My bet is we bottom from here. NFA.
Recently gold has been suffering a prolonged period of drawdown.
The reason given by the common narrative for the sell-off since late August was basically that oil prices are too high and rising, therefore pushing inflation and inflation expectation higher, resulting in a hawkish Fed that is poised to hike rates and in fact did indeed hike rates.
From August to the end of September, this all kind of track, with 10 year treasury yield running higher by 68bps, aggressively pricing in rate hikes. However, since October, something rather odd has happened. The yield dropped from the peak of about 5.36% on 10/6, to just below 5% as of today 10/23. Meanwhile gold dropped about 4% from around 4200 to currently right around 4000 critical level. The most glaring was yesterday 10/22 when yields were down about 15bps across the curve while gold suffered a large 2.5% sell-off. The correlation between gold and yields has flipped, suggesting that something else is at play other than the inflation and Fed rate hike scares.
Today on 10/23 we saw gold continue to slip, going under the 4000 mark and tried for the entire day to get back above it but failed and closed just under that critical level.
In the past, similar episodes happened as well, most recently back in 2023 during the SVB bank mini crisis and 2020 Covid pandemic. Both of these incidents saw a rather significant liquidity crunch. This time around we are seeing a similar episode in the equity land, in the form of an AI related sell-off, induced exactly by the credit scare of AI debt, which I’m not going into details of since it’s not really the topic of this particular one I’m writing.
This also came off the heels of gold being very speculative over the past year especially in the latter half of 2025 and in the first month of 2026, where gold along with silver saw significant price appreciation over a short period of time.
There has also been news of a nuclear fusion startup company being able to convert mercury into stable gold, which could have served as a trigger for a steeper drawdown over the past few days. But it is highly unlikely for gold to be mass produced from mercury, first because it is highly costly, and second because the resulting gold will be radioactive for more than a decade.
So to me it appears that we have shifted from gold being down due to high oil prices and the anticipation of Fed rate hikes in response to high inflation, to a more positioning driven liquidation phase, which could mark an end of the recent weakness, at least in the short to intermediate term.
And then more importantly, it is the question of the Fed rate hike potential. How likely is it really for the Fed to keep hiking rates?
The Fed has a dual mandate of promoting a healthy labor market and maintaining stable prices. Since August, Warsh and his colleagues have been more vocal on the price stability side, and delivered a rate hike in September. There has been little growth concern with AI investment and capex going strong and no visible labor market distress, at least not in elevated layoffs.
But this dynamic has been shifting over the past 2 weeks, especially in the form of growth scare or AI financing distress. Eventually, investors and creditors that lent money to the neoclouds to fund the AI data center buildout will need to see prospects of higher returns, at least within reasonable time.
With just one hike having been actually delivered and 2-3 more hikes anticipated, we are already seeing distress as investors demand higher returns, pushing neoclouds and hyperscalers bond yields higher and CDS higher, which sets up for the potential of a vicious cycle of higher financing costs leading to worsening credit rating, pushing financing costs even higher, eventually to a point of no return. If the Fed actually keeps up with the hiking cycle as people feared, we might see more distress come out, unless data center buildout proceeds normally and starts to generate required returns soon.
There is another aspect to this as well, namely the labor market. So far we’ve not seen mass layoffs and it has been a few years of low hire low fire regime. But for the AI monetization to succeed, as defined by generating the required return on the massive capex currently spent and in the pipeline, it would seem to me that there indeed must be some layoffs in the labor market.
My key reasoning is that currently what LLM can help us do is almost exclusively in the form of boosting productivity, meaning that it enables us to do more within the same amount of time. But it is not really something that enables us to do things that no amount of human labor is able to do. Trains and railroads and cars enabled us to travel at speeds no amount of pure human labor is capable of, and without planes you can’t just add more labor to make humans fly.
The exception perhaps is with respect to math and programming frontiers, where humans have a limit in the subject fields they can be experts in, while LLMs can be subject level experts in almost any field with a short amount of time. But most white collar work is not cutting edge research, but simple administrative work that requires not a lot of intelligence, which means that for companies it is a choice between LLM and human labor, or more likely a mix between the two but you’d certainly have LLM replacing labor.
So outside of the limited cutting edge research space, there is naturally a conflict between the labor market and AI adoption and monetization. If you want monetization to succeed, you’d have less labor for sure, but even then for AI capex to succeed at a lower labor anchored price, the unit cost must be low enough, so that you’d need mass replacement for monetization to truly work out. Basically, you either have mass layoff for ordinary white collar work, or a failure of AI capex monetization, or worse, both.
All of this is to say that, to me, I consider the probability of Fed hiking rates persistently to be low, because in the short term we are already seeing cracks in the AI capex financing, and in the longer term there will be a natural conflict between AI capex monetization and labor market stability.
So if the Fed is unlikely to persist in the rate hike cycle, inflation is then another input to the price of gold. If inflation remains well-anchored, and the Fed naturally stops hiking rates, gold will see just a normal appreciation, probably matching the inflation rate.
But with oil continuing the rise and for potentially more knock-on effects to come as higher oil and higher refined products price to ripple through to other products and services (for example grocery and air fares), and that the SPR has been depleted to 1982 levels (so we have less buffer), it is hard to imagine inflation actually abating. Not to mention that the Iran situation has not seen any real sign of peaceful resolution yet- in fact there has been news that both sides are preparing for a return to combat, it could very well get much worse.
The Fed is really stuck between a rock and a hard place. My bet is that the Fed will do whatever it takes to prevent a widespread credit event that brings down the entire AI capex and circular financing, even at the expense of inflation. And that is my case for gold.
10/25/2026
