Executive Summary:
Market Stagnation & Macro Pressures: The headline indices are experiencing remarkably low 60-day realized volatility, masking significant internal breadth deterioration, shifting real yields, and moderating global growth expectations as the broader economy absorbs tighter financial conditions.
The AI Capex Reality Check: The AI narrative is undergoing a healthy multiple reset as capital costs surge and token spend flatlines, though the massive anticipated volume demands of Agentic AI suggest this is a period of necessary digestion rather than a definitive market top.
Uranium's Structural Deficit: Despite a punishing 50% sector drawdown, the fundamental uranium supply deficit is widening as soaring mining costs and depleted secondary supplies collide directly with Big Tech's aggressive funding of the nuclear renaissance.
Full Commentary:
The stock market ended a volatile week lower as oil prices spiked amid the prospect of renewed strikes between the U.S. and Iran. The Nasdaq index declined -0.7% for the week, but its -5.2% slide over the past two weeks is more notable as the tech trade continues to lose its luster, surrendering its 50-day moving average and retreating below the 26,000 level. The S&P 500 has now dropped for two consecutive weeks, also breaking below its 50-day trendline.
Yet zoom out slightly and the S&P 500 has effectively done nothing for three months. For all the action that's been packed into a given week, the fact is we just keep trading these same levels, over and over again. As a result, 60-day realized volatility measures at a mere 10% (which easily registers in the bottom quartile of the COVID / post-COVID era). In plain English: the headline index is flatlining, creating a deceptive sense of calm that completely masks the violent tech selloffs and internal deterioration happening just beneath the surface. In other words, the index is the calmest thing in the market, and it's also the least informative.
The Nasdaq remains stuck in the range that has been in place since May, despite all the narratives thrown at this market. 29,000 in futures remains the make-or-break level, with the 200-day moving average sitting just below at around 28,500.
Beneath the surface of this headline stagnation, internal breadth metrics have quietly hollowed out. The McClellan Oscillator has pressed firmly into negative territory, and the expansion of 20-day new lows is now consistently overwhelming 20-day highs. This internal deterioration confirms that the recent weakness is broadening far beyond the mega-cap tech complex.
On the fixed-income side, the bond market is telegraphing its own shift in growth expectations. Treasuries caught a powerful bid into the recent oil spike, driving the 10-year yield briefly below 5% for the first time in a month. This dynamic should provide an ideal backdrop for gold. In practice, however, gold has been just another deeply frustrating asset that refuses to perform as expected. It is currently revisiting lows from earlier this summer, down roughly 15% since late August. While its broader weakness makes sense given the general rise in real rates, what is truly vexing is its inability to rally even in a week like this last one, where yields traded down significantly, and gold still couldn't get out of its own way. Once again, it is an asset class punishing investors who are logically following the fundamental playbook.
As for the economy, it has been my contention that the rise in interest rates has thwarted any part of the economy tied to borrowing. Rising oil and gas prices have been a crimp on consumers who are already experiencing a decline in real wages over the past four months—a trend validated by recent BLS data showing real average hourly earnings dropping 0.7% year-over-year in September, with consecutive negative monthly prints into the fall. So, with the AI capex boom under threat, my skepticism of the resilient U.S. economy is on the rise. AI investment has been the U.S. economy's primary growth engine, and should this falter, I fear there isn't enough oomph in the rest of the economy to pull it along. This vulnerability is only magnified by the recent +140-basis-point tightening in overall financial conditions and the Fed pressing forward with its tightening cycle to fight an energy-driven supply shock.
We all know what fuel prices are doing, but look at industrial and agricultural commodities. Lumber futures are down -21% from their mid-July peak. Industrial metals including copper, zinc, tin, aluminum, and iron ore are down more than -7% from recent highs. Despite warnings of mounting food inflation, the commodity markets point the other way: the S&P GSCI Agricultural index has peeled off -11% since early September, with soybeans, wheat, and corn well off their highs. There are two simple interpretations here: global growth expectations are moderating, and inflation is actively rolling over.
