Sharing another macro essay I wrote titled “A Cycle Built on Borrowed Time”.
It covers the stress building in US liquidity markets, the AI capex boom, the misunderstood debt cycle, India’s long-awaited consumption turn, and the growing constraints around energy and infrastructure.
A quick heads-up: this is a long one. It’s the most thorough memo I’ve written so far, and I think it’s also my best. It’s worth reading slowly.
What changed this month
Over the last few weeks a few quiet indicators have started to move together in a way that is hard to ignore. None of them, on their own, is dramatic. Taken together, they say more about where we are in the cycle than another all time high in the index.
On the plumbing side, the secured overnight funding markets have started to strain. As Luke Gromen has pointed out, repo rates have been ticking higher at the same time that the US Treasury has been rebuilding its General Account from $300bn to close to $1tn. That cash has to come from somewhere. In practice it means more bills, more frequent refinancing, and more demand for short term funding in a system that has struggled to term out its debt.
At the same time, rate expectations have swung sharply. Ritesh Jain notes that the probability of a December rate cut went from almost 90% at the end of October to roughly 30% a few weeks later. The Federal Reserve itself is split down the middle between those who want to cut and those who do not. It is rare to see such a divide inside the committee. Markets have spent most of the last month repricing that shift in expectations rather than responding to any single data point.
Beneath the headline indices the real economy is losing momentum. Job postings on Indeed are back to levels last seen in early 2021. Challenger’s layoff data shows job cuts in October running at almost three times September’s pace. S&P Global counts the highest number of large corporate bankruptcies in about fifteen years. Visitor traffic to Las Vegas is falling at a rate last seen during the global financial crisis. All of these are small straws, but they point in the same direction.
The policy response is already forming. The Kobeissi Letter has been tracking a new wave of fiscal and monetary support: proposed stimulus cheques in the United States, fresh packages from Japan and China, Canada restarting quantitative easing, the Federal Reserve winding down quantitative tightening, and more than 300 rate cuts globally over the last two years. Global broad money is at a record high. In other words, just as labour and demand begin to soften, another round of stimulus is lining up.
This sits on top of an artificial intelligence capex boom that is still doing much of the heavy lifting for headline growth. The same data centre spend that props up reported GDP also raises questions about power, water, and the durability of the assets being built. I will come back to that later in the memo, because it matters for both credit and equity markets.
The U.S. funding system is starting to creak
The clearest stress this month has come from repo market. As Luke Gromen has highlighted, secured overnight funding rates have begun to rise in a way we haven’t seen for years. It may look like a small move on the surface, but repo is the foundation of the entire US funding architecture. When that market tightens, it usually means something upstream is forcing the system to work harder than it wants to.
As I wrote earlier, part of the strain comes from the Treasury’s rebuilding of its General Account. The balance has climbed from roughly three hundred billion dollars to almost one trillion in only a few months. To refill that account the Treasury has had to issue an enormous volume of short-dated bills. More bills mean more refinancing. More refinancing means more reliance on repo. And more reliance on repo raises the clearing rate for every leveraged participant in the system.
That brings us to the marginal buyer of long-end Treasuries. It isn’t China or Japan anymore. According to the Federal Reserve’s own data, the biggest incremental buyer has been a cluster of Cayman-domiciled hedge funds running highly levered basis trades. They own close to $1.8tn of Treasuries, financed largely in repo at leverage ratios that can run 50-100x. When repo rates rise, the funding leg of that trade gets squeezed. If that squeeze continues, they may have to shrink their positions. Shrinking those positions in an illiquid long-end market pushes yields higher, weakens equities, and feeds back into more deleveraging. This is the cycle Gromen has been warning about.
There is also a broader macro consequence. Consumption makes up roughly 2/3 of US GDP, and a meaningful part of consumption growth now depends on asset prices rising. Net capital gains and taxable distributions alone are almost twice the annual growth of personal consumption expenditures. In plain terms, the household sector cannot keep spending if equity markets decline.
AI: Productivity boom and credit risk
Artificial intelligence is often described as the new general-purpose technology, something that can lift productivity across the economy the way electricity or the internet once did. That may be true, but as Luke Gromen keeps reminding, productivity alone doesn’t tell you how the credit system absorbs that shock. The United States is built on a model where employment, wages, and debt service reinforce each other. When technology weakens the wage component, the rest of the structure feels it.
