After the Selloff: Has the Market Become Too Dependent on AI?

After the midweek market rout, the Financial Times reported:

"US stocks shed $3.2tn in a day of chaotic trading which saw the tech-heavy Nasdaq 100 plunge 6.4 per cent after Democrats' sweeping midterm election victory upended Wall Street's mammoth AI bet... The market value drop on the S&P Composite 1500 index, a broad basket of America's publicly traded companies, was the biggest since the turn of the millennium... Selling among chipmakers was brutal, with Nvidia, the company most closely associated with the AI surge, tumbling 9.3 per cent, pushing its market value down more than $400bn."

Reading this reminded me of a post-2026 Midterm Election conference I attended, where several market leaders shared their views on the sharp decline in equity markets. One speaker argued that a significant portion of the S&P 500's recent performance has become tied, directly or indirectly, to AI-related investment, particularly in the semiconductor industry.

That observation struck a chord with me. The AI sector has become deeply interconnected with broader corporate spending, investment plans, and market expectations. As a result, I expect that weakness in AI-related industries will not remain confined to a handful of technology companies. In my view, a sustained slowdown would likely spill over into broader market sentiment and potentially affect other sectors as well.

Looking ahead, I suspect policymakers will face increasing pressure to address public concerns surrounding AI, including issues related to privacy, employment, and safety. One possible response would be the development of regulatory frameworks intended to provide greater oversight and clearer rules for businesses operating in the space.

The challenge is that meaningful regulatory efforts often take time. Policymaking, implementation, and compliance do not happen overnight. Markets, however, can move much faster than the political process. If investor enthusiasm for AI weakens significantly before greater regulatory clarity emerges, the adjustment could be painful.

That does not mean an AI-driven downturn is inevitable. However, as AI has become one of the market's primary growth engines, investors will continue to pay close attention to the risks associated with that dependence. The question is not whether AI will remain an important technology. The question is whether current expectations for AI-related growth are sustainable.

The 2026 Midterms: Why Economic Reality Still Matters

As of November 5, 2026 (~5pm), the midterm election results have yielded a net forty-seat increase in the US House of Representatives for the Democratic Party, which puts their total at 252 and 155 for Republicans. There are still 28 seats remaining to be decided. In all likelihood the Democratic Party will end up with a seat count somewhere in the mid-260s after all votes are counted. A similar pattern emerges out of the US Senate: Democrats have secured 54 seats, a gain of seven seats, compared with 44 seats for Republicans. There are 2 seats remaining to be decided. That means that the Democrats will control what happens in Congress. Their legislative agenda should pass with relative ease through Congress, as all they need is a simple majority to pass bills. But when it comes to overriding a presidential veto, they need a two-thirds majority in both chambers (290 votes in the House and 67 votes in the Senate). Therefore, the hurdle increases significantly to achieve the override.

Against this background, it was not surprising that US stock markets went up a lot on the day after the election (e.g., on November 4, the Dow ended the day at 53,356, up 1,520 from the previous day (November 3); and the S&P ended the day at 7,688, up 120 from the previous day). I believe that was the result of expected gridlock (i.e., nothing of meaningful economic consequence will occur) in Washington DC over the remaining presidency of Donald Trump. I also think markets expected that the Iran conflict would materially subside, leading to a continued decline in fuel prices, and thus taming inflationary pressures and future interest rate increases by the Fed. Or so goes the hypothesis, because today (November 5) the Dow, the S&P500, and the NASDAQ closed deep in the red at about 3.45%, 4.47%, and 5.52% respectively.

The equity market’s behavior could be one of those where you “buy the rumor” and “sell on the news”. Perhaps. But I don’t think so. Granted, it has only been a couple of days since the midterm elections occurred, but I view this as a harbinger of a broader recognition that there are deep structural economic imbalances that no Democratic or Republican agenda can fix in one election cycle. Simply examine the US federal debt and you will see that the upward trend has happened across multiple administrations under different parties.

The Austrian view of the economic cycle, which I lean toward, says that artificially low interest rates lead to a misallocation of capital across the economy. That misallocation is what ultimately creates the boom and bust economic cycle. The artificially low interest rates over most of the 2010s, coupled with the massive increase in liquidity after the COVID pandemic, have led to a massive misallocation of capital. In other words, projects that should not have taken place occurred simply because the cost of capital was relatively low.

