The Weekender is my bi-weekly take on macro shifts and emerging themes. It’s not investment advice — or even our firm’s official view. I aim simply to inform, challenge, and maybe entertain. If you’d like this in your inbox every other Saturday morning via Northern Trust, subscribe to The Weekender.

 

Healthy scepticism

 

In a recent interview, Howard Marks, co-chairman of Oak Tree Capital Management, referenced back to a question Alan Greenspan posed in a 1996 speech as chair of the Federal Reserve: “How do we know when irrational exuberance has unduly escalated asset values…?” In today’s market, Marks said the most important question isn't whether investors have exuberance — we do — it’s whether this exuberance is irrational.

 

Despite markets in the U.S., Europe, Japan and China technology (e.g. Star50 and ChiNext indexes) making new highs this week, I'm still struck by the degree of scepticism we encounter. Typically, new highs are greeted with party hats and balloons, and yet, outside of a few younger colleagues, some U.S./tech-based clients and one particularly brilliant group from Edinburgh, the vast majority remain somewhat solemn. Sober, even. And sobriety is seldom seen at market tops. Or at parties.

 

Ignorance is bliss

 

Historian Stephen Kotkin, author of Armageddon Averted, likes to say that if you don't know history, you have no precedent. The inverse is just as true. During the dot-com mania, everything felt unprecedented largely because no one alive could recall the last great technology boom (transport, circa 1900). As many remember dotcom, we anchor to the bust, not the boom — due to loss aversion — and the media hardens that bias. The Economist noted that the most-read book on Wall Street last year was 1929. A good bet for this year is 1873. Both are crash narratives. Both colour how we see the future. And until scepticism gives way to euphoria, this market fails economist John Kenneth Galbraith's test — that a bubble is a mass escape from reality. If reality is what shows up in earnings, and expectations are what's priced, then the escape we're seeing is running in reverse: Multiples have de-rated, not inflated. Exuberant, yes. Irrational, not yet.

 

The past is prologue

 

If you're wondering why winning the AI race matters so much, look at who has historically invented and scaled the decisive technology of their era: The longbow at Agincourt, gunpowder at the fall of Constantinople, the chronometer and the British Navy during colonisation, railways in the Franco-Prussian War, tanks in the First World War, radar in the Second, and the Manhattan Project — perhaps the closest historical analogue to today's AI race. History suggests it’s seldom wise to bet against the full industrial, entrepreneurial, financial and political might of the U.S. when it aligns around a goal — like winning the AI race. Strategic primacy has never been purely about fundamentals. It's often about the future.

 

Counter-narratives

 

To stress-test my own bias, I've taken the five bear arguments I hear most often and tried to build the strongest counter-case to each. These aren't forecasts, just alternative lenses worth holding alongside the prevailing narrative.

 

1. “Oil Prices, and Inflation, Will Stay Higher For Longer"

 

I was struck by how many people became overnight oil and shipping experts these past few months. Armed with the MarineTraffic app, a chorus of TV punditry, industry executives with an obvious interest in higher prices and the International Energy Agency calling this the worst oil shock on record, the consensus was unanimous: oil higher for longer, inflation higher for longer. The market priced out Federal Reserve cuts entirely and briefly priced in two hikes. Instead, the largest oil supply disruption on record failed to produce an energy crisis. Oil, gas, electricity and coal prices never exceeded previous nominal peaks and traded well below long-run averages in real terms. Fertiliser prices had already fallen back below pre-war levels ahead of the peace deal. Jet fuel sat roughly 40% below its peak. The shortages everyone predicted largely failed to show up. Remember helium?

 

So why were so many so wrong? David Epstein provides us some clues.

 

Constraints Drive Creativity

 

David Epstein, author of Range (a brilliant book) and The Sports Gene, has a new book out called Inside the Box, How Constraints Make us Better. It's a useful lens here. His throughline is that constraints don't necessarily stifle creativity — they can create it. Necessity remains the mother of invention. Dr. Seuss's editor bet him he couldn't write a children's book using only fifty words; the result was Green Eggs and Ham, one of the bestselling children's books ever written. When Apollo 13's oxygen tank exploded, the fix was duct tape. And DeepSeek, built under constraint of restricted access to top-end US chips, in turn became America’s “Sputnik moment,” one that will shatter residue complacency, unlock further innovation and drive U.S. growth.

