This is a scenario. Not a prediction, not abundance fan-fiction. Citrini Research wrote a careful bear case for what AI does to the economy, and the bull case has mostly been left to hype merchants. That's the gap I want to fill. Read this and you'll at least be more attuned to the right-tail outcomes when the economy starts getting genuinely weird.
What follows is the December 2028 Macro Memo, detailing the progression and emergence of the Abundance Economy.
March 9th, 2026 → December 31st, 2028
The S&P 500 closed the year at 7,840. Down 12% from its October 2026 highs, up 23% from the March 2028 lows. The correction scared everyone. It bankrupted no one systemically. And it gave way to a rally built on economics nobody had frameworks for. New business applications with planned wages hit 6.8 million in 2028, nearly four times the previous record. Fertility rate ticked up for the first time since 2007. Obesity rates fell 14 percentage points in four years.
Two years. That's all it took to go from "AI will destroy jobs" to "AI created more businesses than any technology in history." Different jobs, created by different people, in a different shape of economy. Here's the sequence, and why the bears got the sign wrong.
They Watched the Numerator. We Watched the Denominator.
Citrini's piece documents the layoffs with precision. White-collar employment contracted 18% from 2024 peaks. ServiceNow, Zendesk, the cascading SaaS compression, the "daisy chain of correlated bets on white-collar productivity growth." All of that happened.
What the bears missed: the same agentic capability that made existing companies vulnerable made new companies trivially cheap to start. The CIO reviewing a $500k annual renewal and asking "what if we just built this ourselves?" is the same kind of person as the former product manager launching three products in six months. Same capability, different framing.
Job destruction and job creation came from the same place: identical agentic tools, hitting the same workers, generating churn that looked apocalyptic if you only counted the layoffs and miraculous if you counted business formations. Labor force participation for prime-age workers hit a high not seen since the late nineties. People kept working. Most of them just stopped working for the companies that used to employ them.
"Ghost GDP" Found Bodies
Citrini introduced "Ghost GDP": output that shows up in national accounts but never circulates through the real economy. A single GPU cluster in North Dakota generating output previously attributed to ten thousand white-collar workers in midtown Manhattan. The bear concern was that machines don't spend on discretionary goods, so velocity of money flatlines as labor's share of income collapses.
Ghost GDP found bodies through deflation. When services become dramatically cheaper to produce, the same nominal income buys more of them. Legal prep, tax preparation, insurance brokerage, financial advisory, medical second opinions, and educational tutoring all fell between half and four-fifths in price. A family that used to spend eight thousand dollars a year on professional services now spends about two thousand for equivalent or better outcomes. Nominal income dropped. Real consumption capacity barely budged.
The bears confused supply-driven deflation with demand-driven deflation. Cheaper to produce is what rising living standards looks like in the data. We went from spending half of household income on food in 1900 to under a tenth today through that same mechanism. Services were the last bastion of Baumol's cost disease, and they finally joined the long deflation that took a century to play out for goods.
The Intermediation Collapse Was Good
Citrini documented the collapse of intermediation: travel booking, insurance renewals, financial advice, tax prep, real estate commissions. "Agent on agent violence." Billions in revenue extracted by exploiting quirks of human psychology, with those moats dissolving in real time. The framing is what differs. Citrini sees catastrophe. I see liberation.
The "giant rent-extraction layer built on top of human limitations" was never real infrastructure. It was a tax on being human, paid in trillions of dollars because nobody had the time or attention to shop their insurance renewals. When agents started doing the shopping automatically, the extraction collapsed. The spending didn't. It moved into actual goods and services.
DoorDash's margin compression is bad for DoorDash shareholders. It's good for delivery drivers, who now capture the share of the fee that used to disappear into the middle. The white-collar workers who "ended up as delivery drivers" earn more per hour than they would have two years ago, because the platforms taking half their labor value got disrupted. "Friction was going to zero." That's correct. The friction was a tax that the bears mistook for a feature.
