No Path Forward, No Way Back, Imperial Decline Edition
July 17, 2026
When American politicians state that terrorism is a threat, what they need is a mirror and an introductory economics textbook to understand that there are many different ways to destroy a nation. From NAFTA to financialization to crypto grifts, the US has been led from one economic catastrophe to another by people claiming to know what they are doing. While you and I know that AI isn’t going to cause 50% – 60% of the jobs in the US to disappear, what astonishes is that this is considered an acceptable outcome by Donald Trump and the US Congress.
For context, the US unemployment rate at the worst point in the Great Depression was 25%. This led to mass protests, a dramatic rise in the fortunes of organized labor, the White House being surrounded by former soldiers wanting to be paid, and WWII. Most governments begin worry about political instability when unemployment rises to 10%. That Trump and Congress heard the 50% – 60% unemployment projection and still thought AI a brilliant idea is testament to an absence of political accountability. There is something deeply broken in US politics.
Graph: surprise, the class war is real. Illustrated is that capital’s share of national income (GDP) has been rising for five decades as labor’s share has been falling. Oligarchs own the capital stock. Labor owns labor. When payments to capital rise, oligarchs benefit. When payments to labor fall, labor loses. As the graph illustrates, labor has been losing in the US for fifty years. Source: economycharts.
Fifty percent unemployment would instantly overwhelm state unemployment insurance systems. These systems divide their perpetually underfunded pools between the number of people eligible to collect, meaning that no one would receive enough in insurance payouts to survive. Every town, city and state affected would be in full-on economic collapse within a few months. Car loans, student loans and mortgages would no longer be paid. The US would be in full-scale economic collapse within six months.
The political effect of this should in theory be analogous to handing someone an empty gun, telling them that it is loaded, and then forcing them to shoot their spouse. The spouse will live because the gun is empty. But the relationship will have ended the second that the trigger is pulled. Trust will have been permanently broken. With the predicted job destroying impact of AI, Donald Trump and the US Congress pulled the trigger. They are fine with the nation disappearing in a puff of smoke as long as their stock portfolios keep rising. If this reads like ‘America first,’ please chime in.
No government would withstand 50% employment without massive social upheaval. This implies that the current US leadership has no stake in the US beyond what can be extracted through looting. Should this read as harsh, you probably aren’t paying attention. The productivity math rigs the capital versus labor game. Through what are called hedonic adjustments, the value produced by capital (e.g. AI) is systematically overstated. The largest possible value is taken away from labor and assigned to capital.
While Trump has long claimed that he opposed NAFTA, the question now is why? 50% unemployment from AI would be far more socially destructive than NAFTA has thus far been. The ‘capital investment’ theory whereby AI represents an investment in the broader economy falls apart above 8% unemployment. The math has it that at 50% unemployment, a 200% increase in productivity would be needed to avoid an economic decline from AI. Last year, with AI in the mix, this value was 3%. This is not going to happen.
Moreover, the gains from AI (if there ever are any) would be directed to capital while the costs would be borne by the newly unemployed. This is Class Warfare 101. Trump and the ‘communists’ (his term) in Congress see their fortunes being bolstered by destroying the US. But they have the math all wrong. They imagine that the stock market would rise in the midst of 50% unemployment. What they should be imagining is every major city in the US being burnt to the ground by newly immiserated citizens. As the saying goes, three days without food and no government will stand.
Graph: between 1948 and the end of 2025, the annual rate of Real (inflation-adjusted) GDP growth fell from 4% to 2%. While this might not read as problematic, the US hasn’t prospered outside of finance and tech heavy cities since 1980 or thereabouts (46 years). And financialization means that houses that cost $100,000 twenty years ago cost $400,000 today after ordinary (CPI) inflation has been taken out. This is the flip side of crypto grifts and national stock market obsessions. Source: St. Louis Federal Reserve.
