The Long Game

The developers who began repositioning during the IT offshoring wave of 2001-2006 had meaningfully better outcomes than those who waited for the old market to return. BLS data tells the story in two occupation codes: “computer programmer” employment peaked above 700,000 during the dot-com boom and has since fallen to approximately 121,200 — a level not seen since 1980. Meanwhile, software developer employment — the category BLS created in 2003 to capture design, architecture, and requirements work — grew steadily, with 15% growth projected through 2034. The profession did not shrink. It bifurcated. The developers who moved from writing code to specification toward designing systems, understanding domains, and making architectural decisions built careers that lasted. Those who waited for the old programmer role to come back are still waiting.

We are in that window again. And this time, the window is shorter.

The evidence base behind this piece comes from 40 research documents across five independent lenses — historical precedent (16 case studies spanning 150 years), AI impact (productivity data, tool adoption, role evolution), market landscape (hiring trends, compensation, employer dynamics), industry sectors (10 sector analyses), and a generational turnings lens applying Neil Howe’s cyclical theory and historical labor market patterns to the current moment. The strategy that follows is grounded in findings where four or five of those lenses converge. Where the evidence is weaker, that is stated explicitly. Where it genuinely conflicts, both sides are presented.

The compressed evidence base

Before the strategy, the foundation. Three findings carry the highest confidence in the entire research project — four or five lenses independently reaching the same conclusion. If you have read the other pieces in this collection, these will be familiar. If this is your first, they are the minimum context for what follows.

The profession is bifurcating, not dying. Across 16 historical cases of workforce disruption — typesetters, drafters, bookkeepers, travel agents, bank tellers, factory workers — the procedural layer of work was destroyed while the judgment layer was preserved or elevated. The dividing line was always the same: the people who understood the domain survived; the people who operated the tools did not. The current market data confirms the same split is underway in software: junior hiring share at major tech companies has collapsed 78%, from 32% to 7%, according to SignalFire data via Stack Overflow. BLS projects 15% software developer employment growth through 2034 — but the growth is in senior, architectural, and AI-specialized roles, not in routine implementation.

Domain expertise is the durable moat. In every historical disruption studied, domain knowledge was the single strongest predictor of positive outcomes. Travel agents who survived specialized in luxury and complex travel. Photographers survived in niches requiring deep craft knowledge. IT workers embedded in specific industries survived offshoring. MIT CSAIL identifies proprietary conventions, institutional knowledge, and physical-world integration as structural challenges for AI — not temporary gaps. Across 10 industry sectors analyzed, the correlation is near-perfect: the sectors with the highest entry barriers offer the strongest AI insulation, and the barrier is the moat.

No one is coming to manage this transition for you. Government retraining programs have performed poorly across nearly every case studied. Trade Adjustment Assistance participants fared worse than non-participants. The most successful institutional response in the entire dataset — Sweden’s job security councils achieving 90% reemployment — requires infrastructure that does not exist in the US tech sector. Every company is making independent adoption decisions simultaneously, and no coordinating institution exists to manage the pace. Self-directed adaptation is the only strategy supported by the evidence.

The infrastructure thesis

There is a second argument, lower in confidence but important for anyone thinking on a 10-20 year horizon.

Crisis periods have historically triggered infrastructure investment at a scale that creates entire career ecosystems lasting 20-40 years. The New Deal produced the TVA, still operating 90 years later. The Manhattan Project created the national laboratory system, which today employs over 100,000 people at the Department of Energy. The Interstate Highway System took 36 years to build. Peer-reviewed research from the Quarterly Journal of Economics (Garin & Rothbaum, 2025) provides the most rigorous evidence: WWII plant construction had “large and persistent effects on local development,” with affected counties seeing 5-10% earnings increases that persisted for over 40 years. Children from those counties earned $1,200 more annually over their adult careers.

