According to Bloomberg Markets, copper and aluminium advanced on Monday as traders priced out near-term rate-hike risk, a move that also steadied gold after its prior week's gain as rate-hike worries receded. On the surface, this looks like a straightforward relief rally. Beneath it, the U.S. Treasury yield curve is telling a more complicated story — one that serious debt-market historians have seen before.
The Term Premium Signal That Risk Assets Are Ignoring
The term premium — the extra yield investors demand for holding long-dated bonds rather than rolling short-term paper — has been quietly rebuilding in U.S. Treasuries throughout 2025 and into 2026. The Federal Reserve Bank of New York's Adrian-Crump-Moench (ACM) model, one of the most widely cited decompositions of the nominal yield, had kept the term premium deeply negative for most of the post-2008 period, a structural artifact of quantitative easing and global savings gluts. That era appears to be ending.
When the 10-year term premium last sustained positive territory at elevated levels, it was 2007 — the year before the credit crisis metastasised. Before that, elevated term premiums in the late 1990s coexisted with a paradox: short-term rates were rising, yet long-duration capital was flooding into speculative infrastructure projects at yields that implied almost no sovereign risk premium was being demanded. That misalignment — cheap long money chasing yield-insensitive projects — was precisely the fuel behind the Telecom Bubble and Fibre Optic Mania (1998-2002), in which the industry collectively spent vast sums building fibre networks, leaving the large majority of the capacity dark when demand projections proved wildly optimistic.
The structural parallel is instructive. In the late 1990s, a flat-to-inverted short end of the curve coexisted with compressed long-end spreads, creating the illusion that duration risk had been abolished. Today's configuration is different in form but similar in implication: the short end has eased as rate-hike fears fade, but the long end is not following proportionally. The curve is steepening — and not in the benign, recovery-driven way that textbooks describe.
Bear Steepening: The Variant That Precedes Dislocations
There are two ways a yield curve steepens. In a bull steepener, short rates fall faster than long rates, typically signalling that a central bank is cutting into a soft landing. In a bear steepener, long rates rise faster than short rates — or hold their ground while the short end falls — driven by fiscal concerns, inflation risk premia, or deteriorating confidence in sovereign debt sustainability. The current dynamic has bear-steepener characteristics.
This matters enormously for equity valuations. The standard discounted cash flow framework uses long-term risk-free rates as the denominator. When the term premium rises — even as near-term policy rates stabilise — the discount rate for long-duration assets increases. Growth stocks, infrastructure plays, and AI-adjacent capital expenditure projects are all implicitly long-duration bets. A sustained bear steepening regime reprices them downward, regardless of what the Fed does at its next meeting.
For equity valuation and market risk context, the current setup warrants particular attention: the price-to-earnings multiples on U.S. large-cap technology stocks remain historically elevated, priced for a long period of low discount rates that the term premium is now quietly undermining.
[Analyst quote removed — original quote could not be verified against provided source material.]
The Telecom Era Analogy: Misallocation Funded by Mispriced Duration
The telecom mania of 1998-2002 was not simply a story of irrational exuberance. It was a story of capital misallocation enabled by a specific interest rate environment. Investment-grade telecom bonds traded at spreads that assumed the sector's revenue projections were credible. Long-dated corporate paper was issued at yields that, in hindsight, failed to compensate investors for either duration or credit risk. The result was that companies like WorldCom — whose massive accounting fraud was eventually exposed in the wreckage — could raise capital almost without friction, funding network builds that the market could never absorb.
The analogy to today's AI infrastructure buildout is imperfect but non-trivial. Hyperscaler capital expenditure commitments for data centres and GPU clusters are being made on multi-year timelines, funded partly by debt issued into a market where the term premium is still, by historical standards, relatively contained. If the bear steepener dynamic accelerates — if fiscal deficits in the United States, Japan, and across Europe continue to expand the supply of sovereign paper — those infrastructure funding costs will rise. Projects underwritten at 2024 yield assumptions may look very different at 2027 refinancing dates.
The comparison between the dot-com bubble and AI infrastructure spending in 2024 has been made by a growing number of analysts, and the yield-curve dimension of that comparison deserves more attention than it typically receives. See also our current bubble risk indicators for live tracking of term premium and credit spread data.
What Fading Rate-Hike Fears Actually Signal
The immediate market reaction — commodities up, gold steady, risk appetite recovering — reflects relief that the next central bank move is more likely to be a cut than a hike. That is a reasonable near-term read. But the historical record suggests that the transition from a hiking cycle to a cutting cycle is not a safe harbour. It is often the period in which accumulated misallocations become visible.
The current fading of rate-hike expectations is being read as a green light. The term premium is flashing amber. Those two signals have coexisted before, and the resolution has not always favoured the equity optimists. Investors and policymakers watching the full spectrum of historical bubble episodes will recognise the pattern: the moment the short end relaxes is often the moment the long end begins to demand its due.
