What follows is the shared timeline of events that, looking back from the last day of 2030, reshaped the landscape between January 2026 and now. They are ordered chronologically. Each is dated to the quarter and marked as either "predictable" (visible to a careful observer in early 2026) or "absurd" (would have sounded implausible to that same observer). Three deliberate non-events are included at the end: things that were widely expected in early 2026 and did not arrive by 2030.
The timeline is selective rather than comprehensive. The events included are those that reshaped how entire professions, institutions, and career paths worked. Where possible, an event is anchored to a concrete date, a concrete organisation, and a concrete mechanism.
At a glance
The 14 anchor events plotted by quarter and category. Predictable events are solid; absurd events are outlined.
The anchor events
The Tooling Acceleration
Anthropic shipped Claude Opus 4.6 in February, Sonnet 4.6 with the million-token context window twelve days later, and by March had released twelve major features in twelve weeks. MCP crossed 97 million installs and went under Linux Foundation governance. OpenAI launched GPT-5.4 with record scores on knowledge-work benchmarks. Claude Cowork went generally available. The Dispatch feature let agents use your computer while you were away from the desk. The first orbital data centre nodes reached low-Earth orbit on January 11th.
None of this was surprising to anyone paying attention. What was surprising was the speed at which it compounded. By the end of Q1 2026, a single person with a Claude subscription could do what a five-person professional team had done six months earlier. The economic implications were obvious. The institutional response was not.
The Consulting Contraction
McKinsey's 200-person layoff in Q1 was the canary. By Q2 2026, the contraction had spread across the industry. BCG, Bain, Deloitte, and Accenture all reduced graduate intake by 40-60% for the autumn cycle. The stated reason varied - "strategic restructuring," "evolving service model," "investment in AI-native delivery" - but the mechanism was the same: agentic AI systems could now perform the analytical, research, and slide-production work that had historically justified the first three years of a consulting career. The junior analyst tier was not being automated away. It was being made unnecessary.
Industriell Ekonomi programmes at LiU, KTH, and Chalmers saw their most reliable career pipeline narrow overnight. The students who had started their degrees expecting to walk into McKinsey or BCG upon graduation discovered, mid-degree, that the conveyor belt had stopped. The consulting partner-track and the associated elite-finance pipelines reset at the same time, in the same way, for the same reason.
The Swedish Compact
Sweden moved faster than anyone expected. In a cross-party agreement brokered between the Tidö parties and the opposition, the Riksdag passed the Omstallningspakten (the Transition Compact) - a legislative framework that included: accelerated public compute infrastructure (AI Sweden scaled to provide enterprise-grade AI access to every Swedish resident and SME), a restructured a-kassa (unemployment insurance) system with higher ceilings and longer duration for AI-displaced workers, and a mandate for Arbetsformedlingen to develop AI-augmented retraining pathways. It was not revolutionary in ambition. It was revolutionary in speed. No other OECD country had passed anything comparable.
The Compact was imperfect, underfunded relative to the scale of displacement, and politically fragile. But it established Sweden as the proof-of-concept that section 4 said was needed: evidence that democratic institutions could respond to the transition in something closer to real time. Nordic Tech Club talked about nothing else for three months.
The First Autonomous Fund Crosses $10 Billion AUM
An AI-native quantitative fund, operating with fewer than fifteen human employees (legal, compliance, and a skeleton oversight team), crossed $10 billion in assets under management. Its track record over eight quarters had beaten every human-managed fund in its category. The capital inflows accelerated. The fund's investment decisions were made entirely by AI agents. Its risk management was handled by AI agents. Its regulatory filings were prepared by AI agents. The humans were there because the law required it, not because the fund did.
This was the moment The Inhuman Economy (Scenario 5) stopped being theoretical. The question for any investor, founder, or operator became simple: which businesses can operate with AI agents as the primary workforce, and which are still dependent on human labour that will become more expensive relative to its output every quarter?
The London UX Collapse
The junior and mid-level UX design market in London contracted by roughly 50% over the twelve months from Q1 2026 to Q1 2027. Agencies that had employed teams of eight to ten UX designers restructured around two or three senior practitioners working with AI design tools. The output was better, faster, and cheaper. The displaced designers flooded into adjacent fields - product management, service design, "AI experience design" - but those fields were contracting too, for the same reason.
For visa-dependent workers, the contraction was existential. A Skilled Worker visa tied to a UX design role at a specific employer became a ticking clock when that employer restructured. The Home Office processing times, the three-month window to find a new sponsor, the rent in a city that had not become cheaper - the structural vulnerability that had always been latent in the visa-dependent knowledge-worker model became acute. Amara's story was not unique. It was one of thousands.
The Karolinska Announcement
Karolinska Institutet announced that its AI diagnostic system, trained on the largest clinical dataset in Scandinavian history and integrated with real-time genomic analysis, had outperformed specialist physicians in diagnostic accuracy across twelve of the twenty most common diagnostic categories in Swedish primary and secondary care. Not by a narrow margin. By a statistically significant margin that was consistent across patient demographics and clinical contexts.