The A2A Economy That Didn't Show Up
The machine-to-machine economy never took over. In mid-2026, the hot thesis was A2A: agents transacting with agents, humans increasingly irrelevant to the flow of commerce. Some of it materialized. Stablecoin payments did route around interchange, agentic shopping killed habitual brand loyalty, and the intermediation layer collapsed. The full vision (machines primarily serving machines, humans as afterthought) never showed up. It also stopped mattering, because the human side of the economy got more interesting at the same time.
Human needs exploded. When AI handled the drudgery, people didn't sit idle. They discovered they wanted things they'd never had time to want before. Custom furniture. Artisanal food. Personal training. Music lessons. Therapy. Travel. Experiences. Education for its own sake. Time with children.
Velocity of money didn't flatline. Humans freed from drudgery turned out to have infinite appetite for things that can't be digitized: presence, craft, attention. Agentic commerce was supposed to automate consumption. What it did was free up time and disposable income for the kind of consumption that requires presence to enjoy. The A2A economy ended up smaller than projected. The human side of the ledger blew past its forecasts.
The Negative Feedback Loop Had Brakes
Citrini described a "negative feedback loop with no natural brake": AI improves, companies lay off workers, savings flow into AI capability, AI improves. Each individual company's response was rational. The collective result, in Citrini's reading, was catastrophic. Brakes existed. They just weren't where economists were looking.
Every displaced worker who started a business absorbed the shock rather than transmitting it. Falling service costs meant displaced workers needed less income to maintain living standards. Geographic mobility let workers leave coastal markets where rents were strangling them and land in secondary cities where the cost structure was a third of what they'd left. Civic participation gave people meaning and connection that labor-market metrics can't measure. The creative economy turned the long tail into a livable wage. The peptide revolution quietly compounded into higher productivity and lower social costs. None of these mechanisms show up cleanly in economic models. All of them are measurable in the actual data.
The Walzers
It started as a prank. In February 2027, Riley Walz, the developer known for his satirical AI projects, got fed up waiting for a permit to install solar panels on his mother's house in Los Angeles. The application had been stuck in review for nine months. He'd called the planning department forty-seven times. Nothing moved.
So he did what Riley Walz does: he built something unhinged. A full sybil attack of hundreds of AI personas, each with backstories, each with different communication styles, each targeting a different member of Mayor Bass's staff and the planning department hierarchy. They called. They emailed. They filed public records requests. They showed up to every public comment period. They coordinated testimony that turned the permit backlog into a recurring news story. His mother's permit was approved in eleven days. Riley posted the code. The internet did the rest.
Within weeks, people were "walzing" their own local governments. The verb stuck. "We walzed the Austin planning board." "Someone walzed the SF fire marshal." "They got walzed so hard they just started approving everything." The legal status was ambiguous. The agents weren't filing false information; they were asking real questions, citing real regulations, and demanding accountability the law already required. Persistently. At scale. City attorneys struggled to articulate what law was being broken when an AI sent four hundred polite follow-up emails about a permit status.
What started as chaos became the Rebuilder movement. The first wave was adversarial: bureaucracies overwhelmed into compliance. The second wave was collaborative. People realized the same swarms could help officials work through their own systems faster. Some planning departments quietly started using forked versions of Riley's code to process their own backlogs. Permit timelines collapsed across cities. Affordable housing approvals that once took years began closing in months. People with time, tools, and local knowledge systematically disarming civic bureaucracy that had calcified over decades. Some called it civil disobedience. Others called it the most effective government reform movement since the Progressive Era. Riley Walz got a cease-and-desist from the City of Los Angeles. He framed it and hung it in his living room. His mother's solar panels work great.
The Hardware Renaissance
AI collapsed software costs, which paradoxically increased the relative value of things that can't be digitized. Physical objects, tactile experiences, things you can touch. Software abundance made hardware differentiation matter for the first time in a decade.
CNC machining communities exploded. Small-batch manufacturing became viable once AI design tools collapsed engineering overhead. You can prototype a physical product in a weekend for a few hundred dollars. If it works, produce a thousand units without going through Shenzhen. There's an aesthetic dimension to it: an American industrial revival, exposed materials, visible joinery, functional design with craft integrity. After two decades of disposable minimalism, people want things that last. Manufacturing employment ticked up for the first time in four decades, driven by collapsed overhead rather than cheaper labor.
