CONSUMER WATCH
Last week we got Q1 results from several retailers. They were generally good, particularly at the likes of TJX and Ross Stores where squeezed Americans look for lower priced goods.
I was particularly interested In Walmart’s results, nearly $600B in US sales last year, clearly a value conscious retailer catering to most Americans.
The left chart shows the spectacular growth in US annual sales. The right chart plots quarterly sales growth rates which averaged +4.5% YoY in last year (when total US retail sales rose 3.8%) but slowed to +4.1% in the most recent quarter ended in April (when total US retail sales rose 4.5%).
WMT US sales are slowing even while total US retail sales are accelerating. Maybe there are idiosyncrasies to explain.
But WMT management gave us other interesting numbers (this quarter vs same quarter last year):
- Comp sales ex fuel: +4.1% vs +4.5%
- Transactions ex fuel: +3.0% vs +1.6%
- Average ticket ex fuel: +1.1% vs +2.8%
More transactions but average transaction well below inflation. But management says it recorded its highest market share gain in 5 years in general merchandise.
It also said that like-for-like inflation was +4.4% in Q1 (February to April), ex fuel. Last year, when the average ticket rose 2.8%, management said inflation was “effectively flat or slightly deflationary”.
Notice the change?
Notice that volume is down quite a bit?
Total US retail sales were up 4.5% YoY during the same WMT quarter.
Hmmm… Resilient consumer?
BTW, Indeed Job Postings keep weakening (through May 15) …
… even though unemployment claims keep declining. Strange.
S&P Global’s Flash PMI for May said that
Service sector growth remained especially sluggish and is on course for its weakest calendar quarter since late 2023 as new business inflows rose only modestly. Service providers reported subdued demand reflected rising prices and uncertainty, notably among consumer-facing businesses and for exports.
Measured overall, employment fell in May for the second time in the past three months, the rate of job losses reaching the highest since August 2024 due to growing concerns over rising costs and deteriorating demand conditions. (…) service sector jobs were reduced at the second-fastest pace seen since May 2020.
Average prices charged for goods and services rose in April at the fastest rate since August 2022 amid the growing supply scarcities and jump in costs. Goods prices showed a particularly marked rise, the rate of increase hitting the
highest since September 2022, but service sector selling price inflation also accelerated to a ten-month high and was one of the sharpest rates seen over the past four years.
Strange, and scary.
While on strange things:
The Oil Mystery at the Heart of America’s Pressure Campaign on Iran The U.S. government, oil traders and private analysts are divided over how much time Tehran has before it runs out of places to stash its crude
The answer may help determine the outcome of the conflict.
A five-week-old U.S. naval blockade has trapped much of Iran’s oil inside the Persian Gulf, forcing Tehran to pump stranded barrels into rapidly filling storage tanks and aboard a flotilla of ships nearby.
U.S. officials are betting that when Iran exhausts places to stash the oil, it will face a costly, high-risk shutdown of its oil fields, forcing Tehran to blink in negotiations over its nuclear program and the wider conflict.
The U.S. government, oil traders and private analysts, however, are divided over exactly how much time Tehran has until it reaches “tank tops,” industry parlance for running out of storage.
Estimates of Iran’s onshore capacity range widely from 57% to 90% full, meaning Tehran could hit the wall in days—or hold out for weeks. Slowly throttling wells and using idle ships as floating storage are further helping Iran stretch the clock. (…)
At the end of April, President Trump said that Iran had about three days before its oil pipelines risked explosion from being clogged up. Last week, Treasury Secretary Scott Bessent said that Iran’s oil storage had filled up and Tehran would need to shut down oil production.
Analysts say Tehran has more breathing room to store its unsold oil. But there’s little agreement on how much.
S&P Global Energy estimates onshore crude stocks are around 57% full, below the historical range for this time of year. JPMorgan puts the number at 64%, leaving the equivalent of about three weeks of exports before they are full.
On the other end of the spectrum, data provider Kpler estimates onshore tanks to be 90% full, meaning they would be exhausted by the end of the week if the blockade continues. (…)
Shipping analytics company TankerTrackers said on Sunday that three empty tankers slipped through the blockade line in recent days into the Gulf. Together they could handle 1.9 million barrels. (…)
For more than three years after Trump imposed sanctions on Tehran in 2019, Iranian oil exports plummeted. Production also suffered, dropping below 2 million barrels a day. However, output surged back in 2022 and reached multiyear highs earlier this year.