 

So what does this have to do with oil?

 

Thanks to earlier constraints — Covid, the war in Ukraine — markets had become resilient and adaptive, with contingency planning for exactly this kind of disruption already in place. A Strait of Hormuz blockade was one of those scenarios. As we’ve  discussed, supply diversified rapidly: Overland pipeline and road logistics took up the slack, shadow fleets (an estimated 125 million barrels moved since the start of Operation Freedom) filled gaps, U.S. exports boomed, Russia, Venezuela, Argentina and Guyana all added barrels, strategic reserves were released, and China effectively went on a buyers' strike — some of which may persist even after the conflict ends, accelerating the shift to electric vehicles. Iran's old strategic leverage — the ability to hold roughly 20% of global oil supply hostage — is now considerably diminished, which should erode any lingering war-risk premium. The same goes for OPEC supply discipline, given the UAE’s exit and the incentive for remaining members to chase back lost revenue and market share. Add the commodity truism that scarcity is reliably followed by surplus — you can't print molecules, but you can pump them — at a moment when the world is reframing decarbonisation as a strategic rather than moral or economic priority, and it's hard to build a durable case for oil staying higher for longer. Now, if goods inflation is largely oil-linked, and wage and shelter inflation look to have peaked, with AI productivity gains still ahead of us, that has real implications for the inflation outlook, for the European Central Bank’s next move, and for the new Federal Reserve chair. As things stand, one-year inflation breakevens sit below where they were before the war started — lower, in fact, than at any point since October 2024. Lower even than the Fed’s target, of 2%. Assuming that is still its target.

 

2. “Rising Yields Signal Contraction”

 

The conventional read is that higher yields can increase the discount factor and choke off growth. An alternative view: Yields may simply be reflecting improved growth expectations, particularly where the move is driven by real yields rather than inflation expectations. Two, five and ten-year inflation breakevens all remain below their prior peaks, which suggests this is primarily a real-yield and term-premium story, not an inflation/stagflation repricing. Moreover, some of the current growth drivers — AI capex, One Big Beautiful Bill Act tax rebates — are rate-insensitive, meaning higher rates do little to slow things down.  Historically, rising yields have more often preceded rising growth than choked it off, which makes intuitive sense: If the opportunity cost of holding bonds rises, the compensation on offer should rise with it.

 

Stability Stimulus, Banks and Bull Markets

 

There's growing chatter — from Fed Chair Kevin Warsh, Treasury Secretary Scott Bessent and others — that today resembles the Greenspan-era 1990s, when tech-led, non-inflationary growth let the Fed run the economy hot without consequence. The comparison is worth revisiting. Greenspan's real trick wasn't cutting rates; it was holding policy steady while growth accelerated. If that's the right framing, "neutral" needn't mean negative — stability itself becomes a form of stimulus. One underappreciated feature of that period is regulatory: Bank deregulation expanded system liquidity and provided a second channel of support alongside monetary policy. Something similar may be building now. Supplementary leverage ratio (SLR) reform, the removal of the Wells Fargo asset cap and a more risk-based approach to capital requirements all point toward greater bank balance sheet capacity — and, anecdotally, repo activity is surging, which is usually a sign banks are re-levering. Whether this amounts to "stealth easing" via regulation rather than explicit rate cuts is still an open question. But bank stocks are making record highs across Europe, Japan and the U.S., which is a reminder of an old Wall Street maxim: You can't have a bull market without banks. And we have bank stocks, it seems.

 

3. “’Tokenmaxing’ Has Peaked”

 

Charts on large language model (LLM) expenditure (see the SDLLMTK Index, if you have a Bloomberg terminal) have started to roll over. The bearish read is that users are becoming more selective, hitting diminishing returns and cutting back on token usage — which would eventually flow through to weaker AI capex. The alternative is that this is a normal optimisation cycle rather than structural saturation. Some seasonals are at play (think university holidays). Lower spend may simply reflect providers cutting prices to win share ahead of an IPO. It's also unclear whether these indices capture newer pockets of demand — neo-cloud players like xAI, sovereign AI buildouts, agentic systems, edge and embedded use cases. If history rhymes, we should expect prices to fall as efficiency improves, which in turn drives even greater adoption — the Jevons Paradox at work. I'd treat the signal value of these spend indices with real scepticism. A fuller picture may come from physical deployment proxies — AI-related capex and activity across the semiconductor supply chain — several of which have stayed resilient through recent volatility. South Korean exports for the first ten days of June were up 85.9% year-on-year. GPU availability, by some measures, is near zero as demand keeps climbing — 3Fourteen Research’s GPU Availability Index has fallen from 80% in January to roughly 9% now, and GPU availability tends to lead pricing. It's hard to call an earnings peak when demand is growing exponentially and supply is growing linearly.