The Step Changes
Some developments resist trend extrapolation. Johns Hopkins demonstrated lab-grown neural tissue performing pattern recognition at a tiny fraction of silicon's power consumption; commercial organoid co-processors began shipping by mid-2028. Helion hit fusion breakeven in April; Commonwealth followed. Neither is commercially viable yet, but the psychological barrier fell and private investment quadrupled. Google's Willow had demonstrated below-threshold quantum error correction in late 2025; by late 2028, systems with hundreds of logical qubits ran continuously. The technological frontier advanced across energy, computation, and biological engineering simultaneously. That kind of convergence is rare in history. It changes what's conceivable.
The Ambitious Works
In September 2028, San Francisco approved the Liberty project: a 150-foot bronze statue of Justice on Alcatraz Island. Scales in one hand, broken chain in the other. Facing west. Design selected through public competition: four thousand entries. Winner: a 28-year-old sculptor from Detroit, no prior public commissions. Funding came through municipal bonds, private philanthropy, and a crowdfunding round that closed in 72 hours. Museum of migration on the Texas border. Climate memorial in Miami. Penn Station reconstruction in New York.
The pessimistic frame imagined abundance leading to stagnation. What showed up was ambition. People with material security at baseline didn't sink into anomie. They built things, and they bet on beauty and meaning while doing it. By late 2028, the pipeline of ambitious public works was larger than it had been since the WPA.
The 1,000 True Fans Economy
More professional musicians earned a living wage in 2028 than at any point in recorded history. Same for visual artists, writers, and independent game developers. The mechanism is Kevin Kelly's "1,000 true fans," articulated in 2008 and finally real at scale. AI replaced middlemen rather than creators. Distribution costs collapsed. Discovery restructured around genuine engagement instead of attention hijacking.
A musician with two thousand fans paying fifteen dollars a month earns a comfortable salary. No label, no touring, no algorithm dependence. The arithmetic was always possible. The infrastructure that makes it work without friction is what finally exists. The fear was that AI slop would flood the zone. People could tell the difference, and they paid for the authentic side. The market bifurcated into a free commodity layer and a paid human-creative layer. Both grew. The human layer grew faster. Winner-take-all dynamics inverted, and the long tail learned to earn a living.
The Peptide Revolution
Semaglutide and its successors reduced US obesity rates by 14 percentage points between 2024 and 2028. One in seven formerly obese Americans is no longer obese. Downstream cascades rippled through healthcare economics: type 2 diabetes incidence, cardiovascular events, joint replacements, and sleep apnea diagnoses all dropped. Healthcare cost curves bent for the first time in decades, driven by upstream intervention rather than rationing.
AI-assisted molecular modeling enabled a combinatorial explosion of peptide therapeutics. The next generation targets addiction, neurodegeneration, and autoimmune conditions previously untreatable. Rebuilder clusters pointed agent swarms at healthcare bureaucracy, working through prior authorization mazes, documenting denial patterns, and coordinating patient advocacy at scale. More people got access to effective treatments than the system was designed to provide. Structural healthcare reform was forced into existence rather than legislated.
The Fertility Surprise
US total fertility rate ticked up in 2028 for the first time since 2007. The causes were overdetermined: declining housing costs in secondary cities, flexible work arrangements, better fertility interventions, and a cultural optimism that's hard to measure but obvious if you spend time with people. People are having children because they believe the future will be good. That belief, once lost, is hard to recover. Its return is one of the strongest macro signals on the dashboard.
The Oracles
Prediction markets became infrastructure. By late 2027, the accuracy debate was settled. The open question was what to do with probability estimates that kept being right. When Polymarket and its competitors consistently outperformed polls, pundits, and expert panels, the information started flowing upstream into actual decision-making. Insurance companies began pricing policies off market odds rather than actuarial tables. Venture funds used market-implied probabilities to value startups. Governments quietly consulted prediction markets before major policy announcements.