The closest historical parallel is not another tech cycle but what China’s entry into the WTO did to the American industrial belt. A massive wave of cheaper supply, far more efficient production, and fewer constraints on labour and regulation. Unemployment in certain cohorts fell by a third in the years that followed because the entire employment base shifted. The early signs of something similar are already visible in the US. Unemployment among bachelor-degree holders in the 20-24 age group is around 7%. If a comparable shock hits the 25-45 age band, which anchors mortgages, car loans, and most consumer credit, the entire lending system comes under pressure.
The capex cycle itself carries echoes of past bubbles. In the telecom boom of the late 90s, companies borrowed heavily to lay fibre long before they had the revenues to justify it. The shale boom a decade later followed the same pattern: rapid capacity buildout funded by cheap debt, declining well productivity, and poor recovery values on the underlying assets. AI risks the same dynamic but with one important difference. Data centre chips have a useful life of three to four years. By the time the next generation arrives, the current generation is close to obsolete. That might question the recovery value of the capex that is being built today.
The bottleneck is no longer the chip itself but the infrastructure that supports it. Electricity, water, and specialized labour are all becoming constraints. Several data centre projects on the US West Coast have been pushed out toward the end of the decade because utilities cannot guarantee the power. Bloomberg recently wrote about NVIDIA-linked facilities sitting unused because they cannot get the required hookups. Water-cooled chips sound elegant until you have to source and transport the volume of water these clusters require.
For now, much of the AI spend has been funded through retained earnings and equity. That is changing. Companies have begun issuing debt to keep pace with the capex cycle. If the returns on that investment arrive slower than expected, the mismatch between short-lived assets and longer-dated liabilities becomes important.
AI may well raise long term productivity, but in the short run it widens the gap between output and income. That gap sits at the heart of the US credit system.
The U.S. labour market is weakening
The softening in the labour market has been gradual rather than dramatic. The headline unemployment rate still looks stable, yet the underlying indicators tell a different story. Steve Hanke notes that job postings on Indeed have fallen more than 6% YoY, returning to levels last seen in early 2021. Openings have been slipping for months, and the gap between available jobs and job seekers is closing quickly.
Layoff data reinforces this. According to Challenger, Gray and Christmas, US companies cut more than 150,000 jobs in October, nearly triple the number from September. They span technology, logistics, retail, and even sectors that benefitted from the post-pandemic recovery.
Bankruptcy trends echo the same theme. S&P Global reports that more large US companies have gone bankrupt this year than at any point in the last 15 years. The failures are not limited to over-levered businesses. They include firms facing weaker demand, higher financing costs, and declining pricing power.
The stress is now visible in consumer behavior. Las Vegas visitor traffic, often a good proxy for discretionary spending, has fallen sharply, matching rates seen during the global financial crisis. Surveys show household sentiment deteriorating, especially among younger graduates and middle-income households.
None of these indicators on their own mark a turning point. But taken together, they describe a labour market losing breadth and confidence just as stimulus discussions reappear. It is the combination that matters: softer demand, rising layoffs, and a policy environment shifting back toward liquidity support.
Why “bad news for consumers” is becoming “good news for markets”
One of the stranger dynamics in this cycle is the widening gap between the economy that households experience and the one financial markets are pricing. The Kobeissi Letter captured it well: even as the S&P 500 hits new highs and the largest technology companies exceed $20tn in market capitalization, a majority of Americans believe they are in a recession. Young graduate unemployment is nearing 10%, and real disposable incomes remain under pressure.
This divergence matters because it shapes policy. When consumer sentiment deteriorates and labour markets weaken, governments respond with stimulus regardless of whether asset prices are already elevated. That pattern is now global. The United States is preparing direct transfers, Japan has announced a $100bn package, China has approved more than a trillion dollars in fiscal support, and Canada is restarting quantitative easing. Central banks worldwide have cut rates more than 300 times in the last two years, and global money supply has reached a record $137tn.
The irony is that the sectors driving equity indices (large-cap technology, AI infrastructure, and capital-light digital businesses) do not need rate cuts or stimulus. But everyone else does. And because markets are now so heavily weighted toward companies that benefit from liquidity rather than broad economic strength, stimulus meant for households ends up amplifying asset prices instead.
The result is a redistribution effect: nominal asset values rise while real consumer conditions lag. Asset owners gain, wage earners tread water, and the distance between the two widens. It is the logical outcome of a system where financial easing is the default response to economic strain. Markets read weak consumer data not as a warning but as a signal that more liquidity is coming.