Tech & AI: The Capex Reality Check
Roughly 36% of the S&P 500 reports this week, featuring tech giants commanding more than a $10 trillion combined valuation. Keep in mind that 34% of the expected +30% YoY EPS growth rate for the S&P 500 is coming from just two companies (Nvidia and Micron), with a mere ten companies accounting for two-thirds of the total pace.
This arrives as mega-cap tech undergoes a quiet multiple reset. The Magnificent Seven index now trades below a 24x forward multiple—down from its ten-year norm of 29x, and 33x a year ago. Despite this group's massive contribution to overall market earnings, individual valuations are quietly compressing: Microsoft and Alphabet trade at discounts to their historical averages, and Amazon’s P/E multiple is near its lowest level since 2008. That is a meaningful re-rating.
This week is a pivotal, no-win situation. Investors no longer want to hear about aggressive, stepped-up capex plans, yet any deceleration in business spending will immediately punish the “picks and shovels” semiconductor stocks feeding the boom. Pick your poison. Four years into this tech spending boom, there is still very little evidence of a broad productivity breakout. By this stage in the late-1990s Internet boom, the productivity payoff was undeniable (and it still didn't prevent a brutal three-year bear market).
Meta beat earnings last week and the stock still fell -7%. When good results get sold, investors are questioning the AI trade itself. AI demand remains robust, but financing it is becoming increasingly expensive. The market has absorbed more than $500bn of AI infrastructure debt this year, and hyperscaler funding costs have risen roughly 120bps.
Capital is now the whole story. Meta's narrowed $135-145 billion capex range implied a massive Q4 spend, pushing free cash flow after dividends negative. Management noted the need to raise "hundreds of billions of dollars," culminating in a $35 billion stock offering today, with an additional $60 billion share sale program expected around the first quarter of 2027. We all need to come to grips with the enormous level of capital this AI boom is extracting from the markets. Keep in mind, this is occurring alongside Uncle Sam needing an additional $2 trillion per year to fund a bloated budget deficit. All of this raises the question of whether there is enough capital to go around to fund everyone in need. The consequence is the rapid repricing of the cost of capital we are currently seeing play out in the markets.
In a nutshell, the math behind this AI renaissance is getting less forgiving. The demand side of the AI boom, as measured by token spend, has flatlined for three months even as token prices have halved. This implies that the bill for the buildout is rising just as the revenue math gets harder. The market is pricing in the reality that financing costs are higher across debt and equity markets, and AI-native growth rates are slowing (Anthropic revenues dropped from 145% q/q in Q2 to 45% in Q3).
However, I continue to hold the view that we are in a period of digestion (perhaps indigestion) as this life-changing technology becomes more integrated into society. The advent of Agentic AI (think of Meta’s Muse or xAI’s Grokbot, among others) demands 10-100x more tokens than basic inference, meaning the fall in price will likely be offset by a massive increase in volume – call it Jevons paradox in real time. All of this suggests to me that this isn’t peak AI or "THE TOP" for AI and adjacent tech names, but it is clear the market is becoming much more discerning of winners and losers. Keeping some powder dry and respecting the repricing of risk remains the right call as this sorts itself out.
Power Supply & Nuclear Energy: Revisiting the Uranium Thesis
The AI buildout has officially collided with physical reality. As forward tech spending commitments stack up, the market is asking hard questions about the necessary raw materials – specifically power, capital, and a timeline that matches immediate need with long-term efficiency. This has pitted all viable energy molecules in a competitive battle not just to fuel the AI renaissance, but for national security and self-sufficiency. I have been a nuclear energy bull dating back to 2017/2018, and I have the trophies and scars to show for it. But given that the sector is currently in the midst of one of its steepest corrections since we’ve had client capital exposed to the thesis, I think it is timely to revisit the fundamentals.
The story of how we got here is a masterclass in market complacency. For those newer to the thesis, Mike Alkin is the CIO of Sachem Cove Capital Management and arguably the Godfather of the uranium sector; it was through his work that I got turned onto the opportunity. Mike began sounding the alarm on a budding uranium supply/demand mismatch back in 2018. He warned that utility fuel buyers were suffering from severe recency bias, falsely assuming their future reactor needs would easily be met by endless, cheap uranium flooding the market from Kazakhstan. When Mike presented his meticulously researched data to a room full of global nuclear fuel buyers at the 2019 NEI International Uranium Fuel Seminar, he was essentially laughed out of the room. His warnings of a looming structural supply deficit were dismissed as ridiculous.
