The Overcut and the Real Economy
1. The Bill Rarely Lands Where the Cut Was Made
There is a version of this story where companies overcut, the strategy fails, executives learn humility, and labor is vindicated.
That version is emotionally satisfying.
It is also probably wrong.
There may very well be an overcut happening in software. I increasingly think there is. Large companies are cutting not only excess, but context. They are cutting people who understand systems, jurisdictions, customer obligations, architecture, compliance, institutional memory, and all the unglamorous details that keep large platforms from becoming expensive confetti.
But even if the cut is wrong, the company does not always pay first. Labor pays first. Communities pay first. Countries pay first.
Jack Welch stayed rich.
The communities hollowed out by the management theology he helped popularize did not receive the same severance package from history.
That is the part people miss when they wait for comeuppance. A strategy can be destructive and still enrich the people who executed it. A company can damage its long-term capability and still reward the executives who made the damage legible as efficiency. The bill may arrive decades later, after the bonus has vested, after the stock options have cleared, after the town has lost its floor.
This is why waiting for corporate karma is not a career strategy. Karma has a backlog, apparently.
2. Intel as Warning, Not Exception
Look at semiconductors.
For years, the assumption was that outsourcing electronics and fabrication overseas was rational. It lowered costs. It improved margins. It pleased investors. It let the United States keep the higher abstractions: design, software, branding, finance, platform control.
For a while, that seemed brilliant.
Then the geopolitical context changed.
Suddenly the same supply chains that looked efficient began to look precarious. Suddenly fabrication was not some dirty industrial afterthought. It was strategic capacity. It was sovereignty. It was defense. It was energy policy. It was industrial policy. It was the floor beneath everything else.
Now the United States wants fabrication back.
But bringing fabrication back is not like reinstalling an app.
There is no simple button marked “restore manufacturing base.” The tooling is gone. The supplier base is thin. The apprenticeship paths weakened. The tacit knowledge migrated. The ecosystem moved. You can spend billions and still discover that capital is not a substitute for accumulated industrial culture.
Intel is not merely a company story. It is a warning about what happens when a country treats deep capability as optional because financial abstraction looks cleaner on the balance sheet.
The attempted recovery is expensive, uneven, politically subsidized, and still not guaranteed. At times, it makes bankruptcy look less painful. That sounds dramatic until you remember how much money society is willing to burn trying to recreate what earlier leadership casually dismantled.
The lesson is not that semiconductors are uniquely hard.
The lesson is that ecosystems are hard.
And once an ecosystem is gone, rebuilding it is not a procurement problem. It is a generational problem.
3. Software May Be Repeating the Same Mistake
I think software may now be walking the same road, only faster.
The United States is treating software labor as if the work is mostly abstract, portable, and replaceable. The thinking seems to be: AI can generate more code, smaller teams can do more, global contractors can fill gaps, and the remaining elite layer can supervise the machine.
There is some truth in that. Enough truth to be dangerous.
AI really does compress the mechanical tier. It really does make some work cheaper. It really does allow fewer people to produce more surface area. It really does expose how much of modern software work had become scaffolding, glue, boilerplate, paperwork, dashboard polishing, and moving things a few pixels left because apparently that is what Prometheus stole fire for.
But the conclusion being drawn may be wrong.
If software labor is cut too aggressively, what disappears is not only output. What disappears is context density.
The people who know why the system works leave. The people who know why the exception exists leave. The people who remember the regulatory scar leave. The people who understand the customer promise leave. The people who know which part of the architecture is load-bearing leave.
At small scale, this may not matter.
At large scale, it matters enormously.
Large companies operate inside many local worlds at once. They deal with privacy law, labor law, tax law, accessibility requirements, export controls, content moderation rules, data residency, procurement restrictions, payment rails, local politics, language nuance, and the constantly shifting mood of governments that are increasingly tired of American platforms acting like foreign weather systems.
That is not just paperwork.
That is software context.
And context is not easily automated because context is not merely information. It is judgment shaped by proximity, memory, incentives, and consequence.
4. The Center of Gravity Moves
Even if this analysis is right, it does not mean everyone gets saved.
That is the uncomfortable part.
Manufacturing did not vanish when it left the United States. Electronics did not vanish. Semiconductors did not vanish. They moved. The demand remained, but the center of gravity shifted.
Software may follow the same pattern.
The work will not disappear. There will still be software everywhere. There will be more software, not less. AI itself will need software. Robots will need software. Energy systems will need software. Ports, ships, factories, farms, warehouses, hospitals, banks, schools, defense systems, and local governments will need software.
