Market Signal — May 12, 2026

May 12, 2026

Market Signal Macro · Rates · Narrative

MARKET SIGNAL May 12, 2026

Two voices today, but they're both pointing at the same fault line — and the convergence is worth taking seriously. Sosnick and Boockvar are independently arriving at the same question: is this 1999, and if so, what inning are we in? The answer neither will give you is a timing call. What they will give you is a checklist of structural echoes that should unsettle anyone who's been riding the semi/AI momentum trade without thinking hard about the composition of what they own.

THE 1999 QUESTION: GENERATIONAL SPLIT, STRUCTURAL ECHOES

Sosnick frames the current moment as a genuine divide between two investor cohorts. Those who lived through the dot-com bubble are asking whether this rhymes. Those who came of age post-GFC are asking why anyone would think it does. Neither camp is obviously wrong — but Sosnick's framing implies the latter group is operating without the relevant priors.

Boockvar goes further, citing a data point from technician Helene Meisler — a "multi-decade veteran" — who wrote that the gapping behavior in semiconductors is 1999. Not a forecast, but a phenomenological match. More damning is the Goepfert data mine Boockvar flags: this will be only the 4th time in history the S&P 500 hit a record high while fewer than 5% of its members were also making 52-week highs. The prior instances: July 1929, January 1973, December 1999. The message isn't that the market tops here — it's that breadth this narrow at new highs has historically preceded severe mean-reversion. That's a narrow club to join.

EARNINGS QUALITY: THE "OTHER INCOME" PROBLEM

The 27% year-over-year headline EPS growth among hyperscalers in Q1 looks like a blowout — until you read the footnotes. Boockvar surfaces an FT analysis showing a material portion of that growth sits in the "other income" line: mark-to-market gains on private investment portfolios, primarily positions in OpenAI and Anthropic. These are not operating earnings. They are paper gains on illiquid stakes in companies whose own valuations depend on the continued enthusiasm of the very ecosystem reporting the gains.

This is circular in a way that deserves more attention than it's getting. Hyperscalers invest in AI companies, AI companies generate hype that inflates the value of those stakes, those inflated stakes flow through to reported EPS, and the market re-rates the hyperscalers higher on the "earnings beat." The operating business may be strong — but investors pricing off headline EPS are partially pricing a feedback loop, not a cash flow stream.

COMMODITIES: THE SIGNAL THE MARKET ISN'T WATCHING

While everyone stares at Nvidia and the semis, Boockvar flags the move in industrial metals as the cleaner, less-crowded expression of the same thesis. Copper is at a record high. The Bloomberg Industrial Metals Index closed last Friday at its highest level since June 2022. The argument is straightforward: none of the AI infrastructure buildout happens without copper, aluminum, nickel, zinc, silver, and platinum. The physical constraint is real and the financial market is starting to price it.

This is the less-sexy, more-durable version of the AI trade — and one without the valuation circularity problem embedded in hyperscaler earnings. Boockvar has been making this case consistently, and the commodity price action is validating it.

GEOPOLITICAL BACKDROP

Boockvar notes in passing that the Strait remains under "effective closure" — and that equity markets are holding up anyway. This resilience in the face of an ongoing supply disruption is either a sign of genuine macro strength or a sign that the market has stopped pricing tail risks it hasn't yet been punished for ignoring. That's worth watching.

What the overall picture suggests: The high-conviction read here is that the rally's internal architecture is deteriorating even as headline indices hold up — narrow breadth, earnings quality questions, and a valuation regime built partly on reflexive private-market marks. That doesn't tell you when it reverses. But it does suggest that the bulls need more things to go right from here, not fewer. The commodities call from Boockvar is the most actionable near-term signal: if you want AI infrastructure exposure without the valuation circularity, that's where the asymmetry may sit. The macro wildcard remains the Strait — a market that's been shrugging off geopolitical disruption is a market that has borrowed from its own risk buffer.

TL;DR - Breadth is historically alarming: S&P at record highs with <5% of members at 52-week highs has only occurred three times before — 1929, 1973, 1999. Both writers are flagging this, not as a top call, but as a structural caution. - Hyperscaler earnings quality is weaker than the headline suggests: Q1's 27% EPS growth is partly paper gains on OpenAI/Anthropic stakes — a feedback loop, not operating leverage. - Copper and industrial metals are the cleaner AI infrastructure trade: Record high copper and a 4-year high in the Bloomberg Industrial Metals Index offer AI upside without the earnings-quality noise embedded in semis and hyperscalers. - The generational divide is itself a signal: When experienced investors hear echoes of 1999 and newer investors don't recognize the pattern, that asymmetry of historical priors is part of what makes it 1999.

Compiled from 2 sources · 2 items

Steve Sosnick (1)
Peter Boockvar (1)