That track record points to limited lasting impairment from prolonged disruptions, said Kim Fustier, an oil and gas analyst at HSBC, suggesting Tehran can afford to play the long game.
“Expectations may need to be adjusted,” Fustier said. “It’s hard to imagine that a weekslong blockade is going to achieve something different than 3.5 years of little exports did.”
Fertiliser groups cut production as Iran war squeezes sulphur supplies
Fertiliser companies around the world are cutting production of one of the most vital crop nutrients as the conflict in the Middle East upends supply chains, intensifying concerns over future food shortages.
Disruption to shipping in the Strait of Hormuz has hit global supplies of sulphur — required to make phosphate fertilisers that are used on crops including corn, soyabeans, rice and palm oil.
“This situation around Hormuz was in the beginning a raw material problem that has turned into a fertiliser supply shock,” said Faris Derrij, chief executive of OCP Nutricrops, whose parent OCP Group is the world’s biggest phosphate exporter.
Before the Iran war began in late February, about 50 per cent of the world’s sulphur trade passed through the strait.
Phosphorus is one of the three main macronutrients needed by commercial crops to survive, along with nitrogen and potassium. The phosphate market was already constrained before the war because of rising demand for sulphur in other industries such as battery metals processing.
“The phosphate supply situation is grim,” said Chris Lawson, vice-president of market intelligence and pricing at consultancy CRU. “Every major source of phosphate supply is under pressure simultaneously.” (…)
In response to tightening supplies, China has halted exports of key phosphate fertilisers until at least August. (…)
“There is simply no sulphur,” said Christian Wendel, president of fertiliser trader Hexagon Group.
Sulphur prices, which traded at between $150 and $180 a tonne a year ago, have surged to as much as $850-$900 a tonne, with some delivered prices approaching $1,000, he said. Even if they can secure sulphur, the economics for phosphate producers “just don’t work out when you have very expensive sulphur coming in”, said Willis Thomas, head of fertilisers at CRU. (…)
But analysts warn that even if the strait were to reopen to more vessels tomorrow, phosphate markets could take longer to recover because they are more geographically concentrated and heavily dependent on Gulf sulphur flows. Morocco and the western Sahara hold most of the world’s reserves of phosphate rocks, which are treated with sulphuric acid to make fertiliser. (…)
“We believe . . . that already for next year we will see smaller crops,” he said. “I think we really might steer towards famine [in certain places] because there won’t be enough food or enough grains in some parts of the world.”
Just “a little excursion”!
And yet:
Investors Just Can’t Get Enough of Stocks These Days War, inflation and worries about the AI boom haven’t kept stocks from records
(…) Institutional investors are holding 50% more in stocks than their benchmarks, the highest figure since January 2022, according to Bank of America’s latest survey of global fund managers. They are particularly enthusiastic about stocks that would benefit from a reacceleration in economic growth, with holdings of so-called cyclical stocks exceeding those of defensive shares that investors buy for relative stability by the largest amount since 2018. (…)
Underpinning the good vibes: hopes for a peace deal in the Middle East conflict and an earnings season that has eased concerns that profits won’t keep up with stock gains. Companies in the S&P 500 have reported a jump of 28% in first-quarter profits, looking at both actual earnings and expected results from the roughly 5% of companies in the index that haven’t yet reported earnings, according to FactSet. (…)
BofA analysts wrote in a recent note that fund managers’ cash levels saw their biggest monthly drop since 2024, to a level they view as a sign to sell shares. (…)
- The Risk Premium for Holding Stocks Over Bonds Is Vanishing Gap between market’s earnings yield and bond yields has narrowed, a measure that has at times predicted subpar stock returns
Investors are piling into stocks at a feverish pace. But by one measure, equities look as unattractive as they did after the dot-com bubble burst.
That metric is the equity risk premium, often defined as the gap between the S&P 500’s earnings yield—the profit companies generate relative to stock valuations, expressed as a percentage—and that of the 10-year Treasury note. In recent weeks, it has nearly disappeared and is hovering among its lowest levels since the start of the new millennium.
In other words, a rough gauge of stocks’ expected returns is now only slightly higher than what ultrasafe government bonds will produce.