The reason why the US is in this mess is because American leaders misread the unique position that the US was in at the end of WWII as evidence that capitalism actually works. What gave the US its economic prowess was geography, history and luck. By the end of WWII, industrial infrastructure abroad lay in ruins. Because the US was far away, lies between two oceans, and joined the war late, its industrial infrastructure was in better condition after the war ended than before it. To confuse this with grace, intelligence or hard work is a category error.
By the 1980s, the choice for the US was between manufacturing and asset stripping. The leadership at the time (Reagan) chose asset stripping (finance capitalism). With crypto grifts now having reached their natural limit, the US rediscovered war as a business plan— but without the manufacturing base needed to turn military production back into a business. When the political leadership represented the interests of the nation, the New Deal, public schools and a functioning healthcare system were possible. With crypto grifts, only more crypto grifts are possible.
That American capitalism has been redefined as using asymmetrical power to arrange circumstances such that guaranteed gains go to the few while the hollowed-out shell of a looted economy is left for the rest of us, is to empty the term of descriptive value. That this is a result of American imperialism turned inward brings to the fore the source of the social logic of looting. Having been educated in neoclassical economics, almost none of what is today called capitalism fits the theories of academic capitalism. In that version, a large and intrusive state is required to keep capitalism ‘capitalist.’
Monopoly power has long been the economic proxy for the political power of capitalism. In neoclassical theory, a powerful and intrusive state is needed to prevent concentrated wealth from closing the door on new competitors. Neoliberalism thus holds the paradoxical view that 1) economic power doesn’t exist but that 2) the purpose of the state is to support the interests of the economically powerful. The result is that economic ‘freedom’ now means the freedom to loot (e.g. crypto grifts) without legal consequences.
Readers may recall the ‘Great Powers’ framing of the contest for power when the US went to war against Russia in 2022. While the frame makes limited sense with respect to China and Russia, it misses that the US no longer ‘does’ the national interest. The evidence: the few dozen oligarchs whose interests Donald Trump represents are willing to see half of the US population be unemployed through the implementation of AI as long as their fortunes are intact. That they are set up to benefit while the rest of us are set up to lose illustrates the predatory nature of this willingness.
In fact, the oligarchs’ interests are antithetical to those of the US. Trump wants to permanently levitate the stock market because it is the source of the oligarch’s fungible wealth. Bartering a coal mine for goods and services is complicated and cumbersome. But selling shares in the coal mine to the public and then cashing in through stock options is easy, quick and rewarded with tax incentives (discounted capital gains tax rate). The richest 1% owns half of the stock market and the richest 10% owns about 90%. The stock market is central to continuing oligarch control.
To date, the promise of AI has been prospective, that if A, B and C happen, AI will produce a social benefit. Maybe. Maybe not. But the people who haven’t yet demonstrated a social benefit are already billionaires. The heads of AI company Anthropic have $15 billion apiece following a recent fundraising round. Do they have a profitable business? No. AI commenter Ed Zitron argues that Anthropic’s margins scale linearly. In other words, costs rise 1:1 with revenues. Implied is that the current trajectory provides no path to profitability. So, in capitalist terms, why are these people rich?
To be clear, the issue isn’t envy over fat wallets. It is concern over the aggregating impact of the mis-allocation of social resources. If the greed-heads want a sixth yacht, who would care except for the social power that their wealth gives them to force their self-serving decisions onto the rest of us. Automation AI and a wee bit of enterprise AI likely represent our shared future. So do the coming financial crash and its economic fallout from AI funny money (circular) financing. The problem with circular financing is that it sets off a game of musical chairs once one or more or the participating parties begins to take money out.
The relevant question is what it is that the American political leadership imagines that the US will do for a living in coming decades. The fantasy that has been sold is that ‘markets’ decide this. In the 1990s, outsourcing was claimed to be the product of nature (wage competition) when it was in fact the result of specific policies promoted by specific interests. American industrialists imagined that they could rid the US of organized labor by ridding it of jobs that pay. Now, with crypto grifts being what it is that Americans ‘do,’ the solution from above is political repression.
