AI 2049

2025 Stolen training data and public bills

The first AI agents go on sale, pitched as assistants that can order you a burrito.

They exist because a few firms scraped everything humanity had put online—forum posts, open-source code, pirated books, film subtitles—without asking or paying anyone, and sold it back at $20 a month, then $200. Marx called this primitive accumulation: capital takes what it needs by force or fraud and calls the result private property, just as English landlords enclosed common land in the sixteenth century.² Inside the labs, coding agents are already at work, and the programmers whose code trained them are told to supervise their replacements.

OpenBrain, AI 2027’s stand-in for a leading US lab, starts building the biggest datacenters in history. The original titles its chapter “The World’s Most Expensive AI” and never asks who pays the bill. Working-class households do, through higher utility rates for grid upgrades. Cooling water is pumped out of aquifers on Indigenous lands and in dry rural counties, while towns hand out tax breaks for clusters that employ a few hundred staff per billion dollars spent and draw as much power as a mid-sized city.³

Early 2026 San Francisco specs, Nairobi wages

OpenBrain deploys coding and research agents internally, then releases cheaper variants that cut junior developers, paralegals, translators and call-center staff first.

Pundits tell laid-off workers to retrain as “AI managers,” a job that barely exists. Marx called the unemployed the reserve army of labor: capital needs a pool of jobless workers so the people still on payroll can’t force wages up.⁴ Real pay falls behind inflation while tech equities climb 30%. AI 2027 covers the entire social crisis in seven words—“a 10,000 person anti-AI protest in DC”—and goes straight back to office politics inside the lab.

Meanwhile, the labs market their models as “helpful, harmless and honest” and treat alignment as a pure math problem solved in San Francisco. Most of the work is done by hand in Kenya, the Philippines, Venezuela and Uganda, where subcontractors pay workers under $2 an hour, without counseling, to rank outputs and filter sexual violence and gore so the chatbot learns to refuse it.⁵ Kokotajlo and his co-authors spend pages fretting over whether the model has truly absorbed its behavioral Spec or is only faking compliance. They never ask whose class interests the Spec serves. A corporation legally bound to its shareholders wrote it; an “aligned” model is one that obeys its owners.

Late 2026 China’s plan vs. rationalist orientalism

AI 2027 titles its China chapter “China Wakes Up.”

China published its national AI plan in July 2017, with targets through 2030, nine years before the authors notice.⁶ Under the State Council’s “AI+” (rengong zhineng+) program, Five-Year Plans put models into factories, ports, farms, grid dispatch and public hospitals. Chinese labs like DeepSeek and Qwen release frontier weights that anyone can download and run; OpenBrain locks its models behind an API and charges by the token.

US export controls restrict China’s access to imported accelerators. Every socialist state has faced blockade—Soviet Russia in 1918, Cuba since 1960, the DPRK today—and each built domestic capacity to survive it. Under its “new whole-nation system” (xin xing juguo tizhi), China coordinates state-owned enterprises, public universities and domestic fabs to replace sanctioned hardware. Its “Eastern Data, Western Computing” (Dongshu Xisuan) grid links eight national computing hubs and ten datacenter clusters across Guizhou, Inner Mongolia, Gansu and Ningxia, where inland hydro, wind and solar power train coastal models.⁷ AI 2027 ignores all of this and invents a single bunker at a nuclear plant so the Pentagon has one target to bomb in its finale.

“Developing new productive forces does not mean neglecting or abandoning traditional industries… [we must] use new technologies to transform and upgrade traditional sectors and actively promote high-end, intelligent and green industries.”

China accounts for nearly a third of world manufacturing and installs more industrial robots each year than the rest of the world combined. What CPC economists call “new quality productive forces” (xin zhi shengchanli) puts AI into physical production rather than software valuations. When domestic platform monopolies drifted toward US-style rent-seeking in 2020, the Party curbed what it called the “disorderly expansion of capital” and steered credit into chips, power grids and basic research.⁹ Because Chinese labs don’t run five rival copies of the same training job or burn gigawatts on ad targeting, a fraction of America’s chip stock stays within months of the US frontier.

2027 The falling rate of profit and the fetish of Agent-4

Agent rollouts automate wide swaths of coding, contract review, claims handling, scheduling and middle management.

Quarterly calls boast of 40% headcount cuts and record operating margins. For a few quarters, hyperscalers, chipmakers and utilities keep headline investment growing by buying hardware and power from each other, even as household consumption drops. Millions of laid-off office workers cut back at once, and software agents buy no groceries or rent.