The primary culprit: a global bond rout powered by inflation fears that has pushed Treasury yields higher. The Iran conflict and the closure of the Strait of Hormuz have lifted the price of oil roughly 60% this year and drastically altered investors’ expectations for interest-rate cuts, once considered a given in 2026. (…)
The earnings yield for the S&P 500—based on projected earnings over the next year—has dropped in recent weeks as share prices jump and Wall Street revs up a weekslong stock-buying spree.
“There’s a little bit of a disconnect at the moment between the bond market and the equity market,” said Don Calcagni, the chief investment officer at Mercer Advisors. “This is telling us that inflation fears are growing, and valuations are stretched.” (…)
The difference between the S&P 500’s earnings yields and that of bonds shows how much investors are compensated for the additional risk of owning one over the other. The idea is that, in the long term, stocks need to promise a higher reward than bonds—if not, the safety and predictability of Treasurys would outweigh the risk of buying shares that can lose some or all of their value.
Some investors shrug off the measure, arguing it isn’t very useful to compare the fixed return of a bond with the potentially unlimited upside of a company’s earnings growth. They add that valuation metrics such as the equity risk premium, when considered alone, are rarely a good reason to dump equities or hold off on buying.
But in the past, measures of the major index’s earnings yield relative to bond yields have proved a useful guide, particularly over a longer time horizon.
The Nobel Prize-winning economist Robert Shiller, for example, has highlighted the tight relationship between his preferred risk-premium measure, which adjusts past earnings for inflation, and the index’s future excess return versus bonds in data going back more than a century.
A notable exception has occurred recently: The S&P 500’s annualized excess return over the past 10 years has been far above what would have been expected based on where Shiller’s “excess CAPE yield” stood a decade ago. The divergence underscores just how much stocks have been able to defy the normal laws of gravity in a period marked by a pandemic, inflation and a sharp rise in interest rates.
The equity risk premium dipped below zero early last year, when a similar combination of higher Treasury yields and soaring equity valuations nudged the metric into the red. The last time the measure stayed negative for an extended time period came in the aftermath of the dot-com bubble.
The debate among equity investors now is whether stocks’ furious race back to records is justified by surging profits at America’s largest companies. (…)
The opposing argument is that the AI boom is just beginning and that companies will see an even bigger boost to earnings as they sell, adopt and improve the technology—powering broader economic growth in the process.
“Stocks aren’t cheap, but they’re not horribly expensive,” said Jeff Blazek, multiasset co-chief investment officer at Neuberger, adding that his team expects a softer approach on interest rates from the Fed than the broader market has forecast. “We like bonds, but we also like stocks.” (…)
But those who are optimistic for equity returns in 2026 caution that their outlook is contingent on a resolution to tensions in the Middle East. Jeff Buchbinder, chief equity strategist at LPL Financial, said that he and his colleagues have started to refer to oil prices as the “chart of truth.”
“The price of oil tells us how the negotiations are going,” he said. “If we’re still looking at $100 oil in late summer, then the formula changes.” (…)
The always useful Ed Yardeni gets into this thorny debate: “Then again, exuberance may be turning a wee bit irrational. S&P 500 consensus expected long-term earnings growth (LTEG) soared to 21.9% during the week of May 21. That’s the highest reading on record, except during the pandemic period.”
While wearing these rosiest glasses, one should take time to ponder the AI boom:
How AI backlash could cost investors (Axios)
AI backlash is mounting. Executives are getting booed. Workers are threatening strikes, and protests are frustrating data center development — but that doesn’t seem to worry investors who are raining money on companies in the business.
AI hate could slow adoption of the technology, posing an underappreciated risk for investors buying into the current frenzy.
The latest: Even SpaceX’s prospectus — which is threaded with AI hype — warns that the backlash is a real threat:
“If AI technologies are perceived to be significantly disruptive to society, it could lead to governmental or regulatory restrictions or prohibitions on their use, societal concerns or unrest, or, both, any of which could materially and adversely affect our ability to develop, deploy or commercialize AI technologies and execute our business strategy.”
Community outcry against AI was a key topic in recent meetings that Morgan Stanley strategists had with investors recently in the U.S., they wrote in a note Monday. The two important areas of concern: job loss and electric bills.
- “These issues may increasingly become a part of the political landscape and could result in greater pushback to data center growth.”