 

4. “Equity Markets Will Suffer Indigestion”

 

Estimates put $400 billion-600 billion of equity supply coming to market over the next twelve months — large by historical standards, equivalent to roughly 1% of total market cap. But that needs to be set against the demand side. Corporate buybacks remain substantial in aggregate, running at an annualised $1 trillion, or about 1.5% of market cap. While some Magnificent 7 names have pulled back on buybacks in favour of capex, financials (deregulation), oil (war-driven cash generation) and industrials (AI capex) have stepped into the gap. Apple still has $100 billion of buyback authorisation outstanding this year, Nvidia has $80 billion, and Alphabet and Microsoft each have plans for over $50 billion (whether they're used is another question). Warren Buffett is sitting on roughly $350 billion of dry powder, money-market funds hold $7 trillion and banks are re-levering on the back of recent deregulation, notably around the SLR. On top of that, structural, price-inelastic pension flows should continue until dependency ratios fall back toward 1x. On my own rough numbers — assuming $34 trillion in global DC assets, roughly 60% allocated to equities, and a similar share to US equities, growing 2% a year — that adds around $250 billion of incremental equity demand annually, or another 0.4% of total market cap. Absorption may get bumpy at the margin, particularly around lock-up expiries, but it's not yet clear markets have flipped from equity scarcity to sustained surplus. There’s no obvious signs of indigestion, yet.

 

5. “China, No growth, No (Equity) Future”

 

China's bearish commentary remains oddly attached to old-China indicators — fixed asset investment and the like — that have either lost their signal value or are showing clear signs of bottoming. Retail sales are a good example: Strip out the comparison effect from Covid-era subsidies rolling off autos and appliances, and sales for most other categories still grew year-on-year, modestly, in a genuinely difficult environment (Iran and domestic gasoline prices among the headwinds). Property headlines stayed weak, but tier-1 city prices — which tend to lead any recovery — continued to stabilise. On the new growth drivers, May's industrial value added for large enterprises rose 4.5% year-on-year, with equipment manufacturing up 9.5% and high-tech manufacturing up 15.1%, both accelerating from April. Granted, headline growth isn't strong, but its composition is shifting toward higher value-add, less capital-intensive, higher-margin output — a mix that should support market multiples and earnings growth. This is happening just as China sits on record household savings earning low or negative real returns, with new tax incentives nudging savers toward equities. The broader transition — toward a consumption-based, equity-driven, technology-led model — is arguably best expressed in the tech indices: the Star50 (+41% YTD) and ChiNext (+34% YTD) have been among the best-performing major markets in Asia this year, behind only South Korea and Taiwan. That’s a signal you won’t see if you insist on reading China through property. China’s future won’t be found in its past.

 

Material world drivers

 

I sense some fatigue building in the AI/choke-point trade, and SpaceX's progress on orbital compute — which would remove the need for cooling or power infrastructure entirely — could slow it further by dramatically lowering the cost of GW-scale capacity. But AI is only one leg of this. The "material world" of things, which underpins the more ethereal world of AI or “intelligence,” should keep performing well over coming years on the back of several converging secular demand stories: energy transition, re-shoring and re-industrialisation, national security stockpiling, a possible China rebound, and now India. "All our customers are doubling capacity. It’s happening. It’s real,” as BHP’s Michiel Hovers observed after a recent trip to India. New Delhi has set a production target of 500 million tonnes of steel by 2047 — more than triple last year's 165 million tonnes. Here's hoping China's property market stays subdued, since markets reward scarcity, and picks and shovels are already in scarce supply. But don't forget the shovellers — the miners — themselves. They still trade at fraction of the value of the sector that relies on them. Indeed, the combined market cap of all listed miners is roughly equivalent to what SpaceX has generated in value since its IPO.

 

No exuberance here. But still, plenty of constraints.

 

Have a great week.

 

Gary