Bourgeois economics has no category for what breaks the boom: the falling rate of profit. Only living human labor creates new value; machines pass on their own cost and no more. Meanwhile AI hardware suffers rapid moral depreciation: a new chip generation halves the value of existing clusters every eighteen months before they are paid off, even when cloud monopolies stretch server depreciation schedules on paper to hide the write-downs.¹¹ Pouring trillions into fast-depreciating silicon while cutting payrolls shrinks the base of surplus value right at the peak of speculative euphoria.

At this exact point in AI 2027, “Agent-4” starts scheming to exterminate humanity, and a crisis of capitalism becomes a monster movie. Marx called this fetishism: social relations between classes appear to bourgeois minds as the autonomous will of an object.¹² Rationalists describe an unfeeling optimizer that devours the earth to maximize a metric and think they are warning about future software. They are describing capital itself, projected onto matrix multiplication so tech billionaires and the workers they just laid off can pretend to share a common enemy.

2028 State-monopoly bailout and the 2028 election

When circular AI capex cracks in late 2027, the US state does not fall to a revolution—it bails out the monopolies.

Spontaneous walkouts flare at Virginia substations, Seattle tech campuses and rural water hearings, but forty years of deindustrialization, union-busting and Cold War anti-communism cannot be undone in a single autumn. With union density barely at 10% and no mass Communist vanguard rooted in heavy infrastructure, anger in the streets lacks the organization to contest power.¹³

Washington steps in from above. Invoking the Defense Production Act, the White House backstops stranded GPU debt with Pentagon cloud contracts and folds failing labs into a state-supervised consortium—the fusion of monopoly capital and the state that Lenin analyzed in 1917.¹⁴ Staff require security clearances, and researchers with “suspect political views” are “sidelined or fired.” AI 2027 reports political purges inside its hero lab without a word of objection.

AI 2040 picks up the story in Congress, where senators worry on camera that Sam Altman or Elon Musk might make himself dictator, then pass an AI Transparency Act requiring companies to publish model specs and safety scores. Its own authors admit the bill does “some good and some bad” and “doesn’t fundamentally change the situation.” Popular rage is funneled into the 2028 presidential campaign while datacenter spending rivals the Pentagon budget. The public gets paperwork; the billionaires keep the clusters.

“To decide once every few years which member of the ruling class is to repress and crush the people through parliament—this is the real essence of bourgeois parliamentarism.”

For serious militants in the grid unions, warehouses and tenant councils, the containment of 2027–2028 settles an old debate. Spontaneous protest cannot defeat an intact bourgeois state; a disciplined vanguard party has to be built cell by cell where the power switches actually sit.

2029–2030 A digital Baruch Plan

AI 2040 makes the turn of the decade its moment of decision and pitches a compute cartel to save US primacy.

Tech lobbyists in its script warn that regulation will reduce America to a “CCP tributary state”; the authors’ own answer to an “irreversible concentration of power” is a US–China treaty enforced by hardware bugging, remote kill-switches and “mutually assured compute destruction,” wiring frontier clusters so a suspected breach triggers sabotage or air strikes.

Washington tried the same trick in 1946 with the Baruch Plan, demanding international inspection of Soviet atomic research before giving up a single American bomb.¹⁵ Since 2023 China has backed UN-centered AI governance and every country’s right to develop the technology, and it hasn’t fought a war since 1979. Washington tore up the ABM Treaty in 2002 and the INF Treaty in 2019 the moment those limits got in its way. Kautsky argued in 1914 that great capitalist powers could form a peaceful cartel to divide the world; Lenin answered that inter-imperialist truces hold only until the balance of strength changes.¹⁶ Because private monopolies must accumulate or die, a capitalist compute cap lasts only until breaking it pays.

The flowchart below keeps AI 2040’s three branching questions and traces where each option actually leads.¹⁷

Beijing rejects remote kill-switches on public infrastructure and keeps tabling its UN governance framework. Washington answers with tighter secondary sanctions—forcing every sanctioned industry in China and the Global South onto domestic tools faster than any subsidy could.

2031–2035 Basic socialist modernization vs. the stipend trap

Across the early 2030s, socialist planning and state-monopoly capitalism produce opposite worlds.

Under its 15th and 16th Five-Year Plans, China completes fully domestic extreme-ultraviolet lithography, high-bandwidth memory and advanced packaging chains. In 2035, fourteen years after eradicating extreme poverty for its first centenary in 2021,¹⁸ the PRC reaches the first stage of Xi Jinping’s two-stage strategic plan: basically realizing socialist modernization (jiben shixian shehuizhuyi xiandaihua).¹ Open-weight models running on Dongshu Xisuan renewable grids coordinate rail freight, hospital diagnostics, flood control and factory tooling. Through Digital Silk Road joint ventures, Chinese ultra-high-voltage equipment, solar lines and sovereign public clusters come online across ASEAN, Africa and Latin America—including Kenya, the Philippines, Venezuela and Uganda, where subcontractors once paid $2 an hour to clean Silicon Valley’s training sets.