But the best software work may not remain centered where the abstraction is highest.
It may move toward where the physical systems still are.
This is the uncomfortable possibility: software follows the real economy. Not the pretend economy, not the asset-price economy, not the slide-deck economy, not the engagement-funnel economy. The real economy. The one with machines, ports, grids, factories, crops, ships, concrete, water, labor, weather, maintenance, energy, and failure modes that cannot be patched with a rebrand.
Countries that still own physical systems will have things to automate.
Countries that still build will have problems worth solving.
Countries that still maintain industrial capacity will need software to coordinate, monitor, optimize, govern, and defend that capacity.
The extraction may be smaller.
The impact may be greater.
That is not a bad trade if you care about the craft.
5. Abstraction Begets Abstraction
5.1 Goodwill Is Not Infinite
The United States still has enormous strengths. It has capital, universities, military depth, research institutions, cloud platforms, media reach, deep markets, and a long memory of being the place ambitious people wanted to go.
But goodwill is not infinite.
A country can ride its accumulated reputation for a long time. So can a company. So can a platform. So can a person. But reputation is a stored asset, not a renewable energy source. If you keep drawing it down without replenishing it, eventually even the most loyal people notice the account is empty.
Tourism, immigration sentiment, foreign student flows, trade relationships, and international goodwill all matter here. They are not separate from technological leadership. They are part of the atmosphere that lets leadership breathe.
If that atmosphere thins, the old software model weakens.
The danger is not only that jobs leave. The danger is that the United States gets trapped in debt and financialization, riding the coattails of its existing goodwill while the real economy beneath it keeps thinning.
5.2 The Leverage Addiction
Once financialization sets in, it is hard to reverse.
Abstraction begets abstraction.
Profit from leverage is addictive. It is cleaner than manufacturing, faster than apprenticeship, more scalable than maintenance, and easier to explain to investors. Why rebuild a hard industrial base when you can create a financial product? Why maintain a factory when you can maintain a valuation? Why own the messy floor when you can own the interface?
This is not simply a capitalist versus non-capitalist problem. It is broader than that.
Society itself becomes addicted to leverage.
Asset owners like it because numbers go up. Professionals like it because abstraction pays. Governments like it because financialized growth produces tax receipts without forcing a confrontation with deeper decay. Consumers like it because the system can feel wealthy even while the productive floor erodes beneath them.
5.3 What Economists Call the Beast
Economists and historians have names for pieces of this pattern: shareholder value, financialization, rentier capitalism, monopoly stagnation. The names differ, because apparently every discipline needs its own spellbook, but the direction is similar.
When productive investment becomes harder, slower, dirtier, or politically neglected, capital retreats into financial channels where returns are faster, more abstract, and less tied to broad labor participation.
That is the polite academic version of the trap. The country can look rich while losing the means to reproduce its own richness.
5.4 Arrighi and the End of Ascension
Giovanni Arrighi’s work is especially useful here. In his reading of long capitalist cycles, financial expansion is not simply a sign of strength. It is often the late-stage pattern of a hegemonic economy after material expansion becomes harder to sustain.
Production, trade, and industrial expansion create the original power. Then, as margins compress and competitors catch up, capital increasingly shifts toward finance. The old center can remain rich for a long time by managing claims on value rather than creating the next productive base.
That is the trap. Financialization can look like victory while it is quietly marking the end of ascension. The asset values rise. The financial sector grows. The professional class learns to speak the language of leverage. The nation still appears dominant because money still flows through it. But the floor beneath that dominance may already be moving elsewhere.
5.5 Why This Matters for Software
This is why the Arrighi frame matters for software. If software becomes another layer of financial abstraction rather than an applied tool tied to production, infrastructure, logistics, energy, governance, and maintenance, then it may participate in the same late-cycle movement.
It can remain profitable while becoming less generative. It can keep producing asset values while producing fewer durable pathways for labor. It can become brilliant at managing claims on the future while weaker at building the future itself.
The trouble is that leverage without a floor eventually becomes fragility.
The chart rises.
The country hollows.
Everyone calls it innovation until the supply chain breaks.
6. No Country Escapes the Floor Forever
No country that fully decouples from manufacturing has easily reversed decline.
That statement probably sounds too sweeping, and history always has exceptions, caveats, and annoying graduate students waiting in the bushes with counterexamples. Fine. Let them have their bushes.
The broader pattern still matters.