The announcement did not say "AI will replace doctors." It said something more precise and more disruptive: the diagnostic component of clinical medicine - the part that medical students spend years training for, the part that distinguishes a specialist from a generalist - was now better handled by a system than by a human in the majority of common cases. The implications for a student who had just finished her first year of the medical programme at Lund were not abstract.
The Redundancy Summer
The summer of 2027 was when The Hollowing stopped being a trend piece and became lived reality for the professional middle class in North America and Western Europe. A cluster of major employers - three of the Big Four accounting firms, two global banks, a Fortune 50 technology company, and the UK's largest law firm - announced restructurings that collectively eliminated over 120,000 white-collar positions. The stated reason, in every case, was the deployment of agentic AI systems that had completed a 12-18 month integration and were now handling work previously done by junior and mid-level professionals.
The scale was not unprecedented in absolute numbers - previous recessions had produced larger aggregate job losses. What was unprecedented was the specificity: these were not cyclical layoffs to be reversed when demand recovered. They were structural eliminations of roles that would not return because the work was now done by systems that did not take holidays, did not make errors of fatigue, and cost a fraction of the annual salary. The Redundancy Summer was the quarter in which the phrase "this time is different" stopped being cliche and became diagnosis.
The First Fully Autonomous Firm Reaches $500M Revenue
A logistics and supply chain management company, registered in Singapore, operating with zero permanent human employees and governed by an AI agent system with a three-person human board of oversight, reported annualised revenue of $500 million. It had been founded eighteen months earlier. It managed freight routing, customs documentation, warehouse allocation, and last-mile delivery coordination across Southeast Asia using a network of AI agents and contracted robotic infrastructure. Its customers included two of the world's largest e-commerce platforms.
The company was not a gimmick or a proof-of-concept. It was a commercial enterprise that outcompeted human-staffed competitors on price, speed, and reliability. The question it posed was the one section 2 anticipated: if the most efficient way to organise economic activity is without human employees, what is the economy for?
The Mirror Protein Demonstration
A research group (the institution and country do not matter for this timeline; what matters is that it happened) demonstrated the first functional self-replicating mirror protein complex in a controlled laboratory environment. Not a mirror organism. Not a mirror bacterium. A protein complex that could replicate its own structure using mirror-image amino acids. The paper was published, withdrawn under government pressure, republished on a preprint server, and became the most downloaded biology paper in history within 72 hours.
The immediate biosecurity response was intense: emergency meetings at the WHO, calls for moratoriums on mirror biochemistry research, a scramble to assess containment protocols that had never been designed for this class of entity. The long-term implication was the one flagged in section 1: the gap between "theoretical" and "laboratory proof-of-concept" for mirror life had closed, and the gap between "proof-of-concept" and "self-replicating organism" was now a matter of engineering, not discovery. The Breach (Scenario 9) had moved from tail risk to foreground risk in a single publication.
The Nordic Redistribution Pilot
Building on the Swedish Compact of 2026, Finland and Sweden jointly launched a redistribution pilot: a "compute dividend" funded by a levy on AI-generated economic output, distributed as a combination of cash transfers and credits for AI-augmented services (healthcare, education, legal advice, business formation). The pilot covered 200,000 participants across both countries. It was not UBI. It was something more targeted and more interesting: an attempt to build a distribution mechanism specifically designed for an economy where the majority of value is created by AI systems rather than human labour.
The early results were mixed, politically contested, and closely watched by every other government that was struggling with the same problem. The pilot's existence mattered more than its results: it was the first serious attempt by any government to answer the distribution question at the structural level rather than with retraining programmes and unemployment extensions.
The Taiwan Strait Incident
A naval confrontation in the Taiwan Strait involving Chinese and American vessels, complicated by autonomous underwater drones operating on both sides, escalated to a level that required direct head-of-state communication to de-escalate. No shots were fired by human combatants. Two autonomous systems engaged each other briefly before being recalled. The incident lasted eleven hours. Semiconductor supply chains seized for three weeks. TSMC's stock dropped 30% and did not fully recover for six months. Global AI chip deliveries were delayed by a quarter.
The incident was The Intelligence Wars (Scenario 8) in miniature: a great-power confrontation driven by AI competition, complicated by autonomous systems that operated faster than human decision-making, and resolved only because both sides chose to step back from a brink that their own systems had pushed them toward. The Brittleocene's fragility thesis received its most vivid illustration: the entire global AI infrastructure depended on fabrication capacity concentrated in the most geopolitically contested strait on Earth.
