Nobody Said No

The wizards of Wall St are at it again. From the people who brought you ABP, CLO, and MBS, a new game in town has emerged. But the same risks come with it.

Jensen Huang, Chairman of NVIDIA, has the street in the palm of his hand. His proposal to reshape the financing of the AI build-out was a big ask, but in his words, "nobody said no." In an effort to put up a firewall between his company and its customers, he's creating a SPV (Special Purpose Vehicle) for funding the sales of his chips to end users. We see this mechanism in other industries - aircraft leases

being the most comparable example.

But the risks to the investment in these instruments _ (CBP, chip-backed paper?) _ remain the same. As experience shows, the amount of leverage applied to any asset class often reaches its breaking point. And when that asset is something as ephemeral as "compute" - the noun, not the verb - it is even more risky than a 30-yr mortgage ever was.

But in the mania of a gold rush, a miner will mortgage his soul for a good shovel, and there will always be someone willing to sell him one.

For now, the AI infrastructure growth and adoption phase is going as smoothly as it picks the low-hanging fruit, such as coding. Eventually, the race to build an AI ecosystem will come into conflict with the economy's ability to pay for it. And as we saw with the Internet bubble, the promise of the technology often far exceeds the reality of the benefits. Just as lending standards were relaxed to offer jumbo mortgages to undeserving borrowers back in 2008, so too will the purveyors of compute lend to a bad credit at some point.

This is not today's risk, however. We are in the early innings of the AI buildout, and the first few nuggets of gold are being unearthed as we speak. Many AI applications are already producing lower costs and higher output. The software industry is rapidly pivoting to incorporate AI into its products, and the beaten-down stocks — like Microsoft — have recovered nicely.

Just file away this issue of my blog for future reference. Once we get closer to the end of the $3tn buildout, then we can worry. For now, you will only be left behind if you say 'no'.

Risk Model: 2/5 - Risk Off

Just when we were getting started, the overbought stock market is making it hard to go 'all in."

The Copper market is lagging the gold rally, and this signal of economic strength is suddenly weakening. On top of last week's tepid labour market data, there is something of a summer doldrum feeling to the economy. And if you like/don't like the oil headlines, wait 30 minutes and they will change. Copper specs and positioning have been stretched for a few months now, and tariff-related news may cause a setback shortly.

AAII sentiment has improved as implied volatility has subsided to low levels, but it has tended to lag the market's direction this year. Protection against a possible downdraft during the upcoming geopolitical and election-related phase is cheap. I would be a buyer of puts here. As shown in the SKEW chart (cost of puts vs. calls), protection has gotten cheaper. Seasonality is not in favour of taking any outsized risks over the next few months.

CBOE 3-mo Volatility Index

CBOE SKEW Index