- The growing opposition to data centers, along with some projects getting canceled, is “sapping confidence” among investors, per a client note from Jefferies, Axios’ Madison Mills recently reported. (…)
The American CEO of London-based Standard Chartered apologized today for his comment that the bank would replace “lower-value human capital” with artificial intelligence.
Communities around the country are bristling at the physical manifestation of AI — the sprawling data centers believed to raise local electricity costs and contribute little to local economies.
There’s also rising concern over AI misuse, misinformation and data privacy, per the Morgan Stanley note. (…)
The big picture: History is littered with examples of backlash against a technological or economic threat that ultimately proved futile. We see you, Luddites!
There’s AI resistance. Investors are noticing, but for now, the money machine chugs on.
Pope Leo urged governments to slow down the development of AI systems in his first major document, released on Monday, warning that they spread misinformation, prioritise conflict and risk leading the world down a path of unending war. (…)
The first U.S. pope called for ownership of AI data not to be left solely in private hands, for policy-makers to protect the rights of workers and keep children safe from the technology, and urged the cooling of competition between AI companies.
“What is needed is a more active political involvement that is capable of slowing things down when everything is accelerating,” said Leo in the text, entitled “Magnifica Humanitas” (Magnificent Humanity).
The pope called for “robust legal frameworks, independent oversight, informed users and a political system that does not abdicate its responsibility.”
Encyclicals are one of the highest forms of teaching from a pontiff to the Church’s 1.4 billion members. (…)
“Humanity is slipping into a violent culture of power, where peace no longer appears as a responsibility to be taken on, but as a fragile interval between conflicts,” he said. (…)
“We cannot rule out the possibility that some leaders may consider armed conflict as an effective way of diverting attention from domestic problems and a cynical tool for managing difficulties,” he stated. (…)
Everything seemed set for a photo op with tech and AI CEOs surrounding President Trump on Thursday as he signed a much-anticipated executive order on AI and cybersecurity.
But it fell apart hours before the order was to be signed, as a top Trump adviser and some tech executives gave it a big thumbs down. And the president didn’t really want to regulate AI in the first place.
Any further delay of the order means more time for infighting and for the text to get bogged down in disagreements among different parts of the government and industry.
Behind the scenes: Before the order was to be signed, Trump, AI adviser David Sacks and some in industry discussed it, sources familiar told Axios.
- The main reason the signing was delayed was that Trump “just hates regulation,” one source familiar said, adding that Sacks also “hated it.”
- “The whole thing was unnecessary” and “just something doomers wanted,” the source added.
Meta CEO Mark Zuckerberg, xAI CEO Elon Musk and Sacks all spoke with Trump between Wednesday night and Thursday morning.
“I didn’t like certain aspects of it. I postponed it,” Trump told reporters in the Oval Office on Thursday. “I think it gets in the way of — you know, we’re leading China, we’re leading everybody, and I didn’t want to do anything to get in the way of that lead,” he said. (…)
We shall see how all the above will impact compute supply growth but it better be not too much, per the Economist:
Strip out the splurge on information-processing equipment and software—the categories most closely tied to AI—and the picture looks grim.
Over the past four quarters non-residential fixed investment excluding ai-related categories has contracted at an annualised rate of roughly 3%, compared with average growth of more than 5% in the previous decade.
Investment in industrial and transport equipment has fallen by more than 2% over the past year. Manufacturing construction is down by 20%.
In total, non-AI investment is running about $130bn below its trend from the last decade. This capex recession is reducing gdp growth by roughly 0.4 percentage points.
Somebody got lucky…
BTW, Walmart had this AI info last week:
AI-powered Sparky’s strong customer adoption resulted in a fourfold sequential increase in units purchased, broadening its use from general merchandise to everyday essentials. Weekly active users of the Sparky AI shopping agent increased over 100% from the previous quarter, with average order value 35% higher for Sparky users; units purchased through Sparky grew more than 4x sequentially.
And from Intuit:
Amid all the noise surrounding Intuit’s big layoffs and stock plunge Wednesday, the CEO of the $100 billion firm also told investors in a call that the company will soon change the way it charges customers for upcoming AI features that link them to experts such as accountants.
Intuit customers can currently use its AI features as part of their flat-fee subscriptions, but chief executive Sasan Goodarzi said the company will switch to consumption-based prices for the new AI features set to be released in August. The company declined to specify details about the new features but said they will be embedded across its apps.
