The Federal Reserve closed 2025 by cutting rates by a quarter point for the third time this year. The decision was not unanimous: nine of the board members voted in favor while three voted against the rate cut. The decision was made as the US economy continues to reel from tariffs, mass deportations, cuts to government spending, and stalled hiring. To make matters worse, official price and labor market data remain murky since the government shutdown.

On the other hand, the US economy expanded at the fastest pace in two years, increasing 4.3 percent. This has produced a confusing picture of the real state of the economy. However, things become clearer when we take into consideration the fact that much of this growth is driven by spending in the AI sector.

On its face, the US economy appears pulled in two directions: on one side, severe inequality, inflation, slowed hiring, and high youth unemployment; on the other, a seemingly unstoppable tech sector, wealthy stockholders responsible for a disproportionate share of consumption, and expanding federal spending on defense and border enforcement technologies reliant on artificial intelligence.

In the words of a Brookings Institution economist, we have a “two-track economy.” The “AI gold rush generates excitement and papers over a drift in the rest of the economy.” According to the Budget Lab at Yale, investment in AI accounts may reach 2 percent of the United States’ gross domestic product (GDP) this year, or the equivalent of $1,800 per person. Analysts at Deutsche Bank have gone as far as to argue that the United States would be close to a recession  this year if it weren’t for tech-related spending.

But the image of a two-track economy oversimplifies the reality. These two halves, which may initially seem distinct, exist within the same system of uneven capital accumulation, and they are held together by access to cheap credit. The “winning” or “growing” half relies on the extraction of rents from the other. As such, Donald Trump’s repeated pressure to cut rates takes on deeper significance.

First, despite the undeniable pace of technological progress and capital expenditure, the gains from AI innovations are concentrated within an increasingly enclosed system. Tech giants monopolize access to scarce or irreproducible resources, from water and the power grid to human social relations. The owners of these companies act simultaneously as landlords of digital terrain and utility providers.

This allows them to extract surplus rent, while the users and workers who train the systems and generate data receive none of its returns. They face precarious work conditions, layoffs, surveillance, and exclusion from the systems they help create. The AI economy is not bifurcated but reliant on the dispossession of low-wage workers and the precarious labor markets in which they are embedded.

Skeptics of the AI gold rush often say the bubble will burst and that the speculative finance that fuels it is crisis-prone and parasitic. They point to the sector’s extreme concentration and circular financing as signs of its fragility. The seven largest tech companies are responsible for 60 percent of the gains in the S&P 500 this year, and actors in the AI economy increasingly fund, supply, and sell to one another in a closed, mutually dependent loop.

For example, Oracle, Nvidia, CoreWeave, and SoftBank trade $1 trillion worth of AI deals among themselves, and Amazon is in discussions to invest $10 billion in OpenAI, which signed a $38 billion deal with Amazon Web Services this November. Writing in the New York Times, economist Natasha Sarin has said that this “situation is worse than having all your economic eggs in one basket. It’s closer to putting all your eggs in one basket and stomping on all other baskets.”

However, it is wrong to dismiss finance so easily as self-destructive and external to production. Even if specific investments fail to yield profits, the US financial sector has shown the ability to repeatedly adapt and innovate to give these companies endless access to credit or render their debt sustainable.

The $29 billion financing package for Meta’s data centers exemplifies this dynamic. Global asset managers Blue Owl and PIMCO combined debt and equity into a special purpose vehicle that funds the project while keeping the debt off Meta’s balance sheet. Meanwhile, tech companies and their financiers alike celebrate Fed rate cuts, which pushes up stock prices and lowers the costs of borrowing further.

Households participate in the same debt-driven system. Facing inflation, layoffs, and stagnating wages, they take on debt to sustain consumption and withstand the rising cost of living. Finance acts as the connective tissue: preserving profitability for some, while embedding labor extraction and inequality for others. It is a disservice to understand this as peripheral to the economy’s functioning, when it is central to its reproduction.

When these dynamics deepen political, ecological, and social instability, tech giants turn to the state to secure their position and rents. In addition to monetary policy, the state funnels taxpayer money into public-private partnerships, many of which are AI technologies. This occurs across federal and state governments but most notably in defense and security spending.

Tech firms also benefit from permissive regulations and targeted exemptions. For example, the AI half of the economy is exempt from Trump’s tariffs. A recent article in Bloomberg described the president as having carved out a “VIP AI express lane where no one is hitting the brakes because there are no tolls.” This growing interdependence between tech and the state feeds state repression and retrenchment and further undermines the conditions necessary for working-class solidarity and resistance.

Understanding the economy as a single system helps clarify the resilience of the current arrangement and the stakes of monetary policy decisions. We cannot be distracted by narratives of a two-track economy or the sudden collapse of a bubble. Instead, we must understand how this growth actively extracts from middle- and low-income households. AI investment is not holding up the fragile half of the economy; it is one of the drivers of its underdevelopment.