The consensus for 2026 is comfortable and crowded. Wall Street strategists cluster around 7,000 to 7,300 on the S&P 500, and the BofA fund manager survey shows the same base case: another year of gains, powered by the same tech complex that carried 2024 and 2025. The interesting question is not whether the consensus is right. It is what happens if the capital cycle underwriting it stalls — because the index and the AI build have become the same trade.
Start with the risk hierarchy the buy-side has actually written down. According to the BofA fund manager survey, 45 percent name higher inflation and interest rates as the top risk for 2026, and 26 percent name an AI capex slowdown. That second number is the one worth staring at. A year ago, a slowdown in hyperscaler spending was not a mainstream tail. Now more than a quarter of managers treat it as the dominant scenario risk — ahead of unemployment above 5 percent (17 percent), a US-China trade rupture (9 percent), and the midterms (2 percent). The market has internalised that the index and the capex cycle are the same object.
From capital-light to asset-heavy
The structural point, made cleanly by Maverick Equity Research, is that TMT (technology, media and telecom) is no longer the capital-light margin machine of the 2010s. It is becoming asset-intensive — data centres, custom silicon, power contracts, cooling — and that shift will compress margins in the near term regardless of how the demand side resolves. This matters for index math. An S&P 500 whose earnings weight is concentrated in a handful of hyperscalers is now an index whose margin trajectory depends on depreciation schedules and utilisation rates, not just gross profit on software seats. The multiple the market pays for that cash flow should, in principle, be lower. It is not.
The sharpest stress test of that future is OpenAI, and HSBC's numbers are the ones to hold in view. The bank estimates OpenAI needs at least $207 billion in new equity funding through 2030, against cumulative operating losses of roughly $500 billion between 2025 and 2030, with a $76.5 billion loss in the terminal year even as revenue reaches around $220 billion. Free cash flow, on HSBC's arithmetic, turns positive only around 2030. That is a private company absorbing more equity over five years than the market capitalisation of most S&P 100 members — and it is the demand anchor for a meaningful share of the capex the index is being rerated on.
The index and the AI build have become the same trade. If one clears, both clear. If one stalls, both stall.
The dossier is one-sided: every named forecaster in the cluster sits on the bullish side of 2026 at medium conviction, and HSBC's OpenAI numbers are presented as base-case rather than stress. That unanimity is itself the signal. Readers should treat the operationalisable claims accordingly — Wall Street strategists carrying the S&P above 7,300 by year-end 2026, HSBC's $207 billion OpenAI equity raise clearing by 2030, and OpenAI's $220 billion 2030 revenue print — as a linked chain, not independent bets. The 26 percent of managers flagging a capex slowdown are pricing the scenario in which the chain breaks at its weakest link. That is the trade worth watching, not the consensus target.
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.