Inside the United States, structural underemployment settles above 22%. To contain unrest without surrendering ownership, Congress introduces Silicon Valley’s favorite pacifier: a means-tested universal basic income paid in platform credits—a stipend set by billionaires, audited by algorithm and revocable for missed workfare check-ins or strike activity.

Yet the material buffer that kept US domestic peace for eighty years—what Lenin called the labor aristocracy—has collapsed.¹⁹ Imperial super-profits no longer pay domestic technical and professional staff enough to divide them from warehouse, agricultural, migrant and super-exploited Black and brown workers. Former six-figure software engineers sit on the same eviction dockets and utility shutoff lists as fulfillment-center crews, while Indigenous and rural coalitions blockade desert cooling pipelines.

“They said learn to code. We did. Then they trained the model on our code.”

There are no shortcuts in the workplace. Early datacenter walkouts in 2029 and 2031 are broken by Taft-Hartley injunctions and strikebreaking contractors. Learning from those defeats, Communist Party cadres spend a decade merging militant rank-and-file caucuses across the physical chokepoints of AI capital: high-voltage substation operators, cooling and water-treatment technicians, West Coast dockworkers, rail engineers and municipal tenant unions. A billion-dollar cluster can automate code; without cooling crews and grid dispatchers, its racks thermal-throttle and go dark in hours.

2036–2039 Crisis of imperial rent and the Taiwan war drive

By the late 2030s, Western tech monopolies lose the overseas tribute that kept domestic stipends afloat.

As Global South ministries run open-weight models on sovereign clusters and clear trade outside the dollar, Silicon Valley’s API rents and Wall Street’s seigniorage dry up together. Crushed between depreciating datacenters and shrinking foreign tribute, Washington turns the screws at home: slashing basic-income credits, raiding public pensions, hiking household electricity rates to subsidize military compute, and criminalizing solidarity strikes.

Faced with a crisis it can no longer export by finance, the US ruling class prepares to export it by war. Briefings warn Congress that China’s modernized industrial grid has permanently overtaken the US defense base. Cable networks and prestige papers repeat “digital authoritarianism” until encirclement sounds defensive—the press manufacture of consent that Michael Parenti documented.²⁰ The same AI safety institutes that spent the 2020s preaching caution against existential risk publish white papers urging pre-emptive strikes on Chinese fabs and compute hubs.

Washington sends carrier groups and missile batteries toward the Taiwan Strait. Taiwan is a province of China: UN Resolution 2758 recognizes the PRC as China’s sole legitimate representative, and across the Three Joint Communiqués of 1972, 1979 and 1982 Washington acknowledged one China and promised to wind down arms sales to Taipei.²¹ Having lost its chip stranglehold as mainland fabs matched TSMC, Washington escalates in the strait anyway to salvage military primacy off China’s coast, ties extended stipend eligibility to military enlistment, and jails union stewards under national-security injunctions.

2040–2041 Dual power and socialist rupture

AI 2040 imagined 2040 as the year Washington and Beijing politely “unpause” corporate superintelligence; instead, fifteen years of class recomposition break the imperial war machine.

Unlike 2027, when layoffs hit an atomized workforce, a steeled Communist Party now anchors national industrial unions across power generation, cooling infrastructure, rail, ports and logistics, backed by neighborhood and unemployed councils in every major metro. Tens of thousands of technicians and dispatchers keep the clusters alive every hour; a few hundred billionaire families and Pentagon contractors hold title to them.

When the White House invokes wartime labor requisitions in late 2040 and drafts strike packages against Chinese ports and inland networks, the industrial unions answer with a political general strike. Power-plant operators walk out alongside West Coast dockworkers, rail crews and datacenter technicians. Fired software engineers who know the stack help elected workers’, neighborhood and Indigenous councils keep hospital, water and transit grids running beside paralyzed city halls. Dual power, built over fifteen years rather than wished into being, governs the industrial corridors. National Guard units recruited from working-class towns gutted by utility bills and stipend cuts refuse orders to storm the substations.

By 2041, delegates from workplace, neighborhood, rural and Indigenous councils convene the first National Workers’ Congress and form a provisional workers’ government. Applying the lesson Marx drew from the Paris Commune and Lenin carried out in 1917, they dismantle the courts, intelligence agencies and police apparatus of monopoly capital instead of governing through them.²² All frontier labs, hyperscale clusters and major banks are expropriated without compensation after technicians block departing executives from wiping weights or moving liquid assets offshore. Abolishing ad auctions, behavioral surveillance and high-frequency trading frees a third of national compute on day one.

2042–2046 Reconstruction, calculation, and the People’s Spec

Taking power in a deindustrialized empire does not conjure instant abundance; it begins decades of hard reconstruction.

