Back Research Notes QE for the Mind: How Artificial Intelligence Is Flooding the Economy with Intelligence Liquidity Published on November 3, 2025 By Jordi Visser The New Liquidity Engine “Coding is no longer a tool, it’s more a substitute for wages,” Microsoft CEO Satya Nadella said that on the BG2 podcast with Brad Gerstner over the weekend. He was describing how artificial intelligence has moved from assisting workers to replacing entire layers of labor. As Nadella later said, “The reality is agents are the new seats.” Those observations capture the essence of what’s unfolding across the economy right now with knowledge workers: productivity itself has become programmable. AI isn’t just an efficiency upgrade, it needs to be thought of as a new form of liquidity, where software replaces headcount and automation and becomes the balance-sheet equivalent of capital. The same way Quantitative Easing (QE) once lowered the cost of money, AI now lowers the cost of work. For more than a decade, corporate America lived off manufactured growth. First came Quantitative Easing (QE) and Zero Interest Rate Policy (ZIRP) , monetary experiments that flooded the system with cheap capital. Companies learned they could boost earnings not by producing more, but by borrowing more. With debt nearly free, financial engineering replaced productivity. Firms issued bonds, repurchased shares, and refinanced liabilities to lift Earnings Per Share (EPS). Even when revenues stagnated, shrinking the denominator (shares) and lowering interest expense inflated the numerator (net income). The decade after GFC was very challenging for investors who chose to fight the Fed. Whenever there would be signs of a liquidity fire, the Fed would ride in to extinguish it with a firehose of QE. This will end badly was a common chorus heard during the period. In a similar fight today, economists, strategists and AI bubblelistas continue to try and play down the adoption of AI and call for a bubble. That chorus of ending badly is back. The similarities between AI and QE go deeper than a two-letter acronym. Artificial intelligence has emerged as the next great liquidity engine but instead of monetary liquidity , it’s operational liquidity . Where QE lowered the cost of capital and ZIRP lowered the cost of borrowing, AI lowers the cost of doing business. C.H. Robinson’s CEO Dave Bozeman captured the shift perfectly on its Q1 2025 earnings report: “We’re not waiting for a market recovery to improve our financial results … This includes continuing to arm our industry-leading talent with innovative tools that help us materially elevate the customer and carrier experience. We are innovating to harness the power of artificial intelligence and driving automation across the full lifecycle of a load, which gives our customers better service, while also helping us improve our performance by automating tasks that free up our talented people to work on more strategic and higher-value work.” By automating thousands of logistics decisions and streamlining workflows, the company achieved double-digit expense reductions in a weak freight market, proof that margin expansion can now come from AI leverage instead of debt leverage . The never-ending skeptics ask how AI revenues will ever justify the spending spree from GPU data centers to enterprise pilots. But that question misses the point. If AI’s gains only come from cost cutting, not new sales, the EPS effect is far greater due to operating leverage but isn’t counted by bears as AI revenues. Every dollar saved flows almost entirely to the bottom line, while each new dollar of revenue is diluted by its own costs. Hence the multiplier effect driving the market right now through rising profit margins. In a world where growth is scarce but margins are king, AI doesn’t need to sell, it only needs to save. Those savings compound like QE-era liquidity, producing the same outcome: accelerating EPS growth without expanding demand. From Zombie Survival to Darwinian Efficiency QE and ZIRP didn’t just rescue the economy after the Global Financial Crisis, they suspended capitalism’s natural selection . By suppressing interest rates and flooding markets with liquidity, policymakers shortened the recession and prevented a deeper corporate washout. But that same lifeline also kept countless zombie companies alive, firms whose profits never covered their interest expense (like the U.S. Government), yet could refinance indefinitely at near-zero cost. Those balance-sheet zombies became artifacts of the financial-engineering era. They survived not because they innovated, but because money was free. Debt was rolled, maturities extended, and investors chased yield in anything that moved. The system became addicted to low-cost credit and artificially high valuations. Bears were frustrated calling the market a bubble back then as they fought QE. AI changes that equation. The new liquidity isn’t financial, it’s intellectual , and it can’t be borrowed. You can’t issue a bond to buy productivity the way you could buy back stock. AI rewards companies that can adapt their processes, not those that can roll their liabilities. The firms that thrived in a zero-rate world by engineering EPS through leverage now face an economy where leverage no longer matters unless it funds intelligence. This is why bankruptcies will rise while the economy remains strong and stocks remain high. This is why there will be many 52-week lows at the same time as 52-week highs. The zombies are finally leaving. In other words: QE prevented failure; AI enforces evolution. ZIRP subsidized debt; AI subsidizes efficiency. Liquidity once prolonged the weak; intelligence now strengthens the strong. The same easy money that once masked operational decay now becomes a liability. Debt-laden firms, built for an age of financial engineering, must now compete against AI-enabled rivals whose cost structures self-improve every quarter. The Mathematics of Synthetic Growth Under QE and ZIRP, companies used cheap money to engineer earnings. The playbook was simple: borrow at 2%, buy back stock yielding 6%, and watch EPS rise even if profits stood still. That was financial leverage. Today, AI achieves the same outcome through operational leverage. Instead of lowering interest expense, AI lowers labor and process expense. Instead of reducing shares outstanding, it reduces the denominator of the cost structure. Both inflate EPS, but AI’s version is stronger because the marginal cost of software is near zero. A mere 5% reduction in expenses yields a 20% EPS jump. To reach the same lift through revenue, sales would need to grow roughly 13–15%, a far harder feat. Every incremental dollar of revenue brings new costs; every dollar of savings flows to profit. That’s the hidden leverage inside AI adoption. It’s the mirror image of QE’s debt-driven era: instead of creating synthetic EPS through the balance sheet, firms are now creating synthetic productivity through the income statement. C.H. Robinson’s transformation illustrates the flywheel: automation and AI-assisted pricing tools are reducing manual labor hours and optimizing freight matching turning cost savings into a compounding EPS engine. AI’s compounding effect on margins functions as QE for the income statement , a self-reinforcing liquidity loop that amplifies earnings faster than traditional growth ever could. Now the bears are choosing a more powerful two letter foe than QE; AI. From CapEx Anxiety to Margin Reality For the past year, investors have obsessed over AI’s price tag. Record data-center buildouts and hyperscaler CapEx have sparked one question: Where’s the revenue? Analysts scour earnings calls for monetization updates, assuming, as in past tech waves, that payoffs appear on the top line. That framework misses the defining feature of this cycle: AI’s value begins with expense contraction , not revenue expansion. In previous decades, QE and ZIRP justified high equity mu