When a country loses manufacturing, it does not lose only factories. It loses supplier networks, applied engineering judgment, maintenance culture, tooling fluency, vocational dignity, industrial apprenticeship, and the daily contact between design and consequence.
Civil engineering weakens when nobody builds.
Aerospace weakens when production knowledge thins.
Semiconductors weaken when fabrication leaves.
Software may weaken when it detaches from the systems it is supposed to improve.
A purely abstract software economy can still be profitable. It can produce platforms, ads, financial interfaces, entertainment layers, subscription tools, and algorithmic optimization. Some of that is useful. Some of it is lucrative. Some of it is even brilliant.
But it is not the same as having a dense, applied software culture attached to physical reality.
The most interesting problems are rarely abstract forever. They eventually touch logistics, energy, housing, healthcare, transportation, manufacturing, food, water, defense, climate, and governance.
If those domains move elsewhere, the frontier moves with them.
7. The Borderless Myth
7.1 Software Looked Weightless
Some might argue that software is soft. Easy to replicate. Easy to export. Easy to run from anywhere. Easy to detach from place.
That was the old story.
But we are finding out that it is extremely not the case.
Software was never truly borderless. It only looked borderless during the brief period when American platforms, global capital, permissive regulation, cheap cloud infrastructure, and a relatively open internet all pointed in the same direction. That period trained people to believe software lived above borders.
Reality is now intruding.
7.2 Borders Are Back Inside the Internet
GDPR was one major signal. It applied European privacy rules beyond Europe’s physical borders when companies outside the Union offered goods or services to people in the EU or monitored their behavior. That was not a small patch. That was a declaration that digital activity could be pulled back into territorial legal authority. [1]
China’s Great Firewall was a different version of the same truth. It showed that a state could build a technical and regulatory boundary around cyberspace, selectively separating domestic internet life from the outside world. The ideology is different. The mechanism is harsher. But the lesson is related: the internet does not automatically dissolve sovereignty. Sovereignty can build walls inside the internet. [2]
Australia’s social media age restrictions sharpen the point again. As of December 10, 2025, age-restricted platforms must take reasonable steps to prevent Australians under 16 from creating or keeping accounts, with large civil penalties available against platforms that fail to comply. [3]
California is moving in the same general direction, though in a more American, lawsuit-shaped way. SB 976, the Protecting Our Kids from Social Media Addiction Act, was signed in September 2024. It targets “addictive feeds” and certain features for minors without parental consent, with age-assurance and parental-consent rulemaking due by January 1, 2027. [4]
The European AI Act adds another layer. The EU now treats AI not merely as software, but as a regulated system with risk categories, obligations for developers and deployers, and requirements around trustworthy deployment. [5]
7.3 Sovereign Stack, Sovereign AI
Then there is sovereign cloud and sovereign AI. The European Commission’s AI Continent Action Plan includes AI factories, gigafactories, and data-center expansion. Its Cloud Sovereignty Framework defines sovereignty objectives for cloud services, drawing on trusted cloud, Gaia-X, cybersecurity, NIS2, DORA, and national cloud sovereignty strategies. [6][7]
These are not isolated policies. They are signs of a broader patchwork forming.
The dream was infinite border, infinite intelligence, infinite mind. Software would flow everywhere. AI would reason everywhere. Platforms would scale everywhere. The world would flatten into an interface.
That dream is meeting law, geography, language, childhood protection, national security, energy constraints, compute scarcity, culture, and distrust.
In other words, reality has entered the chat, dragging several regulators and a customs form behind it.
7.4 Every Boundary Creates Context
This matters because every boundary creates context. Every jurisdiction creates local knowledge. Every regulatory regime creates implementation detail. Every sovereign stack creates architecture that cannot be treated as a generic deployment target.
The more fragmented the world becomes, the less useful generic software abstraction becomes by itself. The valuable work moves toward people and firms that understand the local boundary conditions: what data can move, where compute can run, what minors can access, how models must be audited, which clouds are acceptable, which languages matter, which regulators have teeth, and which political constraints cannot be wished away by a product roadmap.
Software looked weightless when the world temporarily allowed it to act weightless.
That permission is being withdrawn.
7.5 Ambition Is Growing While Context Bifurcates
This fragmentation runs directly against the idea that firms can keep cutting people while still exercising global ambition and influence. Reality is exacting a toll in context. The more jurisdictions split, the more rules diverge, the more sovereign stacks emerge, the more local politics matter, the more expensive it becomes to pretend one lean central team can understand the world from a dashboard.
Context is bifurcating while ambition is growing. That is the contradiction.