Numerous enterprise software firms have already made the switch to usage-based pricing, given the relatively high cost of using models from providers such as Anthropic to power their AI apps.
AI usage is exploding!
But compute supply is not. Just about every AI player is raising prices trying to control runaway demand.
Let’s see if AI can properly analyze the AI boom. Here’s Gemini on that (Gemini’s emphasis):
The tech industry is locked in a severe AI infrastructure and compute crunch, where an explosion in resource-heavy AI usage has officially outpaced the physical capacity of global data centers. This has forced major AI companies to actively ration access, restrict capabilities, and increase prices to manage the overwhelming demand.
Because agentic AI consumes massive amounts of compute, businesses adopting these tools are facing sudden budget shocks. For example, software engineering teams deploying AI coding assistants have regularly burned through their monthly allocations in days, racking up unexpected thousands of dollars per employee in API fees. Nvidia’s applied deep learning team even noted that their compute costs now exceed human employee salaries.
Data center construction simply cannot keep up with software adoption. The supply chain is constrained by several physical realities:
- The Memory Crisis: Modern AI processing is heavily dependent on memory bandwidth. High-demand AI chips have swallowed up global production capacity for DRAM and HBM modules, causing hardware component prices to spike.
- The Power Wall: Massively scaling data centers has run directly into localized power grid constraints, with utility grids unable to supply the gigawatts of electricity required to run and cool these facilities.
- GPU Rental Hikes: Due to shortages, the spot-market rental price for high-end GPUs (like Nvidia’s Blackwell generation) has surged by nearly 50% over brief multi-month windows.
To prevent total infrastructure collapse and maintain platform stability, leading providers are making severe operational cuts:
- Anthropic has faced repeated outages under the weight of Claude’s popularity, forcing them to tighten usage caps and enforce stricter token limits.
- OpenAI executives have admitted to making “tough trades,” including entirely pausing or shutting down heavily anticipated, compute-expensive applications (such as their Sora video generator) to save capacity for core text and reasoning models
The era of effectively “subsidized,” infinite AI compute is drawing to a close. For developers and businesses building on top of AI APIs, compute efficiency is the new competitive moat. Success is shifting away from who has the largest prompt, toward who can write the most efficient code, implement smart caching, and orchestrate smaller, localized models to avoid skyrocketing infrastructure bills.
The pressure is hitting other major hubs worldwide through unique bottlenecks:
China is experiencing a massive domestic demand surge. Daily average token calls jumped from 100 billion to over 140 trillion within roughly two years, a 1,400-fold increase.
Because domestic chips and older hardware are often less power-efficient, Chinese data centers require even more raw electricity and physical space to process the same volume of agentic AI workloads. This has triggered severe localized hardware shortages across Asia’s tech sectors.
Europe is desperate to close the data center gap with the United States, but it has run directly into an infrastructure bottleneck.
In the UK, proposed data centers are stuck in connection queues representing 30 gigawatts of power demand—equal to roughly two-thirds of the entire peak electricity demand of Great Britain. Grid operators simply do not have the transmission lines to move power to where the servers are.
Extreme Costs: European electricity prices for energy-intensive industries are roughly double those in the U.S. and 50% higher than in China. This is forcing European AI developers to face significantly higher operating costs, making the financial math of running compute-heavy agentic AI even more brutal.
With Western and European grids maxing out, major tech companies are aggressively shifting capital to Southeast Asia.
- Singapore’s Crisis: Singapore is the world’s most extreme example of the power crunch. Data centers now consume a staggering 19.5% of Singapore’s entire national electricity supply.
- The Spillover: Because Singapore has had to implement strict moratoriums and sustainability standards on new facilities, the AI buildout is spilling over into neighboring Malaysia and Indonesia. However, the speed of AI deployment is already outpacing those nations’ abilities to build out low-carbon energy infrastructure.
The physical components required to build AI servers are entirely globalized and completely maxed out. High-Bandwidth Memory (HBM)—the specialized, ultra-fast RAM required by AI chips—is entirely monopolized by three suppliers (SK Hynix, Samsung, and Micron). Because of the global agentic AI boom, global HBM supply is entirely sold out well into the future, creating a hard physical ceiling on how many AI servers can be built globally, regardless of a country’s wealth or power grid capacity.
