anyone predicted. The same AI tools that eliminated certain roles also made it dramatically cheaper to start things. The cost of launching a business — software, legal, accounting, marketing, design — fell by 70-80% in eighteen months.

New business formation exploded. The Census Bureau reported 7.2 million new business applications in 2027, shattering the prior record of 5.5 million set in 2021. The majority were filed by recently displaced professionals who now had the skills, the tools, and the extremely low overhead to build something on their own.

A friend of ours was a senior product manager at Salesforce in 2025. She lost her job in the third round of layoffs. Rather than driving for Uber, she spent three weeks building a niche AI-powered compliance tool for small healthcare practices, a market her former employer would never have served because the deal sizes were too small to justify the sales team. She had 200 paying customers within four months. Her revenue exceeded her former salary within eight.

Was she an exception? In 2026, yes. By mid-2027, she was the norm. The collapse in the cost of building and running a business meant that the “minimum viable ambition” dropped to nearly zero. You didn’t need funding, a team, or an office. You needed a laptop, a credit card for inference costs, and a problem you understood.

The economy began creating jobs that didn’t have titles yet. “AI-assisted” became a prefix for virtually every professional services category. AI-assisted architecture. AI-assisted physical therapy program design. AI-assisted supply chain consulting. These weren’t the “prompt engineer” meme jobs that people joked about in 2024. They were substantive roles where a human with domain expertise and AI tools could deliver the output previously requiring a team.

The labor market didn’t stabilize because the old jobs came back. They didn’t. It stabilized because new jobs emerged faster than old ones disappeared. Unemployment peaked at 5.8% in February 2027 and began declining by summer as the new business formation wave translated into hiring.

The historical pattern held, but with a twist. In prior technology cycles, the new jobs took a generation to materialize. The internet disrupted travel agents in the late 1990s, and the new digital economy jobs took 10-15 years to fully absorb the displaced workers. This time, the same technology that caused the displacement also compressed the creation cycle. AI didn’t just destroy jobs faster; it created the replacement jobs faster too.

When Cheaper Meant Richer

The bears’ most compelling argument was “Ghost GDP” — the idea that output could surge while the real economy withered because the gains never reached consumers. This framework assumed that the relationship between output and consumption depended on wages.

It doesn’t. It depends on purchasing power. And purchasing power is a function of both wages and prices.

This distinction turned out to be the entire ballgame.

Even in Q1 2027, when the labor market was at its weakest, something strange was happening in the consumption data. Retail spending volumes were rising even as nominal wages softened. Economists were confused until the price data caught up: AI-driven deflation in services was running at 8-12% annualized. For the first time since the industrial revolution, services were getting cheaper, not just goods.

A household earning $100,000 in 2025 needed $100,000 in 2027 to maintain the same standard of living, in a normal inflation regime. But in the actual 2027 economy, that same standard of living cost $85,000 because healthcare navigation, legal services, tax preparation, insurance, financial planning, and dozens of other service categories had been radically deflated by AI.

Technology-driven deflation is not the same as demand-driven deflation. When prices fall because nobody is buying, it’s a death spiral. When prices fall because the cost of production collapsed, it’s a living standard boom. The 20th century is full of examples: the falling price of automobiles, televisions, air travel, computing, mobile phones. Each time, the deflation coincided with more economic activity, not less, because affordability unlocked demand from populations that were previously priced out.

AI did this to the entire services economy simultaneously. And services are 70% of consumer spending.

The political class initially struggled to communicate this. “Real wages are up” didn’t resonate when nominal paychecks were flat. But household surveys told the story clearly: by Q3 2027, consumer confidence had rebounded to pre-2020 levels, driven not by income growth but by the tangible experience of everything being cheaper.

The “intelligence premium” didn’t unwind. The intelligence tax did. For decades, Americans had paid enormous markups for services that were expensive because human intelligence was scarce. When intelligence became abundant, the tax disappeared. The 30-year-old who couldn’t afford a financial advisor in 2025 had one in 2027 — it just cost $20/month instead of $5,000/year. The small business owner who couldn’t afford a lawyer had access to competent legal analysis for the cost of a Netflix subscription.

This wasn’t Ghost GDP. This was the most progressive economic event in modern American history, achieved without a single redistributive policy. AI deflation was a de facto transfer from the owners of scarce intelligence to the consumers of it.

The Private Credit Scare That Didn’t Metastasize

We’re not going to pretend the financial system sailed through without turbulence. It didn’t. The bears were right that PE-backed software deals were overvalued and that some would default.