Conclusion:

In my view, the Democratic Party's control of the US Congress should not automatically be interpreted as a mandate to promote its political agenda. The midterm election results reflect widespread voter dissatisfaction with the Trump administration. I am deeply skeptical that the Democratic Party will do anything meaningful to try to meet the expectations of the voters. I expect they will pursue partisan agendas that will not help steer the economy to where it should be. The truth is that all grand plans must meet economic reality. That is why I like to analyze what is happening in the economy. There are significant economic concerns, whether they are the size of the federal debt, massive increases in AI capital expenditures (capex), or decades-low consumer confidence, all of these point to profound structural issues whose gravity I doubt politicians fully grasp, or about which they have limited ability to do much. But I could be wrong. I hope I am.

Why the 2019 Repo Shock Happened and Whether It Could Happen Again

The repo market is the financial system’s plumbing, a multi-trillion-dollar market where banks, hedge funds, and dealers lend cash in exchange for collateral. In September 2019, this plumbing suddenly clogged, sending short-term borrowing rates spiking from 2% to 10% in a matter of hours. Take a look at the Secured Overnight Financing Rate (SOFR) for the month of September 2019:

Today, many of the underlying structural pressures that led to the 2019 crisis have returned, but with one critical new safety valve installed by the Federal Reserve.

What Really Happened in September 2019?

The 2019 spike was not caused by wild panic or an economic recession; it was caused by rational, regulation-driven behavioral gridlock:

A Perfect Storm Drained Cash: A major corporate tax deadline and a surge in U.S. Treasury debt auctions pulled over $150 billion in cash out of the banking system in just a few days, right as the Fed was winding down its bond holdings (Quantitative Tightening, or QT).

Dealers Needed Cash: Wall Street primary dealers were flooded with new Treasury bonds and urgently needed overnight cash to finance them.

The “Hoarding” Paradox: The four largest U.S. banks held plenty of cash, approximately $1 trillion at the Federal Reserve. Normally, banks would eagerly lend that cash into the repo market to earn high interest. However, they refused to lend even at 10% return.

Why Banks Held Back: Internal risk rules and post-2008 banking regulations (such as resolution liquidity rules and strict leverage limits) required bank treasurers to keep a hard minimum cash buffer directly at the Fed. Crossing below that limit would trigger regulatory alarms.

The combination of all these factors led to what is called a “fallacy of composition”. In other words, each bank acted rationally by protecting its own cash cushion, but their collective refusal to lend caused the entire overnight lending market to freeze until the Fed stepped in to provide cash directly.

Are We At Risk Today?

Today’s financial plumbing exhibits several unmistakable similarities, along with a few vulnerabilities:

Vulnerability Driver	What Happened in 2019	What is Happening Today
Fed Balance Sheet Run-Off (QT)	QT had been draining excess cash down to a critical tipping point.	QT and massive U.S. Treasury debt issuance have drained the Fed’s cash buffer (Overnight Reverse Repo Facility) from $2.5 trillion to near zero. Questions remain, however, whether reserve levels have fallen enough to approach banks’ minimum comfort level.
Record Supply of Treasury Bonds	Heavy Treasury issuance clogged dealer balance sheets.	Multi-trillion-dollar federal deficits are forcing dealers to warehouse even larger volumes of government debt.
Hedge Fund Leverage (The Basis Trade)	Present at moderate levels	An estimated $1.2 trillion Treasury cash-futures “basis trade” relies on overnight repo borrowing, creating a massive daily demand for cash. And in a period of constrained balance sheets and insufficient cash supply, we’ll likely see a severe funding crisis.
Bank Caution on Cash	Banks guarded cash due to regulatory requirements.	Following the 2023 regional banking turmoil (e.g., Silicon Valley Bank) bank treasurers may be more cautious about maintaining ample liquidity buffers.

While the pressures on overnight cash are just as high today as in 2019, the Federal Reserve introduced a safety valve in 2021 by what is called a Standing Repo Facility (SRF), which allows primary dealers and banks to exchange Treasuries for cash directly from the Fed at a pre-set maximum rate.