From Silicon Valley to Beijing and London, the economic reality is identical: AI models have outgrown the physical infrastructure of the planet. Every major economic bloc is currently treating AI compute and data center energy not just as a tech issue, but as a matter of critical national infrastructure and economic sovereignty.
Is this AI fabulizing on AI problems? I can only say that every AI supplier I use is now rationing me and raised prices.
How serious?
What we are witnessing is not a fatal flaw, but a classic tech sector “infrastructure choke point.” Similar to the early days of the internet when dial-up networks crashed under the weight of early web traffic, the industry is experiencing an abrupt transition phase.
The market will naturally correct itself, and over the coming years, several structural shifts are likely to happen:
The current pace of hyper-spending is unsustainable. Nearly 30% to 50% of planned U.S. data center builds are facing delays or outright cancellations due to power grid limitations and component shortages.
Over the next 18 to 24 months, supply and demand will dramatically narrow, leading to an infrastructure market correction where underutilized data centers force cloud rental prices back down.
When compute becomes expensive, software engineering adapts out of financial necessity.
- Smaller, Hyper-Optimized Models: Instead of routing every basic query to a massive, 1-trillion-parameter frontier model, companies will aggressively transition to Small Language Models (SLMs). These highly compressed models can run locally on smartphones, laptops, or small internal servers, bypassing the data center bottleneck entirely.
- Algorithmic Breakthroughs: Software engineers are pioneering advanced quantization (shrinking model sizes without losing accuracy) and smart semantic caching (reusing previous AI calculations instead of recalculating from scratch), heavily slashing active token costs.
The worst of the hardware supply shock will peak as manufacturing expansions finally go live.
- Memory Shortage Relief: The current High-Bandwidth Memory (HBM) squeeze—where suppliers like SK Hynix, Micron, and Samsung are entirely sold out—is a temporary manufacturing lag.
- Next-Gen Tech: As production lines catch up and transition to next-generation architectures (like HBM4), the raw physical throughput of AI accelerators will dramatically expand, offering significantly more computing power per square inch.
AI companies will stop waiting around for slow, heavily regulated public utility grids to build new transmission lines. Tech hyperscalers are actively pivoting to sovereign power generation:
- Nuclear and SMRs: Tech giants are signing direct purchase agreements with nuclear power plants and investing heavily in Small Modular Reactors (SMRs) to power data centers completely off-grid.
- Geographic Redistribution: Instead of forcing data centers into crowded hubs like Silicon Valley, Northern Virginia, or London, companies will build infrastructure where stranded, unutilized energy already exists—such as near massive solar arrays in deserts or geothermal plants in Iceland.
We are currently living through the peak of the bottleneck, which explains the sudden price hikes and strict token usage caps. However, this crunch will inevitably trigger massive innovations in energy procurement, chip manufacturing, and software efficiency. By the late 2020s, compute supply will catch up, prices will stabilize, and agentic AI will become far cheaper and more sustainable to run.
So, “the AI boom is just beginning and companies will see an even bigger boost to earnings as they sell, adopt and improve the technology—powering broader economic growth in the process.” Probably right, but likely not a smooth ride, with significant bumps right ahead.
We all need shock absorbers.
Speaking of rides:
To infinity and beyond, with the SpaceX IPO Ordinary investors can at last bet directly on AI’s potential — with caveats
(…) The 200,000-word SpaceX prospectus — surely the most audacious ever published — sets out a vision of extraterrestrial data centres, asteroid mining and developing a “lunar economy”. Musk’s ambition is to harness the sun to “power a truth-seeking” AI. Many investors will find it hard to resist strapping in for lift-off.
But the document has caveats as big as the pay deal offered to its founder. The largest, also applicable to OpenAI, is that investment in AI hardware is burning through billions, racking up debt and pushing SpaceX into loss. Then there is the overhang of stock. Musk is subject to a longer than usual 12-month lock-up but other existing shareholders will be able to offload parts of their holdings quickly.
Retail investors’ fear of missing out could offset insiders’ hunger for rapid gains but the consequences for already febrile equity markets are uncertain.
Traditional governance checks are almost entirely absent. (…)
He [Musk] will have a virtually unchallengeable grip on voting rights and the board. Only three months ago, SpaceX absorbed another Musk venture, xAI, developer of chatbot Grok. The long-held goal of colonising Mars has for now been eclipsed by his desire to win the costly AI race. If Musk changes direction again and SpaceX turns out to be a trillion-dollar strategic bait-and-switch, there will be little shareholders can do.
