I was staring at the screens late last week when something clicked. Credit spreads had widened just enough to look attractive on paper, yet the selling pressure never really went away. At the same time diesel futures kept grinding higher and the big technology names that have carried so much of the market started showing cracks. It felt less like a normal pullback and more like the market quietly raising a yellow flag. That combination of cheap-looking credit, stubborn energy costs, and the enormous weight of artificial intelligence infrastructure spending is what keeps me up these nights.
Why Energy Costs Refuse To Ease
Energy is the easier part of the puzzle to understand, at least on the surface. Brent crude has climbed back to levels last seen in May. That alone grabs attention. But the real pressure sits deeper in the distillate complex. Low sulphur gas oil futures, the contract that tracks diesel most closely, have shown almost no relief even as various efforts tried to push prices lower. Diesel matters more than crude for the real economy. It powers trucks, agriculture, and a big slice of industrial activity. When diesel stays elevated everything downstream feels the squeeze.
A lot of chatter claims more oil is now flowing out of the Middle East than before recent conflicts. Some of that may be true. Workarounds and nighttime sailings without transponders probably move more barrels than official trackers capture. Iranian exports have slowed to a trickle under effective pressure, which removes supply from the market and forces buyers elsewhere to pay up. That part is working as intended for geopolitical aims, yet it keeps a floor under prices.
The tougher reality is that liquefied natural gas, diesel, other distillates, and fertilizer have not returned to anything close to pre-conflict volumes. Oil is the most fungible piece of the energy puzzle. Everything else is harder. LNG logistics remain complex. Diesel needs the right feedstock into refineries that still have capacity constraints. Fertilizer shipments are even murkier. Shipping costs have exploded, with some Middle East to Asia routes jumping roughly 250 percent in a matter of months. Longer routes, higher insurance, and general risk premiums all add up. An exchange-traded fund that tracks future crude transport costs exists for a reason. There really is a product for almost everything these days.
Until some form of lasting arrangement takes hold in the Middle East, it is hard to see meaningful relief across the energy complex. A resolution between Russia and Ukraine would help at the margin, yet the Middle East remains the bigger lever. Even after any deal, supply chains would need months to normalize. In the meantime energy stays expensive and that feeds directly into inflation expectations and corporate costs. I keep coming back to this point because it sits underneath every other market conversation right now.
The Quiet Pressure Inside Credit Markets
Credit looks cheap in many names that have already widened. That is the first thing most people notice. The second thing is the growing conviction that those same names can keep getting cheaper unless the pace of new supply slows. Future issuance is the real worry more than outright default risk for many of these credits. That creates an awkward setup. Positions that look attractive on a pure valuation basis can still be expensive to hold if the technical pressure continues.
What stands out is how orderly and selective the widening has been so far. The investment-grade index sits around 63 or 64, still below its March peak near 68. Broader corporate bond spreads reached 89 and then pulled back, remaining under the 93 level seen earlier in the year. Longer-dated investment-grade exchange-traded funds have also stayed inside previous ranges. This is not a broad risk-off stampede. The pressure concentrates in the sectors and names that have the clearest reasons for concern, especially those tied to heavy capital spending.
Still, orderly markets can turn disorderly when supply keeps coming and buyers step back. The old rule applies with extra force right now: far more money gets lost in credit from forced selling than from actual defaults. Orders of magnitude more. That is why a simple gut check matters. If spreads widen another 20 or 30 basis points, will the position still be holdable under risk limits and client pressure? If the honest answer is no, the time to adjust is before the pressure intensifies.
Two separate forces could push spreads wider from here. The first is pure technicals. Heavy supply meets limited demand and selling begets more selling. The second is fundamental. If the massive artificial intelligence infrastructure buildout slows, economic growth slows with it. That would give spreads a genuine reason to reprice higher across a broader set of names. When technicals and fundamentals line up in a market that often feels thinner than it looks, the move can surprise people with its speed.
Far more money is lost in credit from being a forced seller than from default. Literally orders of magnitude more.
I am sitting in a clear yellow-alert stance on credit. Not red. Not all-clear. Just heightened caution and a preference for quality and liquidity.
How Treasuries Are Reading The Same Signals
The ten-year Treasury yield finished the recent week lower by about eight basis points, including a sharp sixteen-basis-point drop on one heavy session. That kind of move rarely happens in isolation. A slowdown in heavy capital spending reduces competition for capital and simultaneously raises the odds of softer growth. Both dynamics support longer-duration government bonds.