ZENDESK MISSES DEBT COVENANTS AS AI-DRIVEN CUSTOMER SERVICE AUTOMATION ERODES ARR; $5B DIRECT LENDING FACILITY MARKED TO 58 CENTS | Financial Times, September 2027

Zendesk was real. So were the half-dozen other LBOs of 2021-2022 vintage SaaS companies that entered restructuring. The $5 billion in direct lending losses were painful for the sponsors and their LPs.

But here’s what didn’t happen: contagion.

The bear case required the software defaults to cascade through the insurance-linked balance sheets and into the broader credit system. The mechanism was supposed to be: defaults impair private credit portfolios → insurance regulators force insurers to raise capital or sell assets → forced selling depresses prices → more defaults → the spiral accelerates into the mortgage market.

Each step in that chain was individually plausible. But the chain broke at the second link.

The software defaults were concentrated in a narrow vintage of deals (2021-2023 LBOs) in a specific sector (horizontal SaaS) with a specific vulnerability (seat-based pricing into companies cutting headcount). The total exposure was roughly $80-100 billion across the private credit ecosystem. Against $2.5 trillion in total private credit AUM, this was a 3-4% loss rate. Unpleasant, but well within the range that “permanent capital” vehicles were designed to absorb.

The insurers took marks. Athene wrote down a portfolio of software-backed loans. It was a bad quarter. But the broader portfolio — real estate debt, infrastructure lending, asset-backed finance — was performing fine, in some cases better than projected, because AI-driven productivity was improving cash flows at the operating companies backing those loans.

The regulatory response was proportionate. The NAIC tightened guidelines on concentration limits for private-rated credit in insurance general accounts, but stopped short of the forced deleveraging the bears had predicted. The regulators were stern, not panicked, because the system-wide capital ratios were adequate.

NEW YORK, IOWA STATE REGULATORS MOVE TO TIGHTEN CAPITAL TREATMENT FOR CERTAIN PRIVATELY RATED CREDIT HELD BY LIFE INSURERS | Reuters, Nov 2027

This headline spooked the market for about a week. Apollo fell 15%. Then investors actually read the guidance, realized it was targeted at concentration limits rather than across-the-board capital charges, and the stocks recovered within the month.

The lesson was a familiar one: financial systems that aren’t leveraged 30:1 can absorb losses. Private credit’s locked-up capital structure, the same feature the bears argued was hiding losses, turned out to be genuinely stabilizing. There was no forced selling, because there was no mechanism to force sales.

The Mortgage Market Held

This was the domino the bears needed to fall for their scenario to reach crisis proportions. It didn’t.

The logic was clear enough: white-collar income impairment → mortgage stress → housing price declines → negative wealth effect → deeper recession. Each link was reasonable in isolation.

But the premise was wrong. White-collar incomes were not structurally impaired. They were transitionally disrupted — a critical distinction.

The labor market turbulence of late 2026 and early 2027 was real, but it lasted about nine months, not nine years. By Q3 2027, the new business formation wave and the expansion of AI-augmented professional roles had restabilized employment for the white-collar cohort. Incomes were different — more variable, more entrepreneurial, less corporate — but aggregate household income in the top two quintiles had recovered to 2025 levels by Q4 2027 and exceeded them by Q1 2028.

Meanwhile, the purchasing power math was working in homeowners’ favor. A household whose income dropped 10% but whose non-housing expenses dropped 20% was better positioned to make mortgage payments, not worse. Debt-to-income ratios actually improved for many households despite nominal income softness, because the denominator in their monthly budget (everything besides the mortgage) had shrunk.

Delinquencies ticked up in exactly the ZIP codes the bears identified — San Francisco, Seattle, Austin, Manhattan — and peaked at levels that were mildly concerning but never approached systemic. The 30-day delinquency rate for prime mortgages reached 2.1% in Q1 2028, up from 1.2% in 2025, well below the 5%+ levels that would signal structural impairment. By the time we’re writing this, it’s already declining.

ZILLOW HOME VALUE INDEX: NATIONAL HOME PRICES +2.3% YOY; SAN FRANCISCO -3%, AUSTIN -2%, SEATTLE FLAT; SUNBELT METROS AND MID-SIZE CITIES LEAD GAINS | Zillow, June 2028

The geographic dispersion was notable. The expensive coastal metros that had concentrated white-collar tech employment softened modestly. But mid-size cities and Sunbelt metros — the beneficiaries of remote work and AI-enabled geographic arbitrage — appreciated. The national index was positive. The $13 trillion mortgage market was not in distress.

The bears assumed that the mortgage market needed the same jobs in the same places at the same salaries to remain solvent. What it actually needed was for households to be able to make their payments. The combination of labor market rebalancing and services deflation made that possible even through the transition.