AI helps create visible surface. It can produce the document, generate the interface, draft the policy summary, translate the copy, scaffold the workflow, and make the product appear locally adapted. But it does much less to own context. It does not live with the consequences. It does not remember why one market rejected a feature, why one regulator cared about a clause, why one customer promise became sacred, or why one old architectural wart is actually a scar over a load-bearing wound.
Software is context.
What is efficient in one context can be terrible in another. An n^2 loop may be unacceptable in a high-scale service, perfectly fine in a bounded local workflow, or even desirable when clarity, auditability, and operational simplicity matter more than theoretical elegance. A feature that improves conversion in one country may create legal exposure in another. A compliance shortcut that works in one jurisdiction may be a business-ending category error somewhere else. There is no universal optimization function hiding under the couch.
7.6 Safe, Fast, Cheap, and Other Fairy Tales
This is where the old tradeoff returns: safe, fast, cheap. Pick two, as the joke goes, though big tech seems to believe it can nail all three with magic AI dust. Fast generation, cheap headcount, globally safe execution. Lovely fantasy. Very laminated.
But fantasy is reliable in one way: it eventually sabotages itself.
The world is getting more fragmented, not less. The local toll is rising, not falling. Every boundary creates context, and every context needs ownership. Cutting the people who own context while expanding global ambition is not efficiency. It is hubris, and hubris has a predictable ending.
History is rich with hints. GE, GM, Intel, and others were once so large, so central, and so apparently inevitable that decline seemed almost impossible from the inside. But they ate their own seed. They optimized the harvest while weakening the soil. They treated accumulated capability as a cost center, then discovered too late that the capability was the company.
8. The Check Valve
8.1 Push Factors at Home
If sovereignty abroad creates pull factors, the United States is also creating push factors at home.
This is not just about workers deciding to leave. It is also about making it harder for people to come in, harder for visitors to feel welcome, harder for students to plan their lives, harder for applicants to trust the process, and harder for highly skilled people to believe that the United States is still the obvious default destination.
In plumbing, a check valve is useful. It allows flow in one direction and prevents backflow. On a boat, a check valve can keep water from returning where it does not belong. Very useful in a bilge. Less charming as national talent policy.
The United States appears to be installing check valves across the talent system.
Tourism is one warning sign. Reuters reported that foreign visitors to the United States fell by 6 percent in 2025 even as global tourism spending rose, with anti-immigration policies cited as one factor steering travelers toward other destinations. International travel spending in the United States was also expected to fall by about 7 percent, or $12.5 billion, in 2025. [8][9]
Visa friction is another. In 2025, the State Department halted new student and exchange visitor visa appointments while preparing expanded social-media vetting, then resumed appointments with stricter screening requirements. Later reporting described proposed social-media requirements for foreign visitors that industry groups warned could have a chilling effect on travel. [10][11]
Permanent pathways are becoming more uncertain too. Reuters reported that the administration paused immigration applications, including green card and citizenship processing, for immigrants from 19 non-European countries. It also reported DHS guidance that past statements deemed extremist could trigger closer scrutiny for immigrants applying for green cards and naturalization, raising free-speech concerns. [12][13]
8.2 Brain Drain Becomes Ecosystem Destruction
This matters because brain drain does not become permanent only when people leave. It becomes permanent when the ecosystem that made their work productive begins to thin.
Hardware did not become expensive to manufacture in the United States simply because wages were higher. That is the lazy explanation, and like most lazy explanations, it is attractive because it fits on a bumper sticker and explains almost nothing.
It became expensive because the ecosystem diffused. Supplier density thinned. Tooling knowledge moved. Production managers, technicians, vendors, machine shops, repair paths, logistics relationships, and tacit process knowledge scattered across longer chains. The supply chain did not merely get cheaper overseas; it became more concentrated overseas. Once concentration moved, speed moved. Iteration moved. Practical judgment moved.
That is the part spreadsheets struggle to price.
Software talent can follow the same pattern.
If enough engineers, researchers, founders, operators, and domain experts stop seeing the United States as the obvious hub, the loss is not just individual. The context network starts to thin. The meetups fade. The labs weaken. The weird expert leaves. The junior who would have learned from that expert never gets trained. The startup that would have formed around that cluster forms somewhere else. The customer context, regulatory context, and industrial context start accumulating elsewhere.
Brain drain then turns into ecosystem destruction.
And once an ecosystem is destroyed, inviting people back is not enough. You are no longer asking talent to return to a dense, working circuit. You are asking them to rebuild the circuit from scratch, usually while pretending the previous generation did not burn the wires for quarterly margin.