The current investment cycle is at historically unprecedented scale: the CHIPS Act ($52.7 billion), the Inflation Reduction Act ($369 billion in clean energy), the Bipartisan Infrastructure Law ($1.2 trillion authorized), plus private AI infrastructure capex of $660-690 billion in 2026 alone from five hyperscalers. Combined, these rival or exceed the postwar infrastructure buildout in inflation-adjusted terms.

But three caveats weaken the thesis substantially.

First, the current AI infrastructure boom is overwhelmingly private, not public. Every major crisis-era infrastructure program — WPA, TVA, Manhattan Project, Interstate Highways, NASA — was government-funded, which gave employment political protection. AI data centers are funded by five companies subject to market dynamics. If AI revenue disappoints, companies will cut. The more accurate historical analogy for private infrastructure overbuild may be the 1990s telecom fiber buildout, which collapsed.

Second, political risk to public programs is documented, not speculative. The IRA faces approximately 60% rollback of energy tax credits via the House Ways and Means reconciliation draft, with an estimated 790,000 jobs at risk. The administration has called the CHIPS Act “horrible.” DOGE terminated $5-7 billion in annual IT contract value, directly shrinking the near-term government IT market.

Third, the same QJE researchers who documented the 40-year persistence of WWII plant effects explicitly warn that “opening a comparable factory today would not have similar effects on local employment and wages” — because contemporary manufacturing faces declining union influence, increased automation, and greater international competition.

What this means practically: The infrastructure thesis is strongest for developers who target the most politically protected and operationally durable segments — government-directed programs, regulated utilities, defense — rather than private AI data center construction. The durability hierarchy holds: infrastructure over applications, government-directed over private, operating roles over construction roles.

The generational turnings lens

There is a framework — contested, unfalsifiable by the standards of professional historians, and yet stubbornly interesting — that offers a structural way to think about where we are and what comes next.

Neil Howe’s generational turnings theory argues that American history moves in roughly 80-year cycles of four “turnings”: a High (institutional confidence, collective building), an Awakening (cultural upheaval, individualism), an Unraveling (institutional decay, cynicism), and a Crisis (old order breaks, new one is forged). Each generation occupies a different archetype — Prophet, Nomad, Hero, Artist — and each archetype enters a different life stage during each turning. The theory places us squarely in a Crisis (the Fourth Turning), roughly analogous to the late 1930s: post-Depression restructuring, pre-climactic mobilization. Howe suggests this crisis will probably resolve “around 2030,” with a new High — a period of civic renewal, institutional rebuilding, and expanding prosperity — emerging in the mid-2030s.

The caveat is real and worth stating plainly. Mainstream academic historians describe the framework as “overly deterministic, unfalsifiable, and unsupported by rigorous evidence.” The archetypes are heuristics, not mechanisms. Individual circumstances — economic class, geography, industry, luck — determine outcomes far more than generational membership. The City Journal review of Howe’s latest book is sharply skeptical, characterizing some of his predictions as “dystopian.”

But here is why it earns space in a strategy piece. The framework’s value is not predictive — it is structural. It provides a way to think about how the same disruption hits different career stages differently, and more importantly, about what might come after the current disruption. The independently verifiable evidence — the Dallas Fed experience premium, the entry-level hiring collapse, the ageism data, the QJE study on crisis-era infrastructure creating 40-year career ecosystems — all fits the pattern Howe describes, whether or not you accept the cyclical theory that generated it. The practical advice below does not depend on Strauss-Howe being correct. But the generational lens helps explain why the same evidence produces different strategic implications depending on where you are in your career.