This is the uncomfortable symmetry of the current regime. The worse conditions look for the median household, the more supportive the environment becomes for financial assets.
DSP’s contra view: The debt story is not what people think
This entire section is credited to Sahil Kapoor from DSP.
Every cycle produces at least one perspective that sharply diverges from consensus. Sahil Kapoor from DSP provides that counterweight this time. While most commentators frame the United States as drowning in debt and heading toward inevitable currency debasement, he argues that the real picture is more nuanced and, in some ways, misdiagnosed.
The hidden truth about U.S. debt
The headline number, federal debt at nearly $38tn, is alarming. But when DSP decomposes the system into its three borrowers (households, non-financial corporates, and the federal government), a different pattern emerges. After the 2008 crisis, households deleveraged aggressively, and corporate borrowing grew at a manageable pace. The only balance sheet that truly blew out was the federal one.
Total non-financial debt as a share of GDP was about 250% in 2009. After rising sharply during COVID, that ratio has fallen back to roughly 246% today, almost the same level as fifteen years ago. In other words, the system is not uniformly over-levered. What has deteriorated is Washington’s balance sheet, not the private sector’s. That distinction is often lost in the broader narrative.
Why foreign central banks really stopped buying treasuries
The popular explanation is that foreign central banks “lost faith” in U.S. Treasuries and switched into gold. Sahil challenges this. If preferences had truly shifted years ago, gold should have risen sharply starting in 2014. It didn’t. For nearly a decade, gold was flat to down.
The more compelling explanation is that EM central banks did not stop buying Treasuries because they disliked them, but because they stopped earning the dollars needed to buy them. After the 2013–14 U.S. shale boom, America dramatically reduced crude oil imports. Countries that once ran large surpluses against the U.S. suddenly saw those flows evaporate. Without dollar inflows, they could not accumulate Treasuries even if they wanted to.
Reframing the USD bear narrative
The U.S. fiscal position is undoubtedly stretched, and long-term sustainability is a legitimate concern. But the private sector is far healthier than the headline numbers imply. Household leverage is contained, corporate balance sheets are stable, and the overall debt-to-GDP ratio has not deteriorated materially relative to the past.
As for the idea that the world is “moving away from the dollar,” DSP argues that this, too, is incomplete. A dollar shortage still exists across large parts of the emerging world. Without sustained surpluses against the U.S., they cannot rebuild Treasury holdings even if they wished. That structural constraint keeps the dollar stronger for longer than the bearish narrative often suggests.
DSP doesn’t dismiss long-term risks. But their work is a reminder that monetary transitions are rarely linear.
India’s consumption slowdown and the turn
The story of India over the past two years is largely the story of its middle class. Consumption accounts for roughly 60% of the economy, and when the middle-income segment slows, the broader economy inevitably follows. That slowdown became visible after the initial post-pandemic rebound faded.
Why India Slowed
A combination of policy and labour-market dynamics created the drag. To finance one of the largest public-capex cycles in recent history, the government increased income-tax collections and GST revenues. Middle-class households absorbed the bulk of that adjustment. Higher taxes could have been offset by strong job creation, but by 2022 the pace of hiring began to weaken. Wage growth in the top 50 listed companies averaged just 3% over three years, while inflation held near 6%, pushing real wages negative. At the same time, household leverage rose sharply. Excluding home loans, households now carry debt equal to roughly 1/3rd of their annual income, among the highest ratios globally. When taxes rise, wages stagnate, and leverage climbs, consumption inevitably slows.
The 6.3 Trillion Rupee Stimulus Reversal
The good news is that policy has pivoted decisively. Since the start of the year, a coordinated easing across taxes, regulation, and monetary policy has redirected an estimated 6.4tn rupees back into household cash flows. Income-tax cuts delivered around 1tn. A suite of GST reductions added another 2tn. The ban on real-money gaming redirected close to 700bn. Expected curbs on F&O trading could release nearly 1tn more into the real economy. The RBI contributed with four rate cuts amounting to a full percentage point, injecting an additional 1.6tn rupees through lower borrowing costs. Together, this is one of the most significant pro-consumption adjustments in recent years. The effects should begin to surface meaningfully from December onward.