These moves have less to do with near-term inflation prints or the exact path of policy rates than with shifting supply and demand balances and a possible change in the growth trajectory. At the same time, French ten-year yields moved above United States ten-year yields for the first time in more than a decade. Sovereign stress has even begun to touch parts of the banking sector in subtle ways. Nothing alarming yet, but another yellow flag on the board.
In my experience the bond market often senses economic softness before equities fully acknowledge it. That does not mean equities must fall hard. It does mean the easy assumption of uninterrupted growth deserves more scrutiny than it has received for many months.
Semiconductors Sit At A Critical Technical Level
The semiconductor sector sits at the heart of the entire artificial intelligence story. Chips are the physical foundation of the compute buildout. Anything that changes the pace of that buildout hits this group first and hardest. The sector also carries an unusually large amount of leveraged exchange-traded product exposure. Those products amplify moves in both directions because they must buy strength and sell weakness to maintain their target leverage. Add in various volatility-selling income strategies and the potential for mechanical pressure increases further.
Right now the main semiconductor index is parked almost exactly on its two-hundred-day moving average. That level often acts as support on first tests. It is easy to forget how sharp the last break of this average was. What looks like a minor dip on a long-term chart was actually a thirty percent drawdown after the market had already climbed from under five thousand to near eleven thousand. Markets frequently test and retest important averages before deciding whether to hold or break. The current setup therefore feels more like a warning than outright panic, yet the mechanical nature of the leveraged products makes the risk asymmetric.
If the index holds and buyers step in, the amplification works in the upward direction. If it breaks with volume, the same amplification can accelerate the decline. That is why positioning around this level feels more important than usual.
The Rising Shadow Of Lower-Cost Chinese Compute
Reports of a large private equity raise by a major Chinese artificial intelligence company, equivalent to roughly fifteen billion dollars, deserve more attention than they have received. Capital of that size can stretch further in a different cost structure and regulatory environment. History offers uncomfortable parallels. China moved from relative outsider to dominant player in telecom equipment over a relatively short period. Similar patterns appeared later in manufacturing and then in refined critical minerals. The possibility that the same playbook is now being applied to large-scale compute capacity is hard to dismiss.
Average token prices continue to drift lower even if the pace of decline has slowed recently. Separating efficiency gains inside United States companies from a rising share of usage flowing through lower-cost Chinese providers is not straightforward. The data is incomplete. The direction of risk, however, is clear enough. Markets still appear to underprice the potential for cheaper Chinese compute capacity to pressure margins and pricing power over the medium term.
This is not a short-term trading call. It is a structural concern that sits alongside the more immediate questions about the pace and funding of the domestic buildout.
Compute Spending As The Central Economic Driver
Strip away the noise and the single largest story facing both markets and the real economy is the scale of capital expenditure tied to artificial intelligence infrastructure. Reasonable estimates put the contribution of big technology capital spending somewhere between half and sixty percent of the total. The compute buildout and the businesses that support it form the bulk of that spending. On the growth side, of the roughly 2.2 percent real gross domestic product growth, somewhere between three-quarters of a percent and a full percentage point can be traced to this activity. Some more aggressive estimates put the contribution closer to half of total growth.
Those numbers are large enough to matter. They influence employment in construction and manufacturing, demand for power and cooling, semiconductor orders, networking equipment, and a long list of secondary effects. They also influence the cost of capital for every other sector because the same pool of savings and credit is being tapped. When that spending is accelerating it supports growth and risk assets. When the pace slows the reverse becomes true.
The market will receive incremental information through earnings reports and, more importantly, through the reaction to those reports. Guidance on capital expenditure plans, comments on power availability, and any signs of slowing customer demand will all be parsed carefully. One particular initial public offering tied closely to the frontier of large language model development is already drawing intense focus. Earlier private market transactions set high bars. The public market reception will tell us whether valuations still have room or whether expectations have run ahead of realistic growth trajectories even after recent pullbacks.
Putting The Pieces Together
Energy prices need a genuine positive development in the Middle East before they ease in a lasting way. Everything else around oil, gas, and refined products is secondary. The volume and cost of the artificial intelligence infrastructure spend remains the dominant driver for credit spreads, equity valuations, and the broader growth outlook. Treasuries appear to be sniffing out the possibility of softer growth, which aligns with a more cautious stance on both equities and credit.
Technically, neither credit nor the semiconductor complex looks comfortable. Credit has already shown selective pressure and the risk of that pressure broadening remains real. Semiconductors sit on a widely watched moving average with leveraged products ready to amplify whatever comes next. The combination does not scream crisis. It does argue for reduced risk, higher quality, and a willingness to let cash earn a decent return while the picture clarifies.