(…) passive investors will end up owning slivers of SpaceX, OpenAI and Anthropic destined for public markets, whether they like it or not. Index providers have been vying to loosen their rules, making room for the AI developers.
This, then, is an epoch-defining test for market authorities, company directors, and the many promoters of these IPOs as they balance their own self-interest against the need to protect ordinary investors. (…)
Wait, wait! From Callum Thomas:
SPCX Index Fast-Track? with the S&P index committee currently consulting on how to treat Mega Cap (i.e. top 100 largest companies by market cap) IPOs, we could see SpaceX soon getting a substantial weight in passive and active fund benchmarks (e.g. on current info it could be as much as a 3% weight if it were included in the S&P500).
Based on Bloomberg Intelligence analysis, this would mean active+passive funds might need to buy almost half of the public float of SpaceX.
Source: @rduboff via @EricBalchunas
Stepping back and looking at the bigger picture, we basically have a wave of mega-cap IPOs expected to come to market. Aside from SpaceX there’s also talk of AI hot shots OpenAI, Anthropic, Databricks (combined set to be at least as big if not bigger than the SpaceX IPO). This all adds up to a bumper year for equity capital markets, and is likely to exceed the crazy stimulus-SPAC boom of 2021.
Source: Goldman Sachs via @BlakeMillardCFA
Zooming further out, and just looking at the number of IPOs, again we can see that things have heated up on the IPO front (and p.s. ever since 2000 IPOs kind of fell out of relative-popularity with the rise of venture capital and private equity, so the 2025-26 numbers are good by post-2000’s standards).
Source: Topdown Charts Professional
Back to the equity risk premium debate:
Three Forces Pushing Rates Higher Across the Curve
Front-end rates are under upward pressure because inflation is higher for longer.
The belly of the curve is seeing upward pressure on yields because of hyperscaler issuance.
And long-end rates are moving higher because of more Treasury supply and less Fed demand.
The bottom line is that three distinct forces are pushing rates higher across the curve, and investors should position for a persistently higher rate environment. (Apollo)
FYI:
In CBO’s projections, the federal budget deficit in fiscal year 2026 is $1.9 trillion and grows to $3.1 trillion by 2036. Relative to the size of the economy, the deficit is 5.8% of gross domestic product (GDP) in 2026 and grows to 6.7% in 2036, which is greater than the 3.8% deficits averaged over the past 50 years.
Rising net interest costs drive much of that increase. The primary deficit, which excludes those net interest costs, totals 2.6% of GDP this year and stays below that level through 2036, when it totals 2.1%. From 2026 to 2036, large and growing deficits cause debt to increase. Federal debt held by the public rises from 101% of GDP this year to 120 percent in 2036, surpassing its previous high of 106% of GDP in 1946.
As all this damage mounts, it’s starting to rattle the safest haven in world finance: the $50 trillion-plus market for Group of Seven sovereign bonds, where long-term yields hit a two-decade high this week. Put simply, investors are starting to worry, like they never quite did even at the post-Covid peak, that higher inflation is here to stay.
They want compensation for that risk and expect central banks will have to raise interest rates to contain it. That’s a conclusion based not just on the blow from the Mideast conflict but the environment it’s landing in. Economies still haven’t wrung all the pandemic inflation out of their system. Governments owe ever-increasing debts as they struggle to rein in spending that adds fuel to the fire.
I don’t know about you but
- AI is not the straight line many think it is
- Interest rates are rising everywhere
- Inflation may prove higher for longer
- The American consumer looks stretched, if not very stretched
- The “little excursion” ain’t over and not particularly fun…
- The Fed put and/or the Trump put currently seem unlikely
And most investors seem eager to ride Elon’s spaceship.
Meanwhile, Beijing is finally getting serious about China’s domestic economy:
China Dangles Better Welfare Access in Cities to Spur Consumers
China issued a plan to open up public services like schools and health care in cities, potentially expanding access for vulnerable migrant workers as part of a broader effort to boost living standards and consumer spending by millions of urban residents.
The State Council, China’s cabinet, published guidelines Friday instructing local governments to dismantle bureaucratic barriers for residents who lack household registration locally. The directive orders cities to expand public school access for migrant children and scrap local restrictions on employee social security.