The top forecasters became minor celebrities. A social network called Oracles emerged: part trading floor, part intellectual arena. The best predictors built followings in the hundreds of thousands. Their track records were public, immutable, and constantly updated. You could see exactly how often they'd been right, on exactly which kinds of questions. For the first time, the future had structural odds in a usable form: calibrated probabilities, posted by people with skin in the game and verified track records, available to anyone who wanted to look. The old coordination signals (vibes, narratives, op-eds) suddenly had competition.
It wasn't without scandal. In August 2028, a prominent Oracle user (handle "CassandraDAO," top fifty on the leaderboard) was caught taking money to post misleading analysis on an energy futures question. The mispricing lasted about eleven hours before other forecasters noticed the anomaly, piled into the correct position, and traced the information back to a suspicious payment. CassandraDAO was blacklisted within forty-eight hours. The entire prediction history got flagged. A reputation built over three years evaporated overnight. Speed of correction was the point. Accuracy was the only currency on Oracles, so a faked track record stayed faked for hours rather than weeks, and the people who tried it ended up obvious.
On the Mortgage Question
Citrini raised the mortgage question: are prime mortgages money good? The thirteen-trillion-dollar market is built on assumptions about income stability that AI disruption challenged. The observed answer in 2028: stress stayed concentrated. San Francisco, Seattle, Manhattan, and Austin took most of it, since they're the markets heavily exposed to tech employment. National delinquency rose but stayed well below 2008.
The mechanism preventing contagion was geographic mobility plus service deflation. Displaced workers who couldn't afford coastal mortgages relocated to places where the same monthly payment bought equivalent housing. Boise, Austin's suburbs, Tampa, and Charlotte absorbed tech workers at dramatically lower cost structures. What looked like a mortgage crisis turned out to be coastal premium deflation. Some prime mortgages didn't pay off. The system held because borrowers moved instead of defaulting, and the destination markets had solvent buyers.
On the Policy Question
Citrini described government paralysis: policy moving at the pace of ideology, not reality. The slowness was real. The slowness was also not the binding constraint. Market-level adjustments (entrepreneurial explosion, service deflation, geographic mobility, civic rebuilding) happened faster than any policy response could have. The Transition Economy Act passed in Q3 2028, providing transfers to displaced workers funded partly by AI inference compute taxation. By the time it passed, the acute crisis had already resolved through organic mechanisms. The Act functioned more as consolidation than rescue. The economy restructured faster than government could intervene.
The Repricing
Citrini was right that human intelligence is being repriced. The premium on cognitive labor has compressed. Work that used to require expensive professionals is now commoditized. The disagreement is over what this means. The bear reading is that repricing is catastrophic. The case I want to make is that it's a liberation, and the rest of this section is why.
For most of human history, survival required selling time. You worked or you starved. The arrangement was coercive by construction. "Dignity of work" was partly cope, with people making meaning out of necessity. When AI collapses the price of cognitive labor, the value doesn't disappear. It moves down the income distribution. Bespoke legal work becomes affordable for small businesses. Real financial planning reaches median households. The advice that used to require a $1,500/hr consultant reaches people who would never have written that check.
The former consultants, lawyers, and advisors aren't useless. They're doing different work: starting companies, building products, participating in civic life, creating art, raising children. Most earn less in dollar terms. Most also report higher life satisfaction. We're running a natural experiment in decoupling income from wellbeing. The early results are encouraging.
What's Actually Happening
Here's the December 2028 picture. The entrepreneurial explosion, the service deflation, the Rebuilders, the hardware renaissance, the ambitious public works, the creative flourishing, the peptide revolution, the step changes in technology, the fertility uptick. All of it is happening, and none of it is hand-waved.
These are specific, observable developments. Together they add up to a civilization learning to use new tools well. The pessimistic frame got the magnitude of change roughly right. The sign was where it went wrong. What we ended up with was renaissance.
❦
But you're not reading this in December 2028. You're reading this in March 2026. The trajectory is not guaranteed. The abundance thesis requires building: building businesses, building civic infrastructure, building health interventions, building creative economies. Passively waiting for the positive scenario would guarantee the negative one.
The good news is that building has never been cheaper. The tools are here, the capabilities are real, and the only open question is whether anyone uses them. The canary is alive. The future is bright.