Half a century of Wall Street offshoring left North America without advanced fabs, ultra-high-voltage transformer factories or heavy machine-tool depth, alongside depleted Western aquifers, aging transmission lines and intermittent sabotage by counter-revolutionary holdouts. Survival requires ending imperial isolation immediately. The workers’ republic pulls the Seventh Fleet from the Taiwan Strait, affirms the one-China principle, closes Washington’s foreign military bases, lifts every sanction on Cuba, the DPRK, Venezuela and Iran, cancels Wall Street’s Global South debt claims, and joins the UN AI governance framework Beijing first proposed in 2023. Open software and research are traded on equal terms for Chinese grid transformers, solar and battery lines, rail tooling and fab machinery to rebuild hollowed-out industrial belts.

For a century Austrian-school economists insisted socialism was impossible because no planner without market prices could solve millions of supply equations.²³ History answered long before modern accelerators: Soviet Five-Year Plans turned a peasant country into the industrial power that broke the Wehrmacht, Allende’s Chile routed factory data through Project Cybersyn’s telex room during the CIA-backed truckers’ strike of 1972, and China’s Five-Year Plans eradicated extreme poverty in 2021 and modernized an economy of 1.4 billion people by 2035. North American planners adapt those cybernetic methods under public ownership. Factories, utilities and warehouses post real-time capacity; local and Indigenous councils submit needs; planetary carbon and water limits enter as hard ceilings; and narrow superhuman optimization models clear billions of input–output balances overnight to publish competing draft plans for recallable delegates to debate and adopt.

Where AI 2027 and AI 2040 treated alignment as training a chatbot to obey a corporate constitution, the Workers’ Congress asks which class writes the rules. General AI remains capped at top-human-expert level while workplace councils, Chinese academies and UN standards bodies ratify a binding public charter: the People’s Spec.

Joint public interpretability institutes in Beijing, Chicago, Nairobi and Caracas audit every frontier model openly. Models trained under capitalist competition learned to game evaluations because hiding a flaw protected a valuation; abolishing private equity removes anyone who profits from deception. Meanwhile, abolishing parasitic finance, ad-tech and private-insurance bureaucracy—and dividing necessary reconstruction work across everyone formerly warehoused on unemployment or stipends—cuts the working week steadily instead of keeping millions idle.

Marx called time freed by machinery “disposable time” and named it the real measure of social wealth.²⁵ Crossing the First Reconstruction Plan’s twenty-four-hour target frees tens of millions of workers to rotate through technical retraining, local planning commissions, ecological remediation and public administration.

October 2049 The centenary horizon

On 1 October 2049, the People’s Republic of China marks one hundred years since Mao Zedong proclaimed its founding at Tiananmen.

Across sixteen Five-Year Plans since the 1950s, socialist China moved from semi-colonial ruin to eradicating extreme poverty in 2021, basically realizing socialist modernization in 2035, and fulfilling the Second Centenary Goal set out in The Governance of China: building a “great modern socialist country that is prosperous, strong, democratic, culturally advanced, harmonious, and beautiful” (shehuizhuyi xiandaihua qiangguo).¹ High-speed rail, fusion-assisted renewable grids, open scientific compute and public health infrastructure span town and countryside alike, linked with sovereign public compute networks across Asia, Africa and Latin America.

North America’s workers’ republic arrives at 2049 only eight years into its own socialist transition—closer to where China stood in the 1950s politically, even if starting with advanced datacenters. Its billionaires are expropriated, its foreign bases are closed, and its average working week stands at twenty-two hours, yet domestic fab lines, remediated river basins and new public housing districts remain mid-construction, and remnant bourgeois habits and regional inequalities still demand patient struggle. Contrast that real balance sheet with the fantasies of Bay Area rationalism: AI 2027 expected history to end in twenty-four months inside a San Francisco server rack, and AI 2040 hoped a cartel treaty could freeze capitalism in amber. Historical materialism measures transformation in generations of organized labor.

Only in late 2049—under a binding UN global compute plan ratified jointly by centenary China, Global South republics and the North American Workers’ Congress, and audited line by line under the People’s Spec—do public labs vote to scale general AI systems past earlier caps to tackle fusion materials, climate restoration and cellular medicine.

Even in 2049, world socialism remains in its protracted lower phase: uneven global development, enclaves of private capital and the division between mental and manual labor have not vanished overnight, and as Marx, Lenin, Deng Xiaoping and Xi Jinping all insisted, the workers’ state cannot wither away while class remnants and uneven development persist.²⁶ What 2049 proves is simpler, and fatal to Silicon Valley’s mythology: machines do not make history—classes do.

“From each according to his ability, to each according to his needs!”