Fast forward seven years and the spot price of uranium has nearly quadrupled from $25 to $90/lb, and the long-term contract price has soared to an all-time industry high of $97/lb. Last week, Mike gave that exact same group an updated presentation. This time, the laughter was replaced by the visibly worried faces of fuel buyers who are now staring down a brutal "seller's market," with producers demanding contract ceilings of $150–$160/lb, or refusing ceiling prices altogether.
How did the "experts" get it so wrong? For a decade, the industry met today’s reactor requirements with uranium produced yesterday. To put the math in perspective: back in 2018, global primary uranium production was hovering around 135 to 140 million pounds, while global reactor requirements were already topping 170 million pounds. Today, despite mines restarting, primary production is only estimated to reach 160 to 170 million pounds for 2025/2026, while global consumption demands are pushing toward 190 to 200 million pounds.
Fuel buyers were lulled into a false sense of security by industry consultants who counted finite inventory drawdowns and a process called "underfeeding" as permanent new supply. In reality, actual newly mined uranium trailed reactor consumption by roughly 58 million pounds between 2016 and 2025. This massive structural deficit was papered over by 172 million pounds of stockpile drawdowns, creating the mathematical illusion of a 115 million-pound surplus.
The problem is that stockpiles are finite. Once they are sold, they are gone. Furthermore, "underfeeding" – which historically provided 11% of global supply – only works when enrichment prices (SWU) are cheap and demand is low. Today, enrichment prices have skyrocketed. Enrichers are now switching from underfeeding to overfeeding, a dynamic that will drastically reduce that source of secondary supply to less than 2% over the next five years. The bridge of cheap secondary supply has been burned.
Compounding the supply shock is a harsh reality check on mining economics. The market incorrectly assumed historical mine costs would dictate future supply. But as Alkin points out, there are three very different costs in this industry: Cash Cost (producing the next pound today), All-In Sustaining Cost (AISC, which includes sustaining capital, exploration, and overhead), and the Full-Cycle Incentive Price (what a new mine must earn to recover development capital and generate a risk-adjusted return). You cannot sustain an industry on cash costs, and you certainly cannot build new mines on AISC alone.
The math is brutal. Kazatomprom, the world's largest and lowest-cost producer, has seen its AISC soar a staggering +280% since 2018 (when stripping out the mask of currency depreciation), yet their realized uranium prices have only risen +99%. Across the board, from legacy giants like Cameco to the newest ISR start-ups, the industry is delivering far less supply than promised – missing targets by 20% to 70% – at costs that are often double the original estimates.
This provides a fundamental validation for why uranium mining companies have been hammered to the tune of a +50% correction (using the URA ETF as a reference) since the end of January.
The cost of mining has increased dramatically, yet the uranium prices miners are realizing in their financial results haven’t kept pace. Okay, that makes sense on the surface. But I’ll contend that investors bailing on uranium miners at current levels are missing the forest for the trees, or have condensed their time horizons to such a degree that they can't see past their own noses. The following chart plots the spot price of uranium (black line) and the long-term contracting price (red line). Both are at cycle highs, with long-term contracting prices sitting at all-time industry highs. Moreover, both have increased by double digits in 2026, yet uranium mining equities are immersed in their second-worst drawdown since I started following the industry back in 2018. The math just doesn’t add up to me, and that spells a massive long-term opportunity for those with the fortitude and conviction to act upon it.
The collision course is set. Global investment banks now forecast deep supply deficits for each of the next 10 years, growing to over 45 million pounds (18% of global demand) by 2035. Yet, astonishingly, utilities have only signed supply contracts for about 55% of their total reactor fuel requirements since 2015.
