Monthly Digest: Waiting for God(ot)

Ancient Greek playwrights, when in need of a conclusion, often used the ‘deus ex machina’: a god winched down to resolve the plot. 

With broader indices struggling for direction, do we need a higher force? Usually, that force takes the form of government or a central bank. The problem is they usually intervene when a crisis is at hand, and it would be churlish to invoke a crisis to break the current logjam. 

The same themes keep recurring: the Iran War; AI optimism and pessimism; central bank policy; market-crisis fears; and UK fiscal politics – and remain far from resolved.

The great gulf

Ships and commodities remain trapped in the Persian Gulf with the ongoing Strait of Hormuz closure. 

Any commentary comes with the caveat that events may have moved on by the time of reading. Still, the sticking points are clear: the US, Gulf states, and most of the world want free passage through the Strait of Hormuz, while Iran sees the threat of closure as both a means of leverage in negotiations and a possible source of future income. The US is determined to neutralise Iran’s nuclear threat; Iran is loath to comply. 

The global economy has cruised relatively unscathed since the war began in February,  risking complacency. Oil consumers drew on strategic reserves, but these have dwindled. Energy source pivots and a modicum of privation are the easy wins that one might describe as trimming fat, but we’re getting closer to the muscle now. 

So why do markets remain relatively calm? The consensus is that neither side has the appetite for prolonged hostilities as the financial and political costs rise, and the US is rumoured to be running low on key ordnance. 

BCA Research’s credible “oil price collar” theory sees Brent crude trapped between $70 and $100 a barrel. At the lower end, tensions tend to rise as the political and economic costs fall; as prices approach the upper end, escalation gives way to de-escalation as the costs rise again. That concept, albeit a fragile one, seems to be holding. 

The crude collar

Mismatched expect-AI-tions

Amara’s Law says we tend to overestimate a new technology's short-term effect and underestimate its long-term impact. Generative artificial intelligence (AI) is showing this today, with extreme consequences for certain share prices. 

The AI push is largely driven by frontier large language model (LLM) developers (think ChatGPT, Claude, or Google Gemini) and the ‘hyperscalers’ building the data centres needed to train and run them. LLM funding has come largely from private equity backers. 

Hyperscaler spending has so far been funded largely by operating cash flow, which is massive for companies such as Alphabet, Microsoft, Meta, and Amazon (with Oracle, of the Big 5, being the notable debt-funded exception). The numbers are mind-boggling: annual capex is forecast to rise from $224bn in 2024 to around $700bn this year and $900bn in 2027.

Problem number one: this level of capex is now outrunning cash flow generation. This leads to massive equity issuance ($85bn for Alphabet, for example). It also prompts huge debt issuance (more than $200bn so far in 2026 in public markets and potentially the same again in private markets and through the backing of special-purpose off-balance sheet vehicles). 

Problem number two: will all this spending generate a return? It must beat the cost of capital and, ideally, match the returns from existing operations. That is a high bar for hyperscalers, whose exceptional performance has been built on scalable, ‘capital light’ business models where the majority of every dollar of extra revenue turns into profit. 

Problem number three: who’s paying for the service? Engagement with LLMs has risen exponentially, but most users are not paying for them. Revenue is mainly coming from ‘enterprise’ customers, or companies. Anthropic’s record revenue growth has yet to translate into a sustainable profit. 

Billing has shifted from an ‘all you can eat’ basis to a ‘per token’ or ‘per task’ model earlier this year. As invoices rose, usage was curtailed or switched to cheaper models for less onerous tasks, many of which are based in China, which is another concern.

This all makes for a very difficult market to negotiate on a short-term basis. Especially as ownership of the shares involved is meaningfully funded by borrowed money and by leveraged exchange-traded funds and options, all of which exacerbate short-term volatility. The forced liquidation of positions by a hedge fund, ironically called Situational Awareness, at the end of July now seems to have been a key contributor to recent gyrations. 

Our central view today is that both the frontier model builders and the hyperscalers are committed to their growth paths, and the demand for computing power will continue to rise as companies and individuals learn to integrate AI-based solutions productively into their operations and lives, especially if it becomes cheaper (as in Jevons’ Paradox). But it will not happen linearly. 

For longer-term investors, swings from overoptimism to unmerited pessimism create opportunities for active managers. 

Central bankers sit on their hands

The latest central bank meetings produced no changes to monetary policy. The new US Federal Reserve chair, Kevin Warsh, has made his committee’s intentions less clear and allowed markets to set rates. Markets have duly obliged, leading to a sell-off in long-dated Treasuries, pushing yields back to levels last seen in 2007. 

In the UK, despite a 6-3 vote to leave the base rate at 3.75%, one more voter joined the ‘hawks’ in July. Yet the tone of accompanying statements was interpreted as relatively dovish. The European Central Bank’s message fell somewhere in between. 

With underlying economies weak in both the UK and Europe and President Trump marking Warsh’s homework, policymakers remain reluctant to increase interest rates. With inflation steadfastly refusing to return to the 2% target, futures markets are pricing in increases later this year. Policymakers are largely ‘looking through’ the war-related inflationary pressures given that: 1) they are the result of a supply shock that is not readily countered by monetary policy; and 2) there are limited ‘second round’ effects, such as higher wage demands. 

Conclusion: Diversification pays off 

Current sentiment swings, amplified by leveraged trading vehicles and strategies, make it easy to look like a hero one day and a failure the next. A diversified portfolio provides a more stable ride and keeps us focused on longer-term investment and planning goals. We still take bigger directional bets where conviction is high, and continue to filter markets for opportunities. AI-related winners are likely to proliferate, including in areas not yet apparent, as adoption of this technology remains in its infancy.

Many media stories about extraordinary gains (or losses) probably reflect leverage and/or poor risk management. We don't borrow money to juice returns, and we remain attentive to portfolio volatility relative to agreed benchmarks as stewards of clients’ hard-earned wealth.