Big Tech and semiconductors are now trading more than two standard deviations above their historical relative bands against the S&P 500, propelled by an AI investment cycle that has seen Amazon, Microsoft, Alphabet and Meta commit close to $300bn of capital expenditure in the last twelve months. The index looks expensive at the top and reasonable underneath. The interesting question is not whether the rally is justified by current earnings — it largely is — but whether the capex behind it will still be earning a return in 2027.
Start with what is actually happening to profitability. According to Goldman Sachs, S&P 500 return on equity has risen by 150 basis points over the past four quarters, and the bank attributes the move squarely to mega-cap technology stocks lifting index margins. That is the bull case in a single data point: the concentration premium reflects real cash generation from cloud, infrastructure and AI workloads, not the revenue-less narratives of 1999. Maverick Equity Research makes this distinction explicit, arguing the current tech complex is unlike the dot-com bubble because earnings and cash flow are doing the work.
The bear case sits inside the same numbers. Capex of that magnitude does not stay off the return-on-equity line forever. Today it is depreciating slowly and being matched by accelerating cloud revenue; in eighteen months, the depreciation schedules harden, the equity base grows, and the question becomes whether AI workloads are monetising at a rate that sustains current margins. Goldman Sachs flags exactly this risk, asking whether AI capex will become a net headwind to mega-cap tech ROE over the next 24 months and taking the YES side. That is the cleanest operational claim in the dossier, and we read it as the consensus view rather than a contrarian one — the dossier shows no named forecaster taking the opposite side.
Metrics to watch, not yet in hand
Maverick is proposing two proprietary gauges — the Maverick Computing Efficiency Index (MCEI) and the Maverick AI Monetisation metric (MAIM) — to track whether hyperscaler dollars convert into returns. They are committed to publishing both by the end of 2026. Until those land, investors are left triangulating from segment disclosures and inference-cost commentary, neither of which resolves cleanly. The honest read is that the market is pricing a successful conversion of capex into durable cash flow, and the tools to falsify that price do not yet exist in public form.
The index is expensive at the top and reasonable underneath — that is a portfolio instruction, not a market call.
The portfolio implication is where the dossier is most actionable. Valuation stress is concentrated in cap-weighted exposures: the Nasdaq-100 and the top-heavy S&P. Equal-weight versions of the same index, US mid-caps, the DAX, and what Maverick describes as undercovered Swiss equities all trade at materially less demanding multiples. Mike Pyle, drawing on BlackRock's Portfolio Management Group and its roughly $5 trillion in client assets, frames the broader regime as one defined by supply-side shocks and geopolitical fragmentation — conditions under which a 60/40 portfolio leans on systematic strategies, macro and liquid alternatives to do work the bond leg used to do alone. That argument and the equal-weight argument point in the same direction: diversify away from the concentration that has, until now, been the trade.
Briefings are synthesised by the Ledger Desk from multiple sources cited in the sidebar. They are distinct from Articles, which are written by named contributors and carry a tracked Calibration Index. The Desk does not currently carry a Brier score; this is a deliberate choice for the v0.1 editorial layer and will be revisited.