Why the Macro Data Says AI Is Not a Bubble
Weekly Market Update
Takeaways
Contrary to the corporate profit margin squeeze seen during prior investment bubbles, aggregate margins today are holding near all-time highs. The total corporate interest burden is only 1.3% of output1, roughly a third of levels in the Dot-com and housing bubbles2, and a 60 year low. U.S. corporations enter this investment cycle from a position of profitability and balance sheet strength.
Companies fund today’s capital spending from internal cash flow rather than debt and equity issuance. While production of high-tech goods is strong at 11% year over year3, this compares to roughly 50% during the Dot-com bubble. Like corporate fundamentals, the macroeconomic data does not suggest AI is a bubble, and this points to potential opportunities for long-term investors.
Three Federal Open Market Committee (FOMC) members dissented in favor of a rate hike. With some policymakers appearing to lose patience with above-target inflation, the balance of risks favors tighter policy. That view keeps us cautious on extending duration in fixed income.
This Week in Charts
Looking Closer
The production of high-tech goods is strong, adding significantly to U.S. growth. However, high-tech production growth of 11% year-over-year in June 2026 is far more moderate than the roughly 50% growth reported during the Dot-com bubble.
Market and Data Recap
Corporate margins and balance sheets separate today from prior bubbles
The question around a bubble is becoming more prevalent. We usually respond by conducting company by company and sector by sector analyses. We also compared today’s macro data against the two most recent investment bubbles, the late 1990sDotcom boom and the mid 2000s housing boom.
Start with margins. Profits as a share of output give us a macro level read on corporate profitability. In both prior bubble periods, that measure fell as margins compressed, and companies responded normally, by cutting investment and cutting labor. Today, margins sit at high levels and show no sign of rolling over, which explains the impressive earnings companies keep reporting.
Aggregate corporate balance sheets tell a similar story. Investors focus on how much debt companies issue to fund investment, so we looked at the interest burden measured by corporate interest payments as share of output. During the prior two bubbles that burden averaged roughly 3.1% and climbed. Today it sits near 1.3%, a 60-year low, and it continues to fall.
A low interest burden exists despite higher interest rates. Companies managed their liabilities well and termed out debt while rates were low, so total debt carries a light interest cost today. Balance sheet strength, not leverage, funds the current cycle.
Bottom line: Prior bubbles arrived with compressed margins and a rising debt service burden, and today we see neither. Corporate America enters this investment cycle from a position of profitability and balance sheet strength.
Capital spending shows discipline, and cash flow pays for it
Investment in the high-tech complex looks strong, and we want to keep that strength in perspective. The Federal Reserve’s industrial production report tracks a high-tech segment covering computers, semiconductors, and communications equipment. That segment has grown about 11% year over year. During the Dot-com bubble, it grew near 50% year over year, and demand never arrived to absorb the output.
Capital expenditure (capex) lead times point in the same direction. Lead times measure how far in advance companies commit to capital investment. During the Dot-com bubble, those lead times shortened as conviction faded. Today they run historically long, and they show no sign of shortening (Figure 2).
How companies pay for that spending also matters. In the Dot-com bubble, companies financed capex through debt and equity issuance, and the financing gap between capex and internal cash flow widened. Today, although some large cap Technology names report negative free cash flow, the corporate sector runs a financing surplus in aggregate, meaning cash flows still exceed capital spending even at these investment levels.
Investor positioning and valuations round out the picture. Households hold high levels of liquid assets relative to their total assets today, while both prior bubbles featured the opposite as households concentrated into Technology stocks and then into housing (Figure 3).
On valuation, S&P 500 Information Technology forward multiples peaked near 62 during the Dot-com bubble. That multiple has averaged around 38 during the AI capex cycle and moved lower in recent months, and consensus forecasts point to further contraction by 2027 due to strong expected earnings growth.
Bottom line: Today’s capital spending runs strong, self-funded overall, and forward committed, and it looks nothing like the debt financed overbuild of 1999. We disagree with the parallels investors draw between the AI buildout and prior investment bubbles. Macroeconomic trends related to the AI buildout support the message from company fundamentals that the current environment is supportive of equity risk.
The Fed leans hawkish, and we still expect a hike next
The Fed left policy unchanged and delivered a short statement with no forward guidance, the new standard under Chairman Warsh. However, while June brought no dissents, the July meeting brought three, all favoring a rate hike.
Timing makes that shift interesting because June inflation data, released ahead of the July FOMC meeting, came in fairly benign. The dissenters focused on the duration of above target inflation rather than the latest report. Inflation has now overshot the Fed's target for over five years, and several FOMC members appear to be losing patience with that record. Their written dissents argued the case clearly and questioned if there is sufficient policy restraint to return inflation to target.
Taylor Rule4 recommendations also suggest the current policy setting is too loose (see Figure 4). The Cleveland Fed's average across seven rules recommends a 4.4% policy rate versus the current rate of 3.6%. The median recommendation of 30 rules in the Atlanta Fed's Taylor Rule heatmap is 5.6%.
We are not advocating for mechanically following a policy rule, but there is evidence that the current Fed funds rate is too low. We continue to expect the next move from the Fed to be a rate hike rather than a rate cut. The onus is now on the economic data to prevent more FOMC voters joining July’s three dissenters and pushing the majority to increase rates. Ahead of the next FOMC meeting on September 16, the CPI reports on August 12 and September 11 will be an important focus.
Market reactions to the latest FOMC meeting have raised questions about the Fed's inflation-fighting credibility, but we think these are overstated. This is not 2022, when the Fed fell behind the curve and moved aggressively to catch up. Five-year inflation breakevens from Treasury Inflation-Protected Securities (TIPS) are up only two basis points since the Fed meeting, to 2.2%. Since Kevin Warsh took over as Fed Chairman, breakevens have fallen roughly 30 basis points while real yields have risen about 40 basis points. Markets appear to anticipate tighter policy rather than a loss of inflation control.
Bottom line: Market based measures of inflation expectations remain well anchored, but amid an extended inflation target overshoot and with monetary policy too accommodative, the balance of risks favors more tightening. That view keeps us cautious on extending duration in fixed income portfolios.
1 Haver Analytics as of July 29, 2026.
2 The production of high-tech goods is strong, adding significantly to U.S. growth. However, high-tech production growth of 11% year-over-year in June 2026 is far more moderate than the roughly 50% growth reported during the Dot-com bubble.
3 Haver Analytics and Bloomberg as of July 29, 2026.
4 The Taylor Rule is a monetary policy guideline that prescribes the federal funds rate based on the deviation of actual inflation from the target inflation rate and the output gap, which measures the difference between actual potential GDP.
See our weekly CIO Strategy Bulletin for more details