Every power station has two numbers. The first is its nameplate capacity, the output it could deliver under perfect conditions. The second is what it can actually be relied upon to deliver at 6pm on a cold, still winter evening when everyone gets home and turns the heater on. Engineers care about the second number. Politicians, press releases and, for a long time, markets have preferred the first. 

That distinction is worth borrowing, because the gap between what was promised and what can be delivered is turning up in more than one place at the same time. It is showing up in electricity grids. It is showing up in sovereign debt, most visibly in French government bonds. And, if Ray Dalio’s latest numbers are anywhere near right, it is showing up in the US Treasury market and in the vast pipeline of debt and equity being issued to fund the AI build out. 

For long term investors, the common thread matters more than any single headline. Markets appear to be shifting from rewarding promises to pricing deliverability. That shift tends to be slow, then sudden, and it changes which assets deserve a premium. 

The 8% problem 

Start with electricity, because the arithmetic is cleanest there. 

In July, Beijing released the renewable energy plan that sits under its 15th Five Year Plan. The headline target is more than 2,800 gigawatts of wind and solar by 2030, roughly 38 times the capacity of Australia’s entire National Electricity Market. That is the number that travels. As Chris Uhlmann pointed out in The Australian last week, the more interesting numbers are buried further down. 

According to Uhlmann’s reading of the plan, China’s planners expect wind and solar to make up around half of installed capacity by 2030 but to produce only around 30% of the country’s electricity. More striking still is how much of that capacity they believe they can count on when the system is under stress. China’s planners call this confidence output, a measure of average available energy across periods of peak demand. Their answer, on Uhlmann’s account, is about 8%. For every 100GW of wind and solar, including farms with batteries attached, planners assume roughly 8GW will reliably show up when it matters. 

China has priced that distinction too. Reforms introduced in January pay coal and gas generators for dependable capacity, value batteries according to how long they can keep delivering, and push wind and solar off guaranteed prices and into the market. China has not stopped building renewables, far from it, but its official messaging has shifted from adding capacity to building a system that can actually integrate it. 

Uhlmann then applies China’s benchmark to Australia’s Capacity Investment Scheme, which targets 40GW of new capacity by 2030, around 26GW of wind and solar and 14GW of mostly battery based clean dispatchable capacity. On his assumptions (four hour batteries and a prolonged lull in wind and sun), that 40GW shrinks to something closer to 4GW of dependable power. Professor Alex Coram’s point, which Uhlmann quotes, captures the engineering problem neatly: “You can’t buffer your way out of ruin.” 

Reasonable people will dispute the assumptions, and energy policy is contested ground. The investment lesson does not depend on who wins that argument. When the system is stretched, reliability becomes scarce, and scarce things get repriced. 

France, sovereign debt and the gap between promise and payment 

Now swap gigawatts for euros. 

Robert Swift, who manages TAMIM’s global portfolios, has rechristened the European problem in his latest commentary. Fifteen years ago the market fretted about the PIIGS. Today the acronym is FROGS: French Republic OATs Go Soggy. The joke lands because the underlying point is serious. The 2012 crisis was resolved largely by deferral, and the fiscal repair job France avoided then is harder now. 

The numbers have moved quickly. In February the spread between French and German 10 year government bonds was around 55 basis points. By 5 October it had blown out to roughly 147 basis points, the widest since the eurozone debt crisis, leaving France paying more than Italy or Greece. French 10 year yields are approaching 5%. Public debt sits at roughly 119% of GDP and the 2026 deficit at around 5.4%, both far beyond EU reference levels. France has cycled through four prime ministers since the snap election of June 2024, and the 2027 budget faces Assembly votes on 20 and 27 October, with Moody’s reviewing the sovereign rating on 23 October. 

Help from Berlin looks less likely than it once did. Chancellor Merz’s CDU suffered historic losses in two state elections in September, and German voters appear more interested in problems at home than in underwriting a neighbour’s entitlements. As Robert observes, the French have a long tradition of expressing displeasure at changes to their entitlements forcefully, which makes the obvious fiscal fixes politically expensive. 

In nameplate terms, France has promised its citizens a level of spending its tax base cannot reliably fund. The bond market is doing what grid planners do, quietly de-rating the promise. 

Dalio’s arithmetic: the same problem, much bigger 

If France is the live case study, the United States is the main event. Ray Dalio’s note of 13 October, which updates the framework from his book How Countries Go Broke, is long and dense, so it is worth distilling. 

His starting point is simple. Countries get into trouble when the supply of debt to be sold is much larger than the demand for it, or when debt service costs rise so far relative to income that they squeeze out other spending. On his numbers, both conditions are now present. 

Marketable US Treasuries outstanding were about US$32 trillion at the end of September, 8% higher than a year earlier, with around US$2.1 trillion of net new issuance expected in 2026. Demand for long dated bonds has softened enough that the Treasury has shifted issuance heavily towards short term bills, roughly 50% of issuance against a historical norm closer to 20%. That makes borrowing cheaper today but means more debt must be refinanced sooner, at whatever rate prevails at the time. 

Interest is the second squeeze. The average rate on existing Treasury debt is around 3.5%, while market rates sit between 4% and 5.5% across the curve. Every maturing bond therefore rolls into a higher coupon, even if rates go no higher. Dalio estimates that a 1 percentage point rise in the average rate eventually adds around US$310 billion a year in interest, and that interest already absorbs about 21% of federal revenue. Add maturing principal and total debt service is running at roughly 200% of revenue. 

The buyers are changing too. Foreign investors hold about 32% of Treasuries and the Federal Reserve about 14%, but recent marginal demand has come increasingly from domestic buyers. Dalio argues that both China and Japan now have reasons to hold fewer US bonds, not more, and that an unwind of Japanese currency hedged positions could be disruptive. 