Economists have long blamed the decades-old household registration system, or hukou, for suppressing domestic demand. Lacking an urban safety net program, China’s vast migrant population traditionally hoards precautionary savings instead of spending. (…)
The structural divide remains stark. While nearly 70% of Chinese live in cities, less than half hold a local hukou, leaving hundreds of millions without equal access to core urban benefits.
Take note. This is a big deal, the key to really unlock China’s housing mess. From my post on Housing in my December series TEN DAYS IN CHINA
China’s housing sector is undoubtedly messy. But it is also very complicated because of the hukou system.
The Chinese hukou is a household registration system that officially identifies a person as a permanent resident of a specific area. It functions as a type of internal passport, linking an individual’s access to social services to their registered location.
Established in its current form in 1958, its primary objectives are to control internal migration, manage social protection, and preserve social stability.
Initially, it was designed to support China’s industrialization by preventing a large-scale exodus from agricultural areas to cities, thereby ensuring a stable food supply and urban order.
A defining feature of the hukou system is its classification of citizens into either “agricultural” (rural) or “non-agricultural” (urban) status.
This designation, which is inherited from one’s parents, dictates an individual’s eligibility for housing and public services.
Urban hukou holders typically have access to better education, healthcare, subsidized housing, and government jobs, while rural hukou holders receive arable land for their livelihood but have limited access to urban benefits.
This has created significant disparities and is often cited as a major source of social inequality in China.
While people can travel and work in places where they are not registered, their lack of a local hukou creates significant barriers.
For instance, a migrant worker from a rural area living in a major city may be unable to enroll their children in local public schools or access the same healthcare benefits as a registered urban resident.
Changing one’s hukou from rural to urban is a difficult process, making these individuals similar to migrants without full rights in their city of residence. Although reforms have been introduced, the system continues to shape economic and social mobility for hundreds of millions.
For Chinese citizens, the ability to purchase a home is strictly gated by hukou status, particularly in top-tier cities like Shanghai, Beijing or Shenzhen, where demand is highest.
In cities like Shanghai, local hukou holders (households) are typically permitted to purchase up to two properties.
In contrast, non-local hukou holders must be married and show proof of local tax or social insurance contributions for several years (often 5+ years) just to qualify to buy a single home.
To combat high inventory of unsold homes, many smaller (tier 2 and 3) cities have relaxed these restrictions, allowing non-locals to buy property more freely, sometimes even offering hukou registration as a perk for buying a home.
Home ownership in China is above 90% (US: 63%), among the highest in the world:
-
87% in urban areas (urban residents with local hukou (non‑migrants). Homeownership for migrants in the destination city is about 14%, but they often have a housing “fallback” in their rural place of registration through family housing/homestead arrangements.
-
96% in rural areas where households are given land and generally self build their house.
A declining population, high ownership and a stiff hukou system make housing a slow growth sector at best. Knowing that, Beijing is mainly trying to help slowly clear the overhang without a major crisis, unlikely given the relatively low housing leverage in China.
As a mitigating factor, the housing overhang has kept rents low in recent years, helping urban migrants’ cashflows.
Urban Chinese (67% of the population) live in apartments, with the average size around 40sqm (430sqf), about 25% the size of the average US home and 15% that of newly built US houses.
This significantly limits the quantity of physical stuff Chinese households can possess compared with American households, effectively curbing consumption of certain types of goods.
Until the government sets up a comprehensive social safety net, the Chinese will keep saving at a very high rate (24.5% in 2024).
Prospects for a meaningful and lasting domestic consumption cycle in China are thus limited by several key factors that don’t seem likely to change rapidly.
What do Chinese do with their money?
-
They save.
-
They travel.
-
They eat out.
-
They buy cars, high end electronics, luxury goods, jewelry, cosmetics.
It means China will continue to send goods abroad, further contributing to global goods deflation or noflation.
Allowing migrants to buy homes in their work cities will create new, potentially large demand for urban housing and promote consumption, even more so if the social safety net is simultaneously improved.
***
***
Highlighting Colbert’s run from a satirical version of himself on “The Colbert Report” to “The Late Show,” the New York Times’ James Poniewozik writes:
“Colbert presided over an era when political TV comedy could take a side and still succeed. Or actually, two eras, which almost perfectly coincided with his two shows: one that parodied politics, one made in a time when politics became a parody of itself.”