Circular reasoning

The current era of high concentration in major indices begins a spiral of ever decreasing circles. And risk supports become interrelated.

Three supports of market pricing and the promise they must keep.

• that passive flows continue, and increase
• that index investment leverage to the anticipated AI payoff, pays off
• and benign credit markets don’t become malignant

This is no under-analysed boom. The rise of concentration in indices and the leverage to AI outcomes is well known. The specific shift we address here is the shift to AI spend being supported by borrowed money not earnings.

The bond market’s unease about it was the subject of the previous piece.

The bond market is renegotiating terms on AI debt. Insuring it costs more than it did. On Oracle, the largest single AI borrower, that cost has gone past two hundred basis points — the highest it has ever been — from 144 in January and 129 in February. The others have moved the same way, less far.

This is not a prediction of stress. The price of insuring debt increases the more there is, and Oracle issues the most. A sustained rise in the cost of that money is a different matter. The equity market shrugs and advances on.

In July S&P cut Oracle to one notch above junk, and put a number on the circularity: roughly half of its $638bn order book is a single private counterparty.

Value is subjective. Price relies on supports. Each one bearing a weight of expectation. And investment expectations across the board, have become unusually uniform. Not in opinion but in the movement of money.

None of this is extraordinary, far from it. But what is unusual is how tightly arranged and interrelated it has all become. And the signals it gives off.

The S&P 500 is more concentrated than at any point in its modern history with ten companies making up forty per cent of the index. Passive flows into market cap index funds accelerate the concentration into megacap companies. Index funds passed active in 2019 and now hold close to two thirds of US equity fund money — $15.2tn against $8.6tn. Which are focused primarily on a single bet. Capital spent on artificial-intelligence capacity will earn its return.

This expenditure is well documented and unparalleled in scale. Now commitments are no longer met by internal cash flow.

Cash flows produced by the greatest of monopoly businesses, the most profitable and cash-generative companies ever built.

Then there is the bid.

Equities are bought, at the margin, by passive funds. Automatic flows which are price insensitive. It is a ‘fixed’ bid that has not faltered and has grown in share to dominate. The US indexes are widely owned and the earnings story at the centre depends increasingly on circular internal commitments.

The credit markets have something to say about it.

THE MARKET CONCENTRATES

Five buyers of AI capacity — Microsoft, Amazon, Alphabet, Meta and Oracle — and one seller of chips.

These drive a meaningful share of the whole index’s earnings growth. Growth dictated by capital spending.

In the quarter to 26 July, Nvidia sold $96.2bn of hardware, twice what it sold a year earlier, and kept $59.7bn of it as profit. It expects to sell $108bn in the next one.

The seven largest are about a third of the index. At the end of 2018 they were thirteen per cent. It is the highest concentration in the top seven since the Nifty Fifty of the early 1970s.

The reversal of that concentration saw large drawdowns in the leaders.

SUPPORT REGIMES

• Passive flows are a stabilising force and unshaken by events. Market cap weighted index funds are a structural driver of concentration.
• Concentrated AI-related spend channels into a singular payoff expectation. The biggest monopolies leaning on their separate businesses to participate.
• Well-behaved interest rates.

The cost of money is now the one under negotiation. The bond market is eyeing up a lot of supply and wants a better price on it.

The shortfall between capital spending and earnings is funded in the bond market. Future spending commitments continue to rise.

The expected earnings stream relies on the promised commitments of the buyers of compute. Some of them AI companies. The start-up leaders are private companies looking to go public. Which, if successful, would be among the fastest aggregations of value in history. The technology is real.

It’s also a spending arms race with a presumed one or three winners. Each competitor is leaning on their empires to do it and, increasingly, the credit markets.

Concentration is less visible in its relationships. It relies on circular internal beliefs. That passive flows are uninterrupted. And they continue to concentrate in US indices. And AI spend accelerates. Everyone within the centre must believe the others’ spending commitments for the future.

Active private funds hold the off-balance-sheet commitments. Which is marked to market on the assumption of perfect liquidity. Passive funds hold the stock. Everyone commits to spending big.

Markets have always had structural fragilities. And regular tests of them are not hypothetical. The concentration and the passive bid are not new. Non-participation in it has been the real risk.

But knowing they exist is valuable. Knowing how they break and seeing it in real time could be priceless.

THE BOND MARKET WEIGHS IN

The three scenarios the owner and active buyer of hyperscaler debt must handicap.

• All is well. Sheer spending might overcome all obstacles. Big investment pays off and transforms the economy. · Low rates forever.
• Muddle through. Earnings take longer than expected. Volatility rocks the boat. · Middling rates with lots of panicky movement.
• ‘Not failure early tech cycle bust.’ An interim hiccup along the way. AI is certain to transform our economy but the profound shifts come after a build out bust. · High rates and rising.

The strength-overcomes scenario is currently under test. Muddle through is the most likely path. Historical precedent would tip, overwhelmingly, in favour of the ‘not failure but…’ outcome. The early cycle clear out. Which is followed by the true transformation promised by the technology.

By any reckoning it is still early in the AI story. The technology will shape everything from here on out. As with all the others that went before it.

MUDDLE THROUGH. KEY BELIEFS.	WHAT IT MEANS — WE MUDDLED THROUGH
Free cash flow across the five turns positive again while capital spending continues	Earnings caught up.
Hyperscaler credit protection stays in range and absorbs new supply	Spreads widening is supply, not stress.
Revenue at the buyers of compute continues to ramp	Circular financing was a bridging loan.
Breadth widens — innovation and smaller cap activity increases	Concentration can unwind in a healthy way.
Passive flows stay positive throughout	The core support remains strong.
Private-equity enjoys continued liquidity	Volatility is tamed.

Should one of the drivers of concentration stutter, disturbances get passed around. This is not to do with sentiment but built into the structure. The coil of concentration. Everything becomes more closely correlated. Market structure can drive drawdowns under certain conditions; sentiment usually catches up later.

‘NO. 3 NOT-FAILURE’ — EARLY INDICATORS	WHY IT MATTERS
Cost of insuring AI debt	Early signs of stress.
Capital spending against cash flow	The gap is what the bond market funds.
Breadth narrowing	Concentration deepening, not unwinding.
Passive fund flows	The support that has never been tested.
Private-market stress	Marks that assume a liquidity nobody has priced.
Interest rates	The discount rate under every one of these.

The supports under this market have been reliable for a long time. Concentration creates correlation and they now all lean on one another.

Return to the three promises at the top. That passive flows continue. That the leverage to AI pays off. That credit stays benign. Each was a separate proposition, held by different people for different reasons. They are now one. Held by the same money, resting on the same handful of names and balanced on the same risk factors.

That is what the tightening coil means. Not that any of them breaks — only that they can no longer break separately. An index fund, a corporate bond fund and a private-credit allocation look like three decisions. They are increasingly one.

And the price being paid for all this? There appears to be lots of borrowed goodwill included.