By Peter Tchir of Academy Securities

Last weekend, in If This Was A Normal Week…, we focused on diesel and credit risk. Here it is, another weekend, and it seems like we once again need to focus on diesel and credit risk. We got to discuss this on Bloomberg TV on Monday. Bloomberg published this segment of the entire interview as a separate clip: AI Debt May Trigger Sharp Sell-off. It was also fun to get on the floor of the NYSE with the Schwab Network where we discussed Yellow Alert for Credit and the Risks of Cheap Chinese Compute.

While the concerns looked a little silly early in the week, after gains for big tech on Monday and Wednesday, they looked a lot more realistic by the end of the week. Thursday’s selling pressure brought the risks back to the forefront (the FT picked up on our view of what was driving markets).

They are two separate stories. Diesel is basically a function of geopolitics and conflict. Credit Risk and Equity Risk are primarily a function of the compute buildout: the pace of the compute buildout, the ability to fund it, and the profitability of the compute once it is funded and built.

Compute Spend is the single biggest story facing the economy and markets.

• Using Grok, it seems reasonable to attribute 50% to 60% of Capex spending on big tech, with the compute buildout and adjacent businesses comprising the bulk of that Capex.
• Estimates, according to Grok, are that of the 2.2% in real GDP growth, somewhere between 0.75% and 1% can be attributed to the compute buildout. At the higher end are some estimates that as much as 50% of U.S. GDP growth is being driven by the spend.

The importance of this compute spend to markets and the economy cannot be overstated.

Having said that, let’s start with energy, because it is “easier” and then we can work our way through credit, equities, and compute.

Diesel, Oil, Fertilizer, Iran, Russia, Ukraine, Shipping, and Refining

With Brent futures hitting their highest level since May, it is impossible to ignore what is going on with oil and gas prices. To be honest, my “go-to” commodity contract is QS (Low Sulphur Gas Oil futures) as it ties most closely to diesel, which is the “real” problem (that along with LNG and fertilizer).

There has been very little relief in this contract despite recent efforts to get diesel prices lower.

There is a lot of “chatter” about more oil flowing out of the Middle East now than before the war started. As with so much right now, it is difficult to figure out the veracity of these claims. Here are my working assumptions on oil and distillates:

• Bypasses and workarounds are getting a good percentage of the oil that once went through the Strait onto the market. Good.
• Strait Traffic Trackers understate how much oil is making it through the Strait as ships run at night, without transponders, etc. Also good.
• Iranian oil flows have slowed to a trickle. Most of the oil they had at sea (outside the Gulf) will have been offloaded by now. The U.S. blockade is working extremely well. Without Iranian oil flowing to China, there is less oil on the market and China has to bid up oil in “normal” markets. Good for clamping down on Iran, bad for the price of oil.
• Unclear that the flow of LNG, diesel, other distillates, or fertilizer has returned to anything close to pre-war levels. Just using Grok, LNG might be 25%, diesel under 50%, and “it throws up its hands in surrender” on fertilizer shipments versus pre-war. Oil is the most fungible and easily moved part of the energy supply chain. LNG is the most complex. Diesel requires the right product being brought into refineries with capacity, and that is a bottleneck. Oil is the “easy” problem; other products are more difficult and not doing well. Bad.
• The cost of shipping is much higher. Longer routes (the bypasses take more time, increasing the cost and time to market). Higher insurance costs, etc. If you want to get all “wonky” you can pull up the FD3CM0 futures contract (Middle East to China shipping cost per metric tonne). For those who want to have more “fun” just check out the ETF BWET.

Until there is some sort of a deal in the Middle East, it is difficult to see the energy complex getting cheaper any time soon. A deal between Russia and Ukraine would help, but the key is the Middle East. In any case, it will take months after a deal for the supply chains to “normalize.”

Credit: Cheap, But Getting Cheaper?

There are two dominant themes that come up when talking credit:

• The names that have widened the most all look very “cheap.” In general, this is far less about concerns about default, and much more about concerns regarding future supply. The corollary of that is that these names look very expensive to short. Could that sentiment mark the bottom in credit?
  • The problem is that despite looking cheap, there is real concern and conviction that unless something changes in the pace of supply, they will continue to get cheaper.
• The pressure on credit is largely in the names and sectors that deserve it. There isn’t “widespread” spread widening. CDX (IG index) is still only at 63.5 (lower than its high of 68 back in March). The Bloomberg Corporate Bond Index OAS (spread) hit 89, still below the 93 it got to in March. LQD (a traded long-dated IG bond ETF) spread only got to 120 this week, lower than where it was as recently as August. So far this is both orderly and directed. There is a clear narrative around the names and sectors that have widened. As a whole, the market is trading in an orderly manner (the “beta” products all seem to be behaving “normally”).
  • One risk is that the cheapness and supply turn that orderly and directed behavior into selling pressure that gains momentum. Sell what you can, not what you have to. Any sort of breakdown in how the beta products are trading. Please read last weekend’s report for these concerns, which have NOT gone away. If anything, they may be intensifying. Maybe this is as good a place as any to reiterate my view that far more money is lost in credit from being a forced seller than from default. Literally orders of magnitude more money lost. This is a good time for a “gut check”: if spreads widen another [pick a number], will I really have the fortitude to hold positions (will management or risk let me hold them)? If the answer is no, it is time to be extraordinarily careful!
  • If the compute spend slows, the economy will slow, which is a “real” reason for spreads to widen. Everything in the bullet point above is “technical” in nature. This would be “fundamental” in nature.
  • If technicals and fundamentals align, in this world of “faux” liquidity, we could be surprised by how quickly spreads widen across the board.