In Howe’s framework, each generation alive today occupies a distinct archetype during this Crisis:

• Boomers (ages 66+) are elder Prophets — their crisis role is moral vision and managed exit, not new construction
• Gen X (ages 45-65) are midlife Nomads — pragmatic, self-reliant, producing “America’s greatest entrepreneurs” according to Strauss-Howe, holding critical systems together under crisis pressure
• Millennials (ages 22-44) are young-adult Heroes — the cohort expected to fight the crisis and build the institutions of the next era, analogous to the GI Generation that built the postwar economy
• Homelanders (under ~21) are childhood Artists — growing up overprotected amid turmoil, entering adulthood during the next High as the most adaptable cohort

The last time this cycle played out, the GI Generation (the previous Hero archetype) entered adulthood during the Depression, fought WWII, used the GI Bill to get educated, and then spent their 30s-50s building the Interstate Highway System, NASA, the military-industrial complex, and the suburban middle class. Over 2 million servicemembers graduated with degrees through the GI Bill, forming the educated technical workforce that built the 1950s-60s economy. The Silent Generation (the previous Artist archetype) grew up during the Depression and war, entered careers during the booming 1950s High, and became the wealthiest and most stable generation of the cycle — despite (or because of) growing up during crisis.

Whether or not you accept the cyclical theory, the structural observation is real: different age cohorts face the same disruption from different positions, and those positions suggest different optimal strategies. What follows is organized by career stage — with the generational lens providing context for the longer horizon.

Strategy by career stage

The same disruption hits every developer. But a 25-year-old entering the industry and a 50-year-old at the peak of their career face fundamentally different strategic landscapes. What follows is organized by career stage, with confidence levels attached to each recommendation.

Junior developers (0-3 years experience) — Homelanders and younger Millennials

The evidence is convergent and severe. Harvard’s study of 62 million workers found a 9-10% junior employment decline at AI-adopting firms within six quarters. Stanford’s Digital Economy Lab found a 20% employment decline for developers aged 22-25 from the late-2022 peak. The Anthropic RCT found that developers who delegate code generation to AI score 17 points lower on comprehension — the skills the market most values (debugging, code review, system understanding) are the skills AI use can erode if not practiced deliberately.

What the evidence says to do (high confidence — 4+ lenses converge):

Survival first, positioning second. Take whatever role builds real systems experience — infrastructure, DevOps, QA automation, security operations, internal tooling. These are the roles least vulnerable to AI displacement and most likely to teach how production systems actually work. Target sectors still actively hiring juniors: healthcare IT, government and defense, cybersecurity, energy and utilities. Smaller companies are better targets than FAANG — they are more likely to hire entry-level and more likely to expose you to the full system rather than a narrow slice. Specialize early. The generalist junior path leads into the surplus segment of a bifurcated market.

Use AI as a learning accelerator, not a bypass. The Anthropic comprehension finding is the critical data point. Build the skills AI cannot replicate — debugging unfamiliar codebases, reviewing someone else’s code, understanding why a system behaves the way it does — by doing them, not by asking a model to do them for you.

What to avoid: Waiting for the junior hiring market to recover to pre-2022 levels. It will not — 54% of engineering leaders expect permanent declines in junior hiring, per a LeadDev survey. Competing for remote generalist positions against global talent. Treating “prompt engineering” as a durable career identity.

The generational turnings lens offers a counterintuitive perspective. The last generation to enter the workforce during a Crisis — the Silent Generation — became the most prosperous cohort of the entire cycle. They grew up during the Depression and WWII, entered careers in the booming 1950s, and spent their entire 30-40 year career in an era of expanding prosperity. If Howe’s timing holds and a new High begins in the mid-2030s, today’s junior developers will be 27-37 — entering their prime career-building years just as institutional stability and economic expansion return. The hardship of career entry is temporary; the skills and resilience it builds are permanent. This does not make the current conditions less brutal. But it suggests that the developers who survive the entry squeeze — building real skills despite reduced hiring — may find themselves positioned for careers far more expansive than today’s market suggests.

Genuinely uncertain: Whether the junior hiring collapse is primarily cyclical or structural. The “junior death spiral” thesis — that organizations eliminating junior hiring will face a compounding talent vacuum by 2027-2028 — has merit. AWS CEO Matt Garman has publicly called it “the dumbest thing I’ve ever heard”. Some recovery is plausible. But timing a career on a thesis is risky. Act on what is knowable now.