Enter Big Tech. The nuclear renaissance is now being fully funded by hyperscalers desperate for reliable baseload power. Over the past four weeks alone, we saw nine major nuclear deals. Google partnered with Constellation and Southern; Amazon deployed $3 billion with Constellation and funded X-energy SMRs; and the DOE offered Vistra a $4.2 billion loan. Globally, the stigma is fading rapidly, with Italy, Switzerland, and Belgium taking active steps to reverse decades-old anti-nuclear policies.
The immediate trade here is the "uprate." Greenfield nuclear remains brutally slow and expensive, but squeezing 5-20% more power out of a reactor that is already licensed, staffed, and on the grid is highly efficient. Yet, despite these massively bullish fundamentals, the physical spot market remains in a standoff. Spot volumes in September fell to their lowest level since December 2009.
Fuel buyers are holding out for lower prices that have a snowball's chance in hell of materializing. As Alkin warned them: the cost of securing supply early is visible, but the cost of discovering a shortage too late may be catastrophic. If fuel buyers refuse to contract at the prices required to incentivize new mines, those mines simply won't be built. The ultimate risk to utilities today isn't paying a higher price for uranium – it is the very real threat that the fuel they desperately need to power this AI infrastructure supercycle will simply stay in the ground.
Let’s round out this week with some closing thoughts. I just spent the past week on the road visiting clients on the West Coast. There was a palpable sense of unease on the geopolitical front and frustration with the strategic execution out of Washington, D.C. Yet, ironically, clients were largely at ease with the markets, even while acknowledging the lack of upward progress over the last several months. My message throughout these meetings was simple: this is a time to remain measured and balanced in one's view of the markets.
We have seen a meaningful repricing across many pockets of the capital markets throughout 2026 – a process that has me licking my wounds in some regards, and grateful in others. Nevertheless, I find myself holding a view not dissimilar to where I've been for several months. I am still leaning bullish, albeit with slightly less conviction. I see the ongoing repricing as a constructive evolution within the largest investment cycle anyone alive today has ever witnessed. Yet, I remain acutely aware of the downside risk: what if this isn't just indigestion, but rather the start of a much uglier unwind in risk assets?
On certain days, considering everything thrown at the S&P recently – energy spikes, sovereign stress, severe factor de-risking – I am impressed by the market's resilience. On other days, as we accumulate these macro pressures, I worry the market has grown overly complacent. After months of ping-ponging around the same levels and chewing up popular positions, I have no issue swimming a bit closer to shore right now. Usually, when risk/reward feels this murky, the odds eventually favor the bull. This is Wall of Worry 101.
The earnings backdrop remains technically supportive, with consensus S&P EPS growth estimates at +29% in 2026, +15% in 2027, and +11% in 2028. If the broad market delivers +15% EPS growth in 2027, the primary trend for stocks will remain higher, even if the gradient flattens. Bear markets rarely materialize in the presence of double-digit earnings growth. Not to mention that valuations have rerated considerably lower throughout 2026 (see chart below) with the S&P trading at its lowest forward P/E (18.7x) since the liberation day lows back in April 2025.
However, the bricks in this particular wall of worry – a $50+ billion quarterly capex bill demanding bond market absorption, credit spreads quietly gapping, and real yields shifting – are the kind that tend to get heavier, not lighter.
It is mentally exhausting to manage a portfolio where your perfectly calibrated thesis gets undercut by cross-asset collateral damage on any given day. But this indigestion is a necessary cleansing of overly optimistic expectations. We are reaching the limits of unfettered fiscal excess and an unprecedented AI cycle with an insatiable appetite for capital. The best path forward is to keep an open mind, respect the chop, and avoid getting overextended on the risk spectrum so we don't squander the compounded returns built over the last decade.
The articles and opinions in "Capital Market Musings and Commentary" are for general information only, and not intended to provide specific investment advice. Performance, dividends and other figures have been obtained from sources believed reliable but have not been audited and cannot be guaranteed. Past performance does not ensure future results. Investing inherently contains risk including loss of principle. Advisory services offered through Casilio Leitch Investments, an SEC registered investment advisor. Copyright © 2023 Casilio Leitch Investments. All Rights Reserved.