Household Balance-Sheet Risks
The challenge is that Indian households enter this recovery from a weak starting point. According to Marcellus’ survey data, 14% have no emergency savings, and roughly half have buffers equal to only 20% of their income. Household savings as a share of GDP are near a fifty-year low. These metrics do not prevent a consumption rebound, but they make it more sensitive to employment trends and policy support. The new stimulus will help, but repairing balance sheets will take time.
The direction of travel is now improving. Policy has shifted toward reflation, liquidity conditions are easing, and early indicators show stabilization in household demand. Whether this translates into a sustained earnings recovery will depend on how quickly consumption resets after two years of pressure.
Energy, infrastructure, and power as the new constraint
One of the clearest signals that the global economic model is shifting comes from electricity. For years, cheap power and abundant grid capacity were taken for granted in developed markets. That assumption is now breaking down. Ritesh Jain’s recent observations make the scale of the issue hard to ignore.
In Oregon, a Berkshire Hathaway–controlled utility that contracted power to Amazon in 2020–21 can no longer deliver the electricity it promised. Similar strains are emerging across the United States. Several NVIDIA-linked data centre projects in California are reportedly sitting dark because the grid cannot support them. Developers in other states are pushing completion timelines out to 2030 simply because they do not expect to receive timely grid connections. The bottleneck is not chips, or capital, or software talent. It is power.
This constraint is becoming visible in household economics as well. Over the past fifteen years, US electricity prices have quadrupled. For households in the bottom income quartile, electricity now consumes close to 30% of take-home pay. In Virginia, political campaigns centred on reducing power bills were decisive in recent elections. Energy affordability has quietly become a frontline economic issue in a country that once treated electricity as an afterthought.
The link to artificial intelligence is direct. AI is an energy-intensive technology. Every incremental wave of compute requires disproportionately more power, cooling, and infrastructure. Unlike the telecom and shale-capex cycles, which at least created long-lived assets, the useful life of AI hardware is short. Chips turn over every few years, but the electricity and water required to operate them are continuous constraints. As data-centre construction accelerates, the grid is becoming the limiting factor on growth.
In macro terms, this is the new scarcity. For four decades, the binding constraint was capital. Today, it is energy infrastructure. The fiscal and geopolitical implications are significant. Countries with surplus electricity, stable grids, and reliable generation will attract the next wave of industrial and digital investment. Regions with fragile grids will face higher costs, delayed projects, and political pressure to subsidize power.
Way forward: Position, don’t predict
The themes across this memo point in one direction: the system is moving toward a world where collateral, energy, and nominal liquidity matter more than forecasts. The task is not to predict turning points but to position portfolios around the constraints that are already visible.
The first anchor is real collateral. Gold, silver, and other hard assets remain the clearest safeguards in a regime where monetary claims expand faster than the income required to support them. They are the insurance against fiscal and geopolitical fragility. In the same vein, equities backed by tangible assets, self-financing models, and low dependence on external leverage offer better durability than capital-light businesses reliant on perpetual liquidity.
The second anchor is power and energy infrastructure. Electrification is becoming the defining bottleneck of this cycle. Countries and companies with stable grids, surplus generation, and efficient transmission networks will capture outsized investment flows.
A related opportunity lies in the AI supply chain, but with a specific lens. The bottleneck is not chips alone; it is electricity, cooling, water, and grid access. Businesses positioned upstream of the compute cycle (power, engineering services, specialized infrastructure, and efficiency technologies) will likely see more durable demand than the end-users of the chips themselves.
India offers its own distinct path. A revival in consumption, supported by the 6.3tn rupee policy reversal, can unlock earnings growth across credit, housing, staples, and discretionary categories. The opportunity is not broad-based yet, but the direction of policy suggests that domestic demand will strengthen over the coming quarters.
Across regions, the likelihood of nominal asset inflation remains high. Global liquidity is expanding, fiscal stimulus is accelerating, and central banks have already demonstrated a willingness to ease at the first sign of labour-market weakness. In such an environment, asset prices may continue rising even as real economic strain persists.
The path ahead does not require perfect foresight. It requires alignment with the structural forces now shaping the cycle. Choose hard collateral over credit, energy capacity over narratives, and domestic demand over speculative liquidity. Positioning, not predicting, is my central discipline.
I would like to credit most of this work to my teachers Ritesh Jain, Luke Gromen, Steve Hanke, Ishmohit Arora, Saurabh Mukherjea, Sahil Kapoor, and The Kobeissi Letter. I wrote this piece simply as a means of consolidating all their knowledge to paint a digestible picture for you all.
