US Treasury yields surge: How much longer can the US economy and stock market hold out? AI may be the most vulnerable link

Global bond yields have climbed to multi-decade highs, yet both the economy and markets have so far exhibited extraordinary resilience. As pressure begins to emerge in areas such as the European sovereign bond market, whether this resilience can be sustained has become the market’s most urgent question.

The latest research report by Goldman Sachs aggregates analyses from the International Monetary Fund, Moody’s Analytics, and Goldman Sachs economists. IMF advisor Tobias Adrian takes a cautiously optimistic stance, believing that as long as the rise in yields does not accelerate disorderly, the global economy is likely to maintain its resilience; Goldman Sachs economists expect the drag of high interest rates on U.S. economic growth to be relatively mild, around 0.2 percentage points by 2027, but if yields remain at current levels, the drag effect will double to more than 0.5 percentage points. Moody’s Analytics chief economist Mark Zandi, however, holds a distinctly more pessimistic view—stating bluntly that the U.S. economy is "fragile," not "resilient," and estimating the probability of the economy completely derailing at roughly one-third.

At the market level, Goldman Sachs strategists overall believe that the banking system and equity markets do not yet pose systemic risks but warn that if yields rise further or rate volatility remains high, risk assets will face increasing pressure. The overarching consensus among various perspectives points to one key issue: can the highly interconnected, leveraged AI investment boom withstand the continued test of a high-interest-rate environment? Zandi issued the most direct warning: if rates remain at or above current levels for another one or two quarters, "AI could be the first link in the economy to break."

The most prominent characteristic of this wave of global bond yield increases is its breadth. William Marshall, head of U.S. Rate Strategy at Goldman Sachs, notes that yields have risen across all major economies, with no obvious single market leader or trigger, and both the front and long ends of the yield curve have hit new highs in tandem.

In terms of driving factors, Marshall attributes this yield ascent mainly to fundamentals: robust economic growth, inflation risks, hawkish pivot by central banks, an ongoing investment cycle, and long-term fiscal risks. Term premium—the extra compensation investors require for holding longer-duration assets—has soared to its highest level in over a decade. Since late September, the sharp escalation in yields has been further amplified by technical factors related to hedging risks, pushing yields beyond what fundamentals alone would suggest.

Looking ahead, Marshall expects yields to retreat as market expectations of easing inflation solidify but emphasizes that structurally higher yields will be "sticky." Energy price volatility and fiscal pressures are the chief upside risks. Meanwhile, the ongoing expansion of AI capex and related bond issuance are holding back demand for duration assets, skewing long-term yields toward the upside.

Goldman Sachs U.S. economist Pierfrancesco Mei classifies the current U.S. economy as resilient, and expects that under the baseline scenario—one more Fed rate hike in December, followed by three cuts starting in late 2027, with 10-year U.S. Treasury yields falling to 4.4% by year-end—the drag from high rates on 2027 GDP growth will be just about 0.2 percentage points.

However, if yields stay at current levels, the drag effect will exceed 0.5 percentage points. By transmission channel, a 30-year mortgage rate above 7% will further constrain housing investment; consumer spending will be reduced by about 0.2 percentage points due to weaker demand for durable goods; and higher interest payments as non-AI companies refinance maturing debt are expected to pressure capital expenditure growth by around 0.3 percentage points.

Goldman Sachs global economists Joseph Briggs and Johan Allen developed a quantitative framework: for every 100 basis-point rise in rates, residential investment in major developed economies drops about 6%, consumer spending declines on average by 0.2%, and corporate capital expenditure falls by about 1.8%, together dragging developed market GDP by about 0.6 percentage points. On the fiscal side, a large structural deficit compounded by high rates is projected to push U.S. debt-to-GDP to 124% by 2035 in the base case—or 132% if current rates persist.

Among the generally upbeat market commentary, Mark Zandi is the most cautious voice. He finds describing the U.S. economy as "resilient" to be "fascinating, even frustrating," insisting the correct term is "fragile."

Zandi highlights the labor market as the clearest sign of weakness. Over the past year, average monthly job gains were about 40,000, but nearly zero excluding the healthcare sector; while the unemployment rate remains low, this is partly due to falling labor participation and, after adjustment, the real jobless rate is in the mid-4% range—above full employment. Real wage growth for middle- and low-income groups is in fact negative, and household savings are being steadily depleted.