This political repression would have different social meaning if the sense was that those imposing it had the people’s interests at heart. But they don’t. We, the people. are viewed as annoying complainers who interfere with stock market gains and persistently rank the people running the place lower than venereal disease and serial killers. And the sense that one of America’s political parties will save us has been replaced with the understanding that nothing in the current political mix is going to do so. We are on our own.
Deindustrialization is a form of looting through Wall Street’s role in asset stripping and pirate finance. Financialization is looting through those closest to the money creation machine (banking) taking the money so created for themselves (e.g. crypto grifts). AI is looting through companies that haven’t turned a profit and plausibly never will receiving trillions in realized and prospective ‘investment’ without having produced a public benefit (like toilet paper or toothbrushes). The common theme here is fortunes made through uncreative destruction. US ‘leaders’ are destroyers, not creators.
Since the plan for what Americans will do for a living going forward at least theoretically involves firing half of the US workforce, how precisely is this supposed to work? The claim is that AI will replace enterprise (call centers, accounting firms, law firms) workers. Nonsense. These companies are creating tiered service where the customers who don’t matter (like yours truly, apparently) hang up and forget that there is a problem because there is no way to fix it except by ceasing to do business with the offending vendor.
AI is tiering precisely this way. Industrial automation and enterprise AI are already quasi-commodity businesses. The AI finance folks (e.g. Ed Zitron) argue that these businesses will never be profitable because their costs (‘compute’) rise with revenues, leaving margins flat. (With flat margins, a $1 rise in revenues means a $1 rise in costs and a $2 rise in revenues means a $2 rise in costs. With these companies currently losing money, rising revenues mean rising losses. The imagined solution is to cut costs, which finds OpenAI cutting prices on its products to stay competitive. But AI is a growth business?
This leaves only the models being sold as ‘thinking machines’ to produce the return on what will soon be several trillion dollars of investment. Are these thinking machines? It doesn’t help that the CEOs of the two main competitors in the thinking machine space, OpenAI and Anthropic, are philosophically ignorant fabulists who don’t appear to understand how the products that they are selling actually work. Anthropic’s Dario Amodei argues that finer gearing for the AI automaton will cause a conversion from dumb machine to human like consciousness. Are cookies set out on Christmas eve for Santa Claus as well?
But little of what these people believe actually matters. The trillions in financing are premised in all of these AI types earning huge profits for their developers. With the enterprise AI sales people backing off of the claim that AI will replace workers wholesale, it is difficult to see where the value proposition that would support fat margins will come from. Having worked for large corporations, it is guaranteed that the only reason that the AI sales people are getting meetings is on the promise that large numbers of workers will be permanently replaced with AI.
While the gearing in ‘thinking machine’ AI can be ever better refined, the product will always and in every case be a better automaton (mechanical robot), not a thinking machine. An analogy is to build a bridge 90% of the way without having a clear understanding of how the remaining 10% will be built. Continental philosophers explained a century ago both how and why building the final 10% (inanimate to animate) of the bridge is impossible. American AI companies built 90% of the bridge with no way to get to the other side. Brilliant?
Again, the reason why this matters is because the US has put all of its hopes for future economic growth into the AI basket. Capital expenditures of one trillion dollars to date, with another two trillion in theory coming behind it, put the wider economy at risk for what at present looks like a stupid gamble. Further, Chinese AI company DeepSeek has partially solved the cost problem, giving it a large pricing advantage over its competitors. As I understand it, DeepSeek’s architectural advantage over US competitors is across the board in automation, enterprise and generative AI.
What DeepSeek did process-wise was to turn a pig in a python into a mouse in a python until the pig hits the digestive tract, at which point it is turned back into a pig. This allows it to move through the python (AI model) with a low expenditure of effort to be revitalized when it is time for dinner (to produce the query response). Because compute is the major cost of the ‘thinking machine’ AI, this gives DeepSeek a cost advantage that US AI companies probably cannot match in the time needed to do so.