The Government Had Less to Do Than Everyone Feared

Perhaps the most surprising development of the past two years was how little the federal government needed to do.

This is not because the government was competent (it wasn’t; the political response was as slow and partisan as the bears predicted). It’s because the economic adjustment happened faster than the legislative process. By the time Congress had formed a bipartisan committee to study AI displacement, the unemployment rate was already declining.

Federal receipts did dip in Q2-Q3 2027, running about 6% below CBO baseline projections. But the recovery in employment, combined with surging corporate tax receipts from AI-driven margin expansion and capital gains tax revenue from the equity rally, restored the fiscal trajectory by Q1 2028.

The proposals for compute taxes, AI royalty funds, and expanded transfer programs remain on the table and continue to be debated. Some version will likely pass, and probably should, as insurance against future disruption cycles. But the urgency has dissipated. The economy didn’t need an emergency rescue because the emergency was shorter than the policy response time.

This isn’t a libertarian parable. Public investment in retraining, community college modernization, and broadband infrastructure played a supporting role. But the primary adjustment mechanism was the market. Companies redeployed workers. Entrepreneurs started businesses. Consumers redirected savings. The feedback loop was positive, and it didn’t need the government to engineer it.

The Abundance Boom

For the entirety of modern economic history, human intelligence has been the scarce input. Every institution in our economy — from the labor market to the mortgage market to the tax code — was designed for a world in which that assumption held.

We are now experiencing something unprecedented: the democratization of that input. Machine intelligence is now a competent and rapidly improving complement to human intelligence across a growing range of tasks. The financial system, initially terrified of what this meant, has repriced — upward.

But the story that matters isn’t in the S&P. It’s in the data that doesn’t make the Bloomberg terminal.

A 22-year-old in rural Arkansas has access to the same quality of legal, financial, and medical guidance that was previously available only to residents of major metros who could afford $500/hour professionals. A small business owner in Lagos can build and ship a software product that competes with tools built by 200-person teams in San Francisco. A retiree on a fixed income has seen her purchasing power rise 15% without any change to her Social Security check, because the cost of the services she depends on has collapsed.

The intelligence premium didn’t unwind. It was democratized. When intelligence was scarce, it accrued to the few who could afford it. When it became abundant, the benefits spread to everyone who consumed it. The owners of compute saw enormous returns — the Mag 7 are collectively worth more than the GDP of most nations. But unlike prior cycles of wealth concentration, the gains did not come at the expense of everyone else. They came alongside the broadest improvement in real living standards in a generation.

This was the historical pattern all along. The Carnegies and Rockefellers got obscenely rich during the industrial revolution. So did everyone else, because steel and oil made everything cheaper. The Bezoses and Zuckerbergs got obscenely rich during the internet revolution. So did everyone else, because information and commerce became nearly free. The pattern was: the builders of the infrastructure capture an outsized share of the gains, but the gains are large enough that the remainder still transforms living standards.

AI is following the same pattern at an accelerated pace.

What the Bears Got Right

The bears were not wrong about everything. They were right that the transition would be painful. They were right that SaaS was overvalued. They were right that intermediation businesses built on friction were in trouble. They were right that PE-backed software was a ticking time bomb. They were right that the labor market would go through a genuine disruption.

Where they went wrong was in their model of human behavior.

They assumed companies would uniformly fire workers rather than redeploy them. They assumed displaced workers would stay displaced rather than adapt. They assumed that reduced spending on one category meant reduced spending overall, rather than reallocation. They assumed that deflation is always contractionary, when technology-driven deflation has historically been the most powerful engine of prosperity. They assumed that the feedback loop ran in only one direction.

The deepest error was in treating the economy as a closed system where AI was a zero-sum substitution for human labor. It’s not. It’s an expansion of the total productive capacity of the system. The output was not “ghost GDP” that never circulated. It was real output that made everything cheaper, which made everyone richer in the ways that actually matter.

The Canary Is Singing

But you’re not reading this in June 2028. You’re reading it in February 2026.

The S&P is near all-time highs. The negative feedback loops have not begun — nor have the positive ones we’ve described. We are certain some of these scenarios won’t materialize exactly as written. We’re equally certain that machine intelligence will continue to accelerate.

As investors, the question is not whether AI will disrupt the economy. It will. The question is whether you believe that the most powerful deflationary force in human history will make people poorer or richer. Every prior technology of this magnitude has produced the latter, and we see no structural reason why this time is different.

The two-hundred-year track record of technology-driven deflation improving living standards is not a guarantee. But it’s the strongest prior we have, and betting against it has been the wrong trade every single time.

The canary isn’t just alive. It’s singing.