Mid-career developers (3-10 years experience) — Millennials

The sweet spot for strategic repositioning. Experienced enough to have portable skills and institutional footing, early enough to invest in domain transitions that take 1-2 years to pay off.

In Howe’s framework, this is the Hero generation — the cohort that fights the crisis and builds the institutions of the next era. The GI Generation analogue is instructive: they entered adulthood during the Depression, fought WWII, and then spent their 30s-50s building the most prosperous economy in history. If the crisis resolves by the mid-2030s, Millennials born 1982-1996 will be 39-53 — precisely the age range when the GI Generation was building the Interstate Highway System, expanding the aerospace industry, and creating the middle class. Career decisions made now should be evaluated against that 2035-2050 horizon, not just 2026 compensation.

A structural caveat on the GI analogy: The timing parallel holds — both cohorts are positioned for peak productivity during post-crisis expansion. But the enabling conditions do not automatically replicate. The GI Generation entered their building years with no student debt, affordable housing, defined-benefit pensions, government-backed education, and rising real wages backed by union density near 35%. Millennials carry the highest student debt burden of any generation and face a labor market where pensions are largely extinct. This does not invalidate the timing advantage — career-stage positioning still matters — but it suggests the “building” phase will be more entrepreneurial and platform-based than the corporate and government-institutional building the GI Generation did.

What the evidence says to do (high confidence):

Domain specialization is the single highest-leverage investment. The convergence across historical precedent (domain knowledge is the strongest survival predictor across 16 cases), market landscape (experienced developers with AI fluency are in the talent-deficit segment), industry sectors (defense, energy, and healthcare offer the strongest moats), and AI impact research (MIT CSAIL identifies domain knowledge as a structural AI limitation) makes this the project’s highest-confidence actionable recommendation.

The sector hierarchy, combining stability, AI insulation, compensation, and accessibility: (1) defense and aerospace if you are a US citizen and geographically flexible — the triple moat of clearances, regulation, and classified domain complexity is the strongest insulation of any sector; (2) energy and utilities for the multi-decade grid modernization runway — $1.4 trillion in grid investment through 2030, with electricity demand growing at its fastest pace since WWII; (3) healthcare IT for the largest addressable market with moderate entry barriers — $480 billion growing to $961 billion by 2030; (4) manufacturing for physical-world integration moats with a projected 3.8 million worker shortfall by 2033; (5) finance for the highest compensation but with tolerance for cyclical instability.

Move upstream from implementation to architecture and system design. Historical precedent is clear: the “move up the stack” strategy worked for IT offshoring survivors, bookkeeper-to-analyst transitions, and drafter-to-engineer transitions. But treat this as a moving target. AI is progressively encroaching on standard architectural patterns. The goal is system-level judgment that requires institutional and domain context AI cannot access — not generic architecture knowledge that AI can replicate.

What to avoid: Remaining a generalist in a software-native company without domain specialization. Staying at a software-embedded employer that treats IT as a cost center with legacy stacks and no AI adoption plan — you gain short-term stability at the cost of long-term employability. Refusing to adopt AI tools. Historical precedent is unambiguous across 16 cases: resistance to the disrupting technology accelerates rather than prevents displacement.

Genuinely uncertain: The pace of enterprise AI adoption determines how long the software-embedded insulation window lasts. Current data leans toward a shorter timeline — banking and financial services hold 20% of global AI market spending, and 88% of organizations report using AI in at least one function. But enterprise adoption has historically been slower than projections.

Senior developers (10+ years experience) — Gen X Nomads

The experience premium is rising in AI-exposed occupations. The Dallas Fed found that returns on job experience are increasing — not decreasing — in highly AI-exposed industries. In computer systems design specifically, wages have increased 16.7% since fall 2022, primarily for experienced workers. Google’s internal study found that senior developers saw the largest productivity gains from AI tools. The combination of deep system understanding and AI augmentation is genuinely multiplicative.