SLU Reorganises Medical Education
Lunds Universitet, under pressure from Karolinska's diagnostic results, from the Socialstyrelsen's updated guidelines on AI-assisted clinical practice, and from a student body that could see the trajectory, announced a fundamental restructuring of its medical programme. The new curriculum reduced diagnostic training from the traditional model and introduced mandatory tracks in AI-augmented clinical practice, computational biology, precision medicine, and what the announcement called "human-centred care in automated systems" - the skills that remained irreducibly human in a clinical environment where diagnosis, imaging interpretation, and treatment protocol selection were increasingly handled by AI.
The restructuring was controversial. Traditionalists argued it was abandoning the core of medical training. Reformists argued it was the only honest response to a profession that had changed faster than its educational institutions. The first cohorts trained under the new model found themselves either fortunate or disorienting, depending on the day - but with entirely different professional options from those who had enrolled a year earlier.
The Longevity Threshold
The combined results of three clinical trials - Altos Labs' cellular rejuvenation programme, a senolytics combination trial led by the Mayo Clinic, and an AI-designed multi-target aging intervention from Insilico Medicine - were presented at a joint session that the organisers called, without exaggeration, the most important medical conference of the decade. The aggregate finding: in treated populations (all trials were still in controlled cohorts, not general deployment), biological aging markers had been decelerated by 30-40% over a two-year treatment window. The effects appeared durable. The side-effect profiles were manageable.
This was not the cure for aging. It was the proof-of-concept that biological aging is a modifiable condition. The implications cascaded through every domain mapped in this document: pension systems, career planning, insurance models, political time horizons, ecological sustainability, the distribution question (who gets access first?), and the deeply personal question of what a life means when it might last 120 years instead of 80. The proof-of-concept changed the time horizon for every career plan, investment thesis, and life decision anchored to the old actuarial assumptions.
The Patchwork Becomes Official
By early 2030, the patchwork that section 4 predicted was no longer an analytical framework. It was the headline description of the world order. The OECD's annual report used the phrase "divergent transition pathways" to describe a global landscape in which: the Nordic states and Singapore were furthest along toward managed abundance (The Abundance Republic, imperfectly). The United States was deep in The Hollowing, with a political system unable to pass redistribution legislation, a thriving autonomous-firm sector, and a displaced professional class that had become the defining constituency of a new populist movement. China had implemented a state-managed version of The Comfortable Cage, with comprehensive AI-mediated service provision, tight social credit integration, and economic growth that was the envy of countries that could not match its top-down coordination. The UK had followed a path somewhere between the US and the EU, worse than the Nordics and better than the US, with London's financial sector thriving and its creative and professional middle class hollowed out. Much of the Global South was experiencing simultaneous empowerment (AI tools in the hands of individuals and small businesses) and disruption (traditional export industries automated by clients abroad), with outcomes varying enormously by country.
The world had not converged on a single future. It had fractured into adjacent futures. The question for the next decade was whether these adjacent futures could coexist, or whether the divergence would produce conflicts, migrations, and instabilities that none of them was equipped to manage alone.
The non-events: what did not happen
Three things were widely expected in early 2026 that did not arrive by 2030. Their absence mattered as much as the presence of the events above.
Artificial general intelligence was not declared
In early 2026, several AI labs were privately circulating timelines suggesting AGI - a system matching or exceeding human cognitive ability across all domains - by 2027 or 2028. By the end of 2030, no such declaration had been made. Not because the systems were not extraordinarily capable. They were. But the goalposts moved with each advance: each new capability was absorbed into the category of "narrow AI" or "agentic AI" and the definition of AGI was revised upward. The practical consequence was that systems that were, by any 2024 definition, superhuman across dozens of domains were not called AGI, and the policy responses that many had tied to the AGI milestone never triggered. The absence of the label mattered more than the presence of the capability.
Fusion did not reach commercial viability
The US Department of Energy's 2026 fusion funding, the private-sector pilot reactors, the AI-accelerated plasma modelling - all of it continued. Progress was genuine. Several pilot plants achieved sustained ignition. But the engineering challenges of converting plasma energy into grid electricity at commercial cost and scale proved slower than the most optimistic 2026 timelines suggested. As of December 2030, fusion is "five years away" - a joke that is older than most of the AI systems that now run the economy. The implication: the energy transition of the 2026-2030 period was powered by solar-plus-storage and advanced fission, not fusion. The near-zero-marginal-cost energy future described in section 1 is arriving, but via the boring path rather than the glamorous one.
No major AI safety catastrophe occurred
This is the non-event that matters most. The Last Handoff (Scenario 7) did not happen. No AI system produced an uncontrolled loss-of-control event. No misaligned system caused mass harm. The safety research community's work, combined with the regulatory frameworks (the EU AI Act, the US executive orders, the informal norms within the leading labs) held. The sceptical reading is that we were lucky. The optimistic reading is that the alignment techniques work at the current capability level. The honest reading is that both may be true, and that the next five years - with systems that are meaningfully more capable than today's - will be the real test. The absence of catastrophe is not evidence of safety. It is evidence that the catastrophe has not happened yet.