Conclusion

The transition toward tighter reserves, combined with massive government bond supply and leveraged hedge fund demand, increases the likelihood of generating periodic volatility and upward pressure on short-term rates. As 2019 demonstrated, having hundreds of billions in cash across the banking system does not guarantee that cash will flow where it is needed, especially when regulations penalize banks for deploying it.

While the Fed’s SRF should mitigate extreme spikes (like the 10% clip we saw in 2019), non-bank financial players (like hedge funds) cannot access the facility directly. I suspect pressure will eventually mount to broaden liquidity access beyond traditional banking institutions. If such support ultimately requires a significantly larger Federal Reserve balance sheet, concerns about long-term dollar purchasing power could become increasingly relevant.

Economic Risk Dashboard: What the Numbers Suggest as of October 30, 2026

Here I am showing what the Economic Risk Dashboard suggests based on data as of October 30, 2026. The data is sourced from the St. Louis Fed FRED database, and I used my proprietary Python program to extract and analyze the data, as shown below. For the conceptual foundation of the dashboard, see my previous post where I explain the framework in greater detail. 

Our macro assessment remains largely unchanged from when we last examined the dashboard on October 2, 2026: The dashboard presents a mixed but generally stable picture of the US economy. At present, the dashboard suggests an economy that is not exhibiting broad-based stress, but one in which consumer sentiment remains exceptionally weak.

There is no change to household health (remains solid), with unemployment at 4.2% and no evidence of deterioration in either the six-month or twelve-month trend. The October 2026 unemployment rate will be reported on November 6, 2026. The consensus forecast is for the rate to remain unchanged at 4.2%, so we don’t expect any meaningful update from what we currently see in our dashboard. Having said that, we will analyze those results to observe how it stands vs. our dashboard.

Credit markets also continue to function normally, as high-yield spreads, investment-grade spreads, commercial paper spreads, and bank lending standards remain well below stress thresholds.

The one area that continues to warrant closer attention is consumer confidence. The University of Michigan Consumer Sentiment Index remains near multi-decade lows, even as other indicators remain stable.

Fed Pauses, Markets Expect Hikes: Will Tighter Policy Reduce Inflation?

As I expected and discussed in a previous post, the Federal Reserve announced on October 28, 2026, that it would maintain its policy rate unchanged. Specifically, in a 10 – 2 vote:

“The Committee decided to maintain the target range for the federal funds rate at 3-3/4 to 4 percent, in support of the Federal Reserve's dual mandate. The Committee is continuing its policy of maintaining ample reserves in the banking system.” 

The next Federal Open Market Committee (FOMC) meeting is scheduled for December 9, 2026. According to market-implied probabilities from the CME FedWatch tool as of October 28, 2026, investors assign roughly a 79% probability that the FOMC will raise its policy rate by 25 basis points at its December meeting.

       

Looking ahead to the January 2027 meeting, investors appear to assign roughly even odds to an additional 25 basis point increase. Current pricing also suggests expectations that rates may remain near that level for much of 2027.

 

Conclusion

The financial markets expect the policy rate set by the FOMC to increase 50 basis points (half of a percent) in the next 3 to 6 months. Whether those rate increases will help 'ensure a timely return of inflation to the Committee's 2 percent goal' remains an open question. My view is that policy rate increases will not help because the elevated inflation we are seeing now has to do fundamentally with a supply shock in energy prices. While those pressures may spread into other sectors of the economy, higher policy rates cannot create the additional energy supply the market demands.

What the Fed's decision does introduce is the risk of a policy error. If inflation is primarily supply-driven, further tightening could unnecessarily weaken economic activity while doing little to address the underlying cause of inflation. In that scenario, policymakers may inadvertently contribute to the very economic slowdown they are attempting to avoid.

Indeed, an argument can be made that the Fed's concern is not necessarily the first-round impact of higher energy prices. Rather, their concern is that households and businesses adjust their expectations, wages, and pricing behavior in response, causing inflation to become embedded throughout the economy. In that case, one could argue that rate hikes may not increase energy supply, but they could still help contain broader inflationary pressures. While that is an excellent point, we must remember the mechanism through which such an outcome would likely occur: slower economic growth. In that sense, the effort to contain inflation may itself contribute to the economic weakness policymakers hope to avoid.

A Mountain of Junk Debt Is Coming Due: Should We Be Concerned?