But the paradox is real: experience is increasingly valuable while becoming harder to sell through conventional employment. Ageism in tech is documented — 61% of tech workers over 45 report age-related employability impacts. Higher compensation makes senior developers targets for restructuring. The employer-employee relationship is structurally hostile at this career stage.

Howe’s framework names this archetype precisely: the Nomad in midlife during the Crisis. Gen X grew up as “latchkey kids” with loose oversight, came of age as alienated young adults during the Unraveling, and now occupy the crisis role that Strauss-Howe describe as “pragmatic, sacrificial leadership” — holding critical systems together while younger generations prepare for the rebuild. The Nomad archetype’s instinct toward self-reliance is now its greatest asset. The systems Gen X builds under crisis pressure may be the foundations that Millennial institution-builders scale in the post-crisis era — the wartime factory managers whose output the postwar corporate builders expanded.

What the evidence says to do (high confidence):

Position as the human judgment layer. Architecture decisions, failure mode analysis, system integration, production reliability — these are the skills AI struggles with most, the skills that retained full value across every historical disruption studied, and the skills commanding premium compensation in the current market. The code review burden is increasing — Faros AI found PR review times ballooned 91% as AI-generated code volumes rise — creating both bottleneck value and burnout risk.

Build a portable practice, not a job dependency. Reputation, client base, open-source contributions, consulting relationships — career capital that survives institutional failure. Historical precedent shows that client relationships and professional reputation consistently predicted survival across travel agents, journalists, IT workers, and photographers. The institutional trust data — layoffs during record profits, CEO pay ratios at 285:1 — makes the case that employer dependency is a structural risk, not a personality flaw.

Mentor strategically. If the junior pipeline remains suppressed, the developers you mentor become scarce mid-level talent. This is both a professional contribution and an investment in future advisory relationships.

What to avoid: Assuming seniority alone provides protection — the market rewards demonstrated specialization over generic seniority. Refusing to adopt AI tools — the risk for seniors is not AI replacement but AI irrelevance, where colleagues who use AI effectively outproduce those who do not. Remaining in a single-employer dependency without portable career capital.

Genuinely uncertain: Whether the “move up the stack” safe zone is stable or shrinking. On a 1-3 year horizon, system design and architecture are clearly AI-resistant. On a 5+ year horizon, the historical precedent lens warns that unlike every prior disrupting technology, AI improves — and no historical case featured a technology that simultaneously automated the execution layer and progressively encroached on the judgment layer. The “move up the stack” escape hatch that worked for IT offshoring survivors may be smaller and shrinking.

For ages 45-51 (younger Gen X): If the crisis resolves by the mid-2030s, you will be in your mid-50s — potentially with 10-15 productive years remaining in the post-crisis High. You may have a foot in the next era. Invest in infrastructure domains where deep experience matters. Stay technically current. The QJE study’s evidence on 40-year career ecosystems born from crisis-era infrastructure investment applies directly to your planning horizon. The Millennials you mentor now may be the ones offering you advisory roles in the post-crisis era — the Nomad-Hero relationship in Howe’s framework is symbiotic: the Nomad’s pragmatism grounds the Hero’s ambition.

For ages 52+ (older Gen X and Boomers): Strategy converges with managed exit. Monetize experience through consulting and advisory relationships. Secure financial runway. Transfer knowledge deliberately. The value you carry — decades of systems knowledge that no AI can replicate and no documentation captures — is real, but it needs to be converted into income through channels that do not require you to compete on hourly coding output. In Howe’s framing, the Prophet archetype’s crisis role is moral and intellectual leadership — and in practical terms, that means writing down what you know, mentoring the developers who will build the next era, and ensuring your understanding of how systems actually work survives your departure from the workforce.

Skill investment priorities

The following is grounded in AI impact and historical precedent evidence, organized by confidence level.