I have found that the most useful discipline in these periods is to separate the narrative from the positioning. The narrative around artificial intelligence remains powerful and long-term constructive. The positioning and the near-term funding requirements create vulnerabilities that can produce sharp moves in either direction. Respecting both sides of that reality tends to keep portfolios intact when the path gets choppy.
Looking ahead, the next stretch of trading sessions will bring more earnings, more commentary on capital plans, and the first real public-market test of one of the highest-profile artificial intelligence companies. Both bulls and bears will find moments that feel validating and moments that feel uncomfortable. That is the nature of markets when the largest growth driver of the cycle is also the largest source of capital demand and technical pressure. Staying flexible, watching the energy complex for any genuine relief, and treating credit as a market that can still get cheaper even when it already looks inexpensive are the practical takeaways that keep me focused.
The yellow flags are up. They are not red. That distinction matters. It is the difference between preparing for turbulence and assuming the plane is already going down. Right now preparation still looks like the wiser course.
Practical Considerations For Portfolios
Anyone managing money through this environment has to balance several overlapping risks. Energy costs that refuse to decline keep input prices elevated for a wide range of industries. Credit markets that look cheap on a historical spread basis can still cheapen further if supply remains heavy. Equity valuations in the technology complex remain elevated relative to many historical measures even after recent softness. At the same time the growth contribution from the artificial intelligence buildout is large enough that any material slowdown would show up in broader economic data.
One practical approach is to favor higher-quality credit over the names that have already widened the most simply because they look cheap. Liquidity matters more than usual when the risk of forced selling is elevated. In equities the semiconductor complex and the broader technology group deserve position sizes that reflect the mechanical amplification risk from leveraged products. That does not mean avoiding the group entirely. It does mean recognizing that daily moves can be larger than the underlying fundamentals would suggest on their own.
Duration in government bonds has looked more attractive as yields have backed off and as growth concerns have quietly risen. The French-United States yield inversion is a reminder that sovereign dynamics can shift in ways that eventually touch corporate funding markets. Keeping an eye on cross-market relationships often provides earlier warning than watching any single asset class in isolation.
Perhaps the most useful mindset is to treat the current period as one that rewards preparation more than prediction. The range of plausible outcomes remains wide. Energy could ease if geopolitics cooperates. The artificial intelligence spending cycle could continue at full pace for longer than skeptics expect. Credit supply could slow if issuers become more disciplined. Equally plausible is a period in which energy stays elevated, capital spending growth moderates, and credit markets demand higher compensation for the uncertainty. Portfolios built to survive the second path while still participating in the first path tend to weather the noise more gracefully.
What The Coming Weeks May Reveal
Earnings season will provide the next batch of hard data on capital expenditure plans and customer demand. Comments about power constraints, delivery timelines for specialized chips, and any signs of pushback on pricing will all be scrutinized. The market reaction to those comments often matters more than the absolute numbers themselves. A company that guides to slightly lower spending but is met with relief can be just as informative as one that raises spending and is met with skepticism about funding and returns.
The highly anticipated public offering of a leading artificial intelligence company will serve as a real-time referendum on how much growth investors are still willing to pay for. Earlier private valuations set a high bar. Any sign that public market participants are more selective about growth assumptions or more focused on the path to cash flow would ripple through the entire complex. Conversely, strong demand would reinforce the existing narrative and likely support risk assets more broadly.
Energy markets will continue to respond to every headline out of the Middle East and every data point on shipping volumes and refining margins. A single meaningful de-escalation could produce a sharp decline in diesel and related products. Until that happens the path of least resistance remains higher prices and continued pressure on the real economy.
Credit markets will keep watching the calendar of new issuance. A quieter period of supply would allow spreads to stabilize and potentially tighten. Continued heavy issuance would keep the pressure on, especially in the names already trading wide. The difference between those two outcomes is large enough to matter for total returns over the next quarter.
I keep returning to the same core observation. The artificial intelligence infrastructure cycle is large enough to drive meaningful economic growth and to create meaningful market distortions at the same time. Respecting both the opportunity and the risks is the only stance that feels durable right now. Yellow alert is not a call to hide. It is a call to stay awake.
Markets rarely move in straight lines when the dominant theme is both a growth engine and a source of capital intensity. The weeks ahead will test how well investors have prepared for that reality. Those who have already adjusted position sizes, upgraded quality, and kept some dry powder will likely navigate the path with fewer forced decisions. That, more than any single forecast, is what tends to separate satisfactory outcomes from painful ones when the flags are yellow and the fog is still thick.