The DeepSeek solution is mathematically clever and likely points the way forward for the US AI companies that survive the next few years. But again, the finance people aren’t going to be in a position to care what happens after loans default and equity values go to zero. The time to pledge, borrow, and cross-collateralize US based AI was after the business case has been demonstrated. FOMO (fear of missing out) is a financial bubble ethos. As Will Rogers put it, once the bubble bursts, people stop worrying about the return on their money to focus on the return of their money.
Where does this leave the US? With 1) an implausible public – private industrial policy that will cause major economic dislocations over the next few years, 2) a commodity AI business that was sold as a high growth (high margin) business, 3) an all new group of entitled billionaires who have never created a product that earned a profit but who now get to tell the rest of us what to do and 4) with a Chinese competitor that built a superior product at lower cost than US firms. As with BYD and the $10,000 EV, who needs a Tesla when you can buy a car that works better for 15% of the cost.
Graph: deindustrialization has been the articulated policy of the US and was codified through serial trade agreements since the 1970s. When it finally took hold around 2001, its impact on the US workforce was utter devastation. In 2016 Donald Trump was elected on his promise to reindustrialize the US. But Trump has the attention span of a gnat and was likely sidetracked by a cheeseburger placed in front of him. The result: manufacturing employment in the US remains about where it was in 2016. Source: St. Louis Federal Reserve.
For those imagining that the US is re-industrializing, think again. Manufacturing employment in the US has fallenunder Trump, not risen. Readers who search will find a body of literature from mainstream economists claiming that the US moved up the value chain in recent decades and that US manufacturing is flourishing. This is (Mitt) Romneyesque fantasy. See my recent Substack articles on the ‘China Shock’ for details. The claim about manufacturing is based on hedonic (qualitative) adjustments made to computer chips. Economists confused a dubious financial statistic (hedonic adjustments made to computer chips) with wider productivity growth.
The math has it that valuing the hedonic adjustment at 100% of the increased functionality while users only use 15% of it overstates the increase in the productivity of capital on its own terms. When applied to AI, assuming that the full theoretical change in the value of capital is 100% of the hedonic adjustment, and therefore allocating 100% of productivity gains to AI, is to overstate the change in productivity from AI (15%) and to understate the change in productivity of labor (85%) in the production process. This same point applies to all technology shocks. The game is rigged for capital.
The fall in US GDP growth from 4% in 1948 to 2% today (graph above) is why the US is saber rattling against China, Russia and BRICS. The punchline is that China, Russia and BRICS didn’t deindustrialize the US. The American ruling class did. China, Russia and BRICS didn’t financialize the US. The American ruling class did. China, Russia and BRICS didn’t force the crypto grift economy onto the US. The American ruling class did. And it is the American AI developers who made generative AI a tool for top-down social control. China is using AI to automate factories. Which nation is authoritarian?
To tie this together, AI seems a weak foundation upon which to build a future economy. In the first place, unlike computers, telephones and the internal combustion engine, AI won’t be transformative. What it does well is digital automation. What is does less well and even then, only in particular industries, is enterprise automation. The executives who imagine that AI will replace workers know absolutely nothing about what ‘their’ workers actually do. The US is to the point where class stratification is making the nation unworkable.
Looking forward, expect and prepare for an AI related financial calamity within the next year or two. This will be the result of AI financing, not AI per se. The catalyst could be the economic consequences of the re-shutting of the Strait of Hormuz. Or it could be the implosion of AI financing in the presence of shrinking promises about what AI can do. Either way, the US has been placed very, very far out on the limb of a dying tree. Trump the destroyer is the product of the American turn toward internal predation. My American neighborhood resembles DRC through economic extraction by external forces.
What are the solutions? The re-localization of agriculture. Rationalizing the consumer economy to match social and environmental constraints. A Federal guarantee that all who are willing to work be provided jobs that pay doing the work that the US needs to be done. Public schools from pre-K – PhD that educate the public in the broad attributes of human civilization. A healthcare system that raises life expectancy instead of lowering it. And a Department of Peace with the authority and purpose of ending US military adventurism abroad. Murder is the business of the mafia. It’s time to get the gangsters out of government.