9. Craft or Lifestyle
So what should a software worker do with this?
Do not wait passively for the old bargain to return. The wide American software middle may shrink. A small number of elite firms can do well while the labor market underneath becomes miserable.
The better question is: where is software still wanted?
Go where there is growth. Go where there are physical systems to automate. Go where the industrial base is being built, not merely monetized. Go where software attaches to ports, factories, grids, ships, clinics, farms, warehouses, public services, and local businesses that still need real tools.
Go where context is dense enough that generic AI cannot cheaply infer it from the global corpus.
This may mean Southeast Asia, East Asia, parts of Europe serious about industrial sovereignty, or sectors like infrastructure, energy, maritime, defense, robotics, manufacturing software, compliance tooling, and healthcare operations.
The old American software bargain let many people have both: high salary, high abstraction, high prestige, and a comfortable distance from the physical systems that made everything possible.
That bargain is fading, just like the American industrial complex, American microchip fabrication, and American electronics. One example can be dismissed as an exception. Two is a pattern. Three is a trend. If you have seen this pattern repeat across manufacturing, semiconductors, electronics, and now software, that is your cue.
Ultimately, this may become a choice: do you want to keep your craft, or do you want to keep your lifestyle?
Preserving the lifestyle may mean accepting more abstraction, more financialization, and more competition for fewer elite seats. Preserving the craft may mean moving closer to where things are still being built, even if the salary, title, or geography changes.
There may be an overcut in software. I think there probably is. But if history is any guide, the correction will not save everyone who was cut. It may not even save the country that allowed the cut. The center of gravity may simply move.
Software will survive. Software labor may not survive in the same place, at the same scale, or with the same bargain.
Follow the floor.
Follow the real economy.
Follow the places where software is still needed because reality is still unfinished.
References
[1] GDPR Article 3, Territorial Scope. The regulation applies to processing in the context of EU establishments regardless of where processing occurs, and can apply to non-EU controllers or processors offering goods or services to people in the Union or monitoring their behavior in the Union.
[2] Encyclopaedia Britannica, “Great Firewall.” Describes China’s Great Firewall as a regulatory and technological system used to monitor, filter, or block internet content and selectively separate Chinese cyberspace from the outside world.
[3] Australian eSafety Commissioner, “Social media age restrictions.” As of December 10, 2025, age-restricted social media platforms must take reasonable steps to prevent Australians under 16 from creating or keeping accounts; civil penalties can reach A$49.5 million for corporations.
[4] California Attorney General, “Protecting Our Kids from Social Media Addiction Act (SB 976).” Signed into law in September 2024, the Act targets addictive feeds and certain features for minors without parental consent, with rulemaking for age assurance and parental consent due by January 1, 2027.
[5] European Commission, “AI Act.” Describes Regulation (EU) 2024/1689 as the first comprehensive legal framework on AI, using risk-based rules for developers and deployers.
[6] European Commission, “AI Continent Action Plan.” Includes AI Factories, AI Gigafactories, and data-center capacity expansion as part of Europe’s AI infrastructure strategy.
[7] European Commission, “Cloud Sovereignty Framework.” Defines sovereignty objectives for cloud services and draws on European trusted cloud, Gaia-X, cybersecurity, NIS2, DORA, and national cloud sovereignty strategies.
[8] Reuters, “Fewer foreigners visited US in 2025 as global tourism spending rose,” January 14, 2026. Reports WTTC estimates of a 6 percent decline in foreign visitors to the United States in 2025 despite global tourism growth.
[9] Reuters, “Foreign travel spending in US to decline 7% in 2025, report says,” May 13, 2025. Reports WTTC estimate that international travel spending in the United States would decline by about 7 percent, or $12.5 billion, in 2025.
[10] Reuters, “Trump administration halts scheduling of new student visa appointments,” May 27, 2025, and “Trump administration tightens social media vetting for student visa applicants,” June 18, 2025.
[11] Reuters, “US social media requirements for foreign visitors could have ‘chilling effect’ on travel, industry group says,” December 15, 2025.
[12] Reuters, “US pauses all immigration applications from immigrants from 19 countries,” December 3, 2025. Reports pause of immigration applications, including green card and citizenship processing, for immigrants from 19 non-European countries.
[13] Reuters, “US DHS to vet immigrants for what it calls extremist views, raising free speech concerns,” April 27, 2026. Reports closer scrutiny for immigrants applying for green cards and naturalization based on past statements deemed extremist.