His conclusion is that we are in the middle of the cycle: rates have already risen a long way, but the usual consequences (weaker equities, wider credit spreads, slower growth) have not yet fully arrived. 

The new marginal borrower is building data centres 

The part of Dalio’s analysis most relevant to equity investors is not about governments at all. 

Treasury is competing for capital with every other issuer, and the fastest growing issuer is the AI complex. Hyperscaler bond issuance reached about US$220 billion in the first half of 2026, more than double the roughly US$108 billion issued in all of 2025. Net US corporate debt issuance is expected to rise from about US$700 billion in 2025 to about US$1.4 trillion in 2027. Non financial corporate equity issuance hit an annualised pace of around US$1.8 trillion in the first half of this year, driven by IPOs, secondary raisings and private fundraising by the frontier AI developers. 

Dalio describes AI corporate debt as the canary in the coal mine. The logic is straightforward. Higher capital spending funded by more borrowing reduces free cash flow; investors demand wider spreads to absorb the supply; required returns on AI investment rise. That does not mean AI fails as a technology. It means the financing chain is becoming more demanding at the same moment governments are asking the same pool of savers for more money. 

There is a pleasing symmetry with the electricity story. Data centres need dependable power, which is precisely the commodity grids are discovering they have overstated. The AI boom is borrowing heavily to build demand for the very thing that is becoming scarce. 

How governments usually get out 

Robert’s commentary includes a useful framework, listing the five ways governments typically reduce excessive debt. Dalio reaches a remarkably similar list from the opposite direction. 

The first is austerity, which is economically sensible and politically unpopular; French streets provide a regular reminder of how unpopular. The second is default, which developed economies avoid for obvious reasons. The third is inflation, which only works if it is concealed, and which is hard to conceal when inflation is already the thing everyone is watching. The fourth is running a global trading empire and effectively appropriating other countries’ assets, the British solution after the Napoleonic Wars and not a realistic option today. 

The fifth is financial repression: steering savers, banks and institutions into government bonds at yields below inflation, often under the banner of prudential regulation. US banks held roughly a third of their assets in government bonds in 1947, under 10% by 2008, and around 20% today. There is room for that share to rise again. Dalio’s list adds a sixth outcome, asset prices simply falling far enough to attract buyers, which is the market’s own solution when policy makers do not choose one. 

The most likely path, on both men’s reading, is some combination of mild austerity, regulation that nudges domestic savers into government debt, and a quiet acceptance of inflation running a little above target. That is not a crisis scenario. It is a slow transfer from savers to debtors. 

What markets may be missing 

Put these strands together and a few second order effects emerge. 

The era of share buybacks as the default use of corporate cash may be ending. When capital is plentiful and cheap, buying back stock is easy. When governments and AI builders are competing for the same savings, companies that need external funding pay more for it, and companies that fund themselves gain a real advantage. 

Long dated bonds may not provide the defensive ballast investors are used to. If the escape route is financial repression, the holders of long duration government debt are the ones being repressed. 

Reliability is becoming a priced asset, in electricity and well beyond it. Dependable generation, firm energy supply, essential infrastructure and businesses with genuine pricing power all share the quality grid planners now pay for: they show up when the system is under stress. 

Risks and counterpoints 

None of this is a forecast of imminent crisis, and there are good reasons for humility. 

Dalio himself notes the one historical exception to his template: a genuine productivity and income boom large enough to grow economies out of their debt. If AI delivers on even part of its promise, today’s arithmetic could look very different in five years. The productivity optimists may yet be right. 

Policy can delay reckonings for a remarkably long time. Japan has carried government debt above 200% of GDP for years without a classic crisis, and the European Central Bank has tools to cap spreads if it chooses to use them. Markets that look stretched can stay stretched far longer than a sensible investor expects. 

On energy, battery costs keep falling and storage durations keep lengthening, and Australian planners would argue that the Integrated System Plan already accounts for firming needs. Uhlmann’s critique is a strong argument, not a settled conclusion. Timing, as ever, is the hardest variable of all. A diagnosis can be correct for years before it is profitable. 

What should long term investors actually do? 

This is not a call to make heroic macro bets or to sell everything and buy gold. It is a reminder to apply the grid planner’s test to portfolios: distinguish what looks good on paper from what can be relied upon when conditions turn. 

In practice, that suggests favouring businesses that generate their own cash and do not depend on cheap external funding; being cautious about valuations that rest on capital staying abundant; questioning whether long dated government bonds will play the defensive role they once did; and paying attention to companies that own scarce, dependable assets, whether that is firm energy capacity, essential infrastructure or a product customers cannot easily do without. 

Robert’s global portfolios remain underweight Europe, but he describes himself as dancing close to the door. If French and European politics produce a credible fiscal plan, European equities could re-rate quickly. That optionality is worth keeping in mind, because the moment the obvious solution is adopted is usually the moment the opportunity disappears. 

TAMIM Takeaway 

French bond spreads have blown out to crisis era levels, Ray Dalio’s numbers show the US debt burden and the AI financing boom competing for the same pool of capital, and even China’s energy planners are quietly discounting most of their wind and solar capacity when the system is under stress. The common thread is a growing gap between nameplate promises and dependable delivery. 

The market may still be pricing many assets as if capital will remain cheap and promises will be met in full. For long term investors, the more useful mindset is the engineer’s: value what will reliably show up when conditions are difficult. Self funding businesses, sensible valuations and genuinely scarce assets tend to hold their value when the arithmetic finally catches up with the rhetoric.

__________________________________________________________________________________________

General advice disclaimer: This article contains general information only and does not take into account your personal objectives, financial situation or needs. Before making any investment decision, consider whether it is appropriate for your circumstances and seek professional advice if required. Past performance is not a reliable indicator of future performance.