Looking at aggregate data, Zandi is not reassured by the roughly 2% real GDP growth rate. He estimates a third of this—around 0.6 to 0.7 percentage points—comes from AI-related activity (including the net contribution of AI capex and wealth effects from AI-related stock performance). This means the non-AI segment of the economy is growing only about 1% to 1.5%—"a pretty bad economic state."

Zandi also describes U.S. fiscal conditions as "the worst I’ve ever seen": a total deficit above 6% of GDP, a structural deficit over 3%, debt around 100% of GDP and projected to rise to 125% within ten years, with federal interest outlays already exceeding $1 trillion and rising rapidly. His risk assessment: about a 25% chance of a severe bond market selloff, and roughly a one-in-three chance of the economy completely going off the rails.

Although Mark Zandi remains highly alert to potential fracture risks in the financial system, Goldman Sachs analysts are relatively moderate in their assessment of banks and the credit market.

Both former IMF Director of Monetary and Capital Markets Tobias Adrian and Richard Ramsden, head of Goldman Sachs Financials Research, believe there is no need for undue concern about banks. Ramsden notes that given U.S. banks’ asset-sensitive balance sheets, higher rates should help boost net interest income; unless there is a broader economic downturn, credit quality is unlikely to deteriorate significantly; and after accounting for rate-driven marks on securities portfolios, most banks should retain excess capital.

In the credit market, Goldman Sachs chief credit strategist Amanda Lynam believes that, amid current robust growth, the impact of high rates on corporate credit is generally manageable and insufficient to trigger a significant rise in default rates. However, she also cautions that if rate volatility returns to the highs of 2022–2023, confidence among yield-seeking institutional investors could be hit, putting substantial pressure on demand for investment-grade credit.

Goldman Sachs chief U.S. equity strategist Ben Snider contends that current yield levels should not prevent U.S. equities from rising further. Although the S&P 500 forward P/E ratio has fallen from 23x to 19x this year, strong earnings growth (over 30% annualized in the past 12 months) has more than offset the impact of valuation compression. Furthermore, the equity-bond relative valuation—measured by the S&P 500 earnings yield minus the 10-year real Treasury yield—is about 230 basis points, and has remained relatively stable over the past two years, indicating that valuations have largely recalibrated to higher yields.

However, Snider highlights two key risk areas: First, if the 10-year Treasury yield rises by more than about 50 basis points in one month (two standard deviations), it would trigger the typical stress pattern seen in historical U.S. equities; second, rate-sensitive groups such as small-caps, real estate, and homebuilders have shown clear divergence, and market breadth is now at its lowest since the internet bubble. Goldman multi-asset strategists Christian Mueller-Glissmann and Andrea Ferrario add that higher bond yields are reinforcing the strategic value of bonds in multi-asset portfolios, with the optimal bond allocation approaching the long-term average of 35% to 55%, but, for now, strong earnings growth still favors equities over bonds.

In current economic and market discussions, AI is both the core engine sustaining growth and the most scrutinized potential vulnerability in a high-rate environment.

From an optimistic perspective, Goldman Sachs analysis sees limited capex risk for AI. The capital expenditure of Hyperscalers still shows strong resilience to high rates and rising costs, accounting for more than two-thirds of U.S. AI investment; MSFT, META, AMZN, GOOGL, NVDA, and other major AI issuers are predominantly rated AAA to AA, with an interest coverage ratio of 27.7 times and net leverage of just 0.5 times, providing ample debt capacity; Lynam also notes that the bar for AI-related bond issuance disruption is extremely high.

However, Tobias Adrian points to three layers of risk embedded in the AI complex: the networked and circular investments among AI firms, meaning shocks to one company’s profits could quickly propagate; the leverage-upon-leverage structures resulting from leveraged ETFs and derivatives connecting AI firms and investors; and, fundamentally, uncertainty over whether strong AI revenue growth can be sustained.

Mark Zandi’s warning is the most direct: "If rates remain at current levels or higher for another quarter or two, AI could be the first sector of the economy to break." He argues that sustained high rates will suppress AI-related stock prices, and the reversal of the wealth effect will further hit consumption, while higher financing costs will make it even harder to raise the capital needed to maintain current investment momentum in AI. At present, AI largely relies on its own momentum and market optimism, but the duration of this state remains highly uncertain.