I’m in a solid “yellow alert” or “caution flag” phase on credit spreads here.

Treasuries

The 10-year Treasury yield finished the week 8 bps lower, including a whopping 16 bps on Thursday.

• Compute spend slowing is good for Treasuries. Less competition.
• Compute spend slowing is bad for the economy and hence good for Treasuries.

These moves have little to do with inflation levels or Fed policy; they are primarily about potential shifts in supply/demand and potential changes in the trajectory of economic growth.

Also, French 10-year bond yields were higher than U.S. 10-year bond yields for the first time since 2013.

We also saw the “sovereign” problem slip into the “corporate” world (or at least the banking sector) this week.

Nothing alarming, but at the same time, another “caution” flag out there for credit risk.

Semiconductors

We will focus on the chip sector for two reasons:

• It is the lifeblood of the compute spend. So, anything we believe about compute spend will impact this sector.
• The relative size of leveraged ETFs in this sector (and I do lump in TQQQ) is quite large. I view this sector as being more exposed to “mechanical” buying and selling due to the proliferation of leveraged ETFs (at the index and single-name level) as a large amplifier of moves in both directions. If you add in “vol suppression” strategies (equity vol selling as part of “income” strategies), you have a further potential catalyst (to the downside right now) if there are any outflows from those sorts of funds. That is a secondary concern, as the leveraged ETFs are my primary concern for this sector.

Which brings us back to a “technical” level.

The SOX index is sitting right at the 200-day moving average. A level that should/could act as support. It is almost difficult to believe that with all the chatter about all-time highs (and the Nasdaq 100 isn’t that far off), we are back to sitting at the 200-DMA for this crucial sector.

It seems almost impossible to believe that the last time it broke this level was in the aftermath of “Liberation Day.” What looks like a tiny blip on the chart is actually a 30% drawdown! It looks like a tiny blip because 18 months ago, the index was at less than 5,000 and it is just below 11,000 now.

It is common that an index will test and retest a level before breaking through (if it is going to break through that support). So more of a warning flag (again) than true fear, but this setup is dangerous, especially given the “mechanical” nature of the leveraged ETFs linked to this sector where they need to buy on up days and sell on down days.

Chinese Compute

Supposedly DeepSeek raised the equivalent of $15 billion last week in a private sale of equity. I think dog years mean that for every year in human terms a dog ages 7 years? Where the heck did that thought come from?

I cannot help but wonder if $15 billion for a Chinese AI company goes a LOT further than a similar raise in the U.S.

I am incredibly concerned that just like in manufacturing, and then rare earths and critical minerals (the processed/refined/smelted versions), China is preparing to dump a lot of Cheap Compute on the world. Oh yeah, as General (ret.) Groen reminds me all the time, Huawei went from nowhere to dominating telecom equipment in much of the world.

Measures of average token price continue to decline (though they have been more stable of late). What is difficult to ascertain is how much of that is from greater efficiency from U.S. companies, versus a greater percentage of token usage coming from China?

The jury is still out on this, but the threat of Cheap Chinese Compute still seems underpriced.

Bottom Line

On energy prices, we need to see a positive resolution in the Middle East. Everything else seems like, more or less, a rounding error.

The amount and cost of the compute spend remains the key driver for credit spreads, equities, and the economy. We will get bits of information from earnings, and maybe more importantly, from the market reaction to those earnings.

If anything, I’d argue that Treasuries are “sniffing out” potential economic weakness, which would fit my cautious view on equities and credit spreads.

I don’t like how credit or the semis are positioned from a technical standpoint.

The entire market is going to be laser-focused on the Anthropic IPO. SpaceX was important and pivotal for markets, but given Anthropic’s importance to the compute build, we will all be looking for any signs of growth potential not priced in or any sign that valuations may have gotten ahead of themselves (even after recent pullbacks in the compute/AI stocks).

Looking forward to a busy start to the week and what should (unfortunately) be another tricky week to navigate (both bulls and bears had plenty of opportunity to feel smart and stupid last week; I know I did, and I don’t see it getting much easier this week).