Invest heavily (high confidence — 4+ lenses converge): System design and architecture. Domain expertise in regulated or physically integrated industries. Code review and debugging, especially of AI-generated output. Security fundamentals. Communication and stakeholder management. These are the skills that retained full value across every historical disruption studied, that MIT CSAIL identifies as structurally resistant to AI, and that command premium compensation in the current market.

Invest moderately (supported by 2-3 lenses): AI orchestration and evaluation-driven development — not just using Copilot for autocomplete, but understanding capabilities, limitations, and how to evaluate AI output systematically. Infrastructure and platform engineering. Cross-system integration. Data engineering and pipeline design. These are the roles where AI is a genuine force multiplier for developers with deep understanding, and where demand is concentrated in the growing segments of the market.

Deprioritize (consistent evidence of declining value): Syntax memorization for code generation. Boilerplate implementation speed. Single-framework specialization without domain context. Pure frontend generalist skills. These are precisely the tasks AI handles best — the procedural layer that has been destroyed in every historical disruption.

Maintain but do not rely on (value is real but potentially temporary): Architectural pattern knowledge — AI is encroaching. Rapid prototyping as a differentiator — AI is faster. Language-specific expertise without system-level understanding — valuable now, uncertain at the five-year horizon.

Employer selection criteria

Not all companies in a good sector are good employers. The evidence supports specific criteria for evaluating where to work:

The worst outcome is a software-embedded role at a company that treats IT as overhead, uses legacy stacks, and will be blindsided by AI adoption — gaining short-term stability at the cost of long-term employability. The market landscape evidence points to companies that treat engineering as a strategic function and invest in modern tooling: Capital One, Goldman Sachs strats, defense technology companies like Anduril. These offer domain stability with technical currency.

The builder advantage

Academic evidence suggests that crisis periods structurally favor people who create new things over people who optimize existing systems.

Research by Manso and colleagues at UC Berkeley, published in the Review of Economics and Statistics (2023), found that firms systematically shift toward exploratory innovation during economic contractions — more new-to-firm inventors, more product innovation versus process innovation, and higher-quality patents. The mechanism: the opportunity cost of exploration is lower during downturns. Separately, research by Ljungqvist and Bias at the Stockholm School of Economics found that startups founded during the Great Recession were 12.1% more likely to survive to their seventh anniversary, with 35.2 percentage points higher employment growth.

The historical evidence is consistent. DuPont boosted R&D during the Depression, producing nylon and neoprene while competitors cut costs. Depression-era companies that built new things — Hewlett-Packard, Polaroid, Motorola — outperformed those optimizing in declining industries. The wartime mobilization demanded builders at unprecedented scale: the Manhattan Project employed approximately 130,000 people in entirely new disciplines; MIT’s Radiation Laboratory grew from 30 staff to nearly 4,000 building radar from scratch.

But the nuance matters. Not all building is equal, and not all optimizing is doomed. The evidence distinguishes four tiers:

1. Building new things in new domains — highest returns. AI infrastructure, government digital platforms, energy grid software, climate adaptation systems.
2. Building new things in old domains — strong returns. AI-augmented enterprise systems, rebuilding government services digitally.
3. Optimizing new systems — solid returns. AI inference performance, cloud infrastructure cost optimization, new platform architecture refinement.
4. Optimizing old systems — poor returns. Legacy CRUD maintenance, pre-AI codebase refactoring in declining enterprise sectors.

The key variable is not builder versus optimizer in the abstract. It is whether you are building or optimizing things the post-crisis economy will need.

The generational turnings lens sharpens this further. In Howe’s framework, the Hero generation’s defining historical role is to build the institutions of the next High. The GI Generation built the postwar institutional framework — the Interstate Highway System, NASA, the military-industrial complex, the corporate career ladder. Millennials are positioned to play an analogous role: building whatever institutional infrastructure the post-crisis era requires. In software, this is already visible in AI infrastructure, governance frameworks, platform standards, and developer tooling ecosystems. Gen X, in the Nomad role, builds the scrappy, functional systems that survive the crisis itself — the foundations that Hero-generation institution-builders scale in the next era. Homelanders, entering under maximum disruption, are crisis-forged apprentices whose entire 30-40 year career will play out during the next High — if the pattern holds.