In a recent Financial Times article, it was mentioned that,

"Almost $2tn worth of speculative-grade debt and loans reach maturity in the next five years, according to Moody’s Ratings, of which $438bn is due by the end of 2028. That might seem like a long way away, but finance departments like to get this sort of thing sorted in plenty of time: by the middle of next year, even a late-2028 maturity will feel uncomfortably close."

And to give you perspective, the FT provided this graph:

As I have mentioned in other posts, the era of easy credit has fueled the large increase in junk debt, almost doubling over the last decade. The FT highlights the concern that much of this debt will likely need to be refinanced at higher interest rates, putting pressure on corporate profitability.

It is difficult to say how this increase in financing costs will affect the viability of individual firms, but certainly I expect some of the companies that make up this population will find themselves in economic difficulties and possibly default. Examining the delinquency rate on business loans in the graph below, which I view as an early warning indicator of corporate defaults, we can see that it remains relatively low compared with most of the past 30 years. While this metric does not capture all the speculative-grade debt referenced in the FT article, it is nevertheless a good holistic measure of default risk.  

Having said that, historical data shows that those rates can change rather quickly and suddenly.

Conclusion

While delinquency rates remain subdued, history suggests that can change rather quickly. The fact that a large amount of junk debt must be refinanced over the coming years should concern us because of the fragility it could create within the economy. If borrowing costs remain elevated, some issuers of junk debt may struggle to meet their obligations when refinancing becomes necessary. It starts with one company, then another, and then another. Before long, a broader credit event can emerge, with the potential to spill over into the wider economy. Before that happens, however, I would expect delinquency rates to begin trending upward. We are not there yet. But "yet" is the operative word.

The End of Easy Money? Reflections on El-Erian's Warning

Mohamed El-Erian, Practice Professor at the Wharton School, Chief Economic Advisor at Allianz, and Chair of Gramercy Funds Management, recently argued in the Financial Times that the surge in European government bond yields is largely an imported shock from US economic volatility. He warned that European governments cannot simply wait for the US economy to stabilize, as an American slowdown could trigger a broader hard landing with severe consequences for Europe. Crucially, El-Erian emphasized that the era of an “artificially suppressed global rate environment is over.”

That statement underscores, in my view, a profound structural inflection point in global macroeconomics. For nearly fifteen years following the 2008 Great Financial Crisis (GFC), major central banks suppressed policy rates near zero (so-called Zero Interest Rate Policy, or ZIRP) and expanded balance sheets through large-scale asset purchases (Quantitative Easing). ZIRP resulted in anchoring the front end of the yield curve, thus lowering short-term borrowing costs for banks, businesses, and consumers. And by purchasing trillions of dollars of long-duration US Treasury bonds and Mortgage-Backed Securities (MBS), central banks aimed to lower the mid-to-longer end of the yield curve by absorbing duration risk (the price sensitivity of bonds to changes in interest rates) from private markets.

By compressing term premia (i.e., the extra compensation investors demand for bearing duration risk over time, rather than rolling over short-term debt) and flooding markets with liquidity, central banks successfully lowered the cost of capital across the sovereign and corporate yield spectrum.

To illustrate the scale of this liquidity expansion, let’s examine the Federal Reserve’s System Open Market Account (SOMA) holdings:

As can be seen, SOMA holdings stood at approximately $500 billion in 2008, peaked at about $8.5 trillion during the pandemic response, and remained around $6.5 trillion by late October 2026. While this balance sheet expansion anchored ultra-low interest rates throughout most of the 2010s, post-2022 macroeconomic realities – characterized by sticky inflation, elevated public debt issuance, and impediments to international flows due to trade barriers (real or perceived) – have structurally shifted the yield curve upward, as reflected in the behavior of the Federal Funds Effective Rate:

Conclusions:

Central banks no longer possess the unconstrained flexibility to slash rates or deploy massive asset purchases at the first sign of economic friction without risking the amplification of inflationary spirals. If El-Erian is correct, and I think he is, structurally higher baseline rates and elevated term premia are here to stay (until evidence emerges to the contrary). Policymakers everywhere, especially in Europe, must adapt by prioritizing productivity, fiscal discipline, and encouraging private investment rather than relying on central bank accommodation to mask underlying economic vulnerabilities. But, as I have stated in other posts, I remain highly skeptical that government officials will take the proper steps to mitigate this systemic risk.