I have been in this industry for 25 years. I have watched three cycles of “this changes everything” arrive and partially deliver. The honest read of the evidence is that the fourth cycle is structurally different — AI improves in ways that offshore teams and automation tools did not — but the career strategy is the same as it has always been: understand what the technology actually does, invest in the skills it cannot replicate, and move before the window closes.

What we do not know

Honest strategy requires honest uncertainty. Three questions remain genuinely unresolved, and career strategy should be robust across their possible answers rather than optimized for any single prediction.

Will cheaper software production expand demand enough to sustain employment? The Jevons paradox — the possibility that cheaper production expands total demand — operated for bookkeepers after spreadsheets and for bank tellers after ATMs. BLS projects 15% growth in software developer employment through 2034, consistent with the optimistic case. But the organizational-level evidence tilts toward “no net improvement” — Faros AI found no significant correlation between AI adoption and delivery improvements across 10,000+ developers. The Jevons paradox may operate for senior and specialized developers while collapsing for junior and generalist developers — demand expanding at the top while shrinking at the bottom. This dual outcome is consistent with the evidence. Career strategy should position for the expanding side regardless.

Will AI capability plateau or continue accelerating? SWE-bench scores went from 2% to 79% in two years. Expert timeline predictions for full autonomous coding keep being revised outward — the AI 2027 project moved its median from 2027-2028 to 2029-2030. And the gap between benchmark performance and real-world impact remains stubbornly wide: METR’s randomized trial showed net slowdowns for experienced developers on real codebases, though their February 2026 update flagged selection bias and called the result a likely lower bound. A sharp plateau would stabilize the “move up the stack” safe zone. Continued acceleration would shrink it. The practical reconciliation: treat AI resistance as a moving target, not a stable property.

Will a forcing event accelerate reallocation? Observable macro indicators — institutional trust at 70-year lows, geopolitical instability, fiscal stress at 101% of GDP — are consistent with a system under severe structural stress. Howe’s framework places us in the late stages of a Crisis turning and predicts resolution around 2030, followed by a new High — a period of civic renewal and institutional rebuilding that he is “super-bullish” about. Whether this produces a discrete forcing event or continues as a gradual grind changes the strategy: a sharp event rewards those already positioned in crisis-relevant sectors; a gradual grind rewards continuous adaptation. And the GI Generation analogue reminds us that the crisis itself — however it resolves — created the conditions for the largest infrastructure buildout and career expansion in American history. The evidence does not resolve the timing. Position for both.

The window is open. It is not permanent.

The window for strategic repositioning is open now but narrowing. Historical precedent shows that early movers in disrupted professions consistently outperformed those who waited. Market data shows the reorganization is already advanced — 58% of job postings now require AI skills (up from 26% eighteen months ago), and junior hiring has collapsed. Industry sector analysis shows that the most insulated positions require 6-24 months of domain investment to access. The developers who begin repositioning in 2026 will have meaningfully better outcomes than those who start in 2028.

The strategy is not complicated. It is the same strategy that has worked across every historical disruption in the research: invest in understanding over operation, in domain expertise over tool proficiency, in judgment over speed. Build a portable practice that does not depend on any single employer. Move toward sectors where software interfaces with the physical world, where regulation mandates human accountability, and where the entry barriers you invest in acquiring become the moat that protects you.

The disruption is real. The timeline is compressed. No one is coming to manage it for you. But the evidence is also clear that the developers who read the landscape accurately and act on what they see — not what they hope — have always come through. The ground is not disappearing. It is shifting. And the people who study the terrain before the ground moves have always been the ones who find solid footing.