“There are an amazing number of good ideas and interesting points made in Unintended Consequences. The thinking underlying it, and the obvious depth of understanding of the author, are very impressive.” - Steven Levitt, coauthor of Freakonomics; 2004 John Bates Clark Medal
“…a very valuable contribution.” - Larry Summers, former Secretary of the Treasury and director of the National Economic Council, president emeritus, Harvard University
“…a must-read for serious students of economic policy.” - Glenn Hubbard, Dean, Columbia Business School, and former Chairman of the Council of Economic Advisers
“…challenges misconceptions that distort our economic debates.” - Arthur Brooks, President of the American Enterprise Institute
“Unintended Consequences represents the most cogent and persuasive analysis of the Financial Crisis to date.” - Andrei Shleifer, 1999 John Bates Clark Medal Winner
“…a very valuable contribution.” - Larry Summers, former Secretary of the Treasury and director of the National Economic Council, president emeritus, Harvard University
“…challenges misconceptions that distort our economic debates.” - Arthur Brooks, President of the American Enterprise Institute
“Unintended Consequences is full of substance, it is one of the must-read books of the year, and once I finish it I will be giving it a second read through right away.” - Tyler Cowen, Professor, George Mason University
“Unintended Consequences provides a provocative interpretation of the causes of the global financial crisis and the policies needed to return to rapid growth. Whether you agree or not, this analysis is well worth reading.” - Nouriel Roubini, New York University; Chairman, Roubini Global Economics
“…serious thinking for serious thinkers. …a thought-provoking blueprint for growing middle- and working-class incomes.” - Mitt Romney, former Governor of Massachusetts
“Unintended Consequences offers deep and well-argued analyses on almost every issue.” - The New York Times
“…a comprehensive explanation of the modern economy.” - Julian Robertson, Founder, Tiger Management
AI Summary.U.S. crude oil production rose from about 5m barrels per day in the late 2000s to nearly 14m by 2025, turning the country from a major petroleum importer into a net exporter. The authors estimate that adverse oil supply news shocks, which once contracted industrial production, now expand it.
Diego Känzig, James Stock and Luca ZanottiNational Bureau of Economic Research
Date Posted:
The US shifted from a large net oil importer to a net exporter around 2010. Computing separate impulse responses pre-and post-2010, Känzig, et al find that adverse global oil-supply news switches from reducing US industrial production to raising it by a similar magnitude.
Has becoming a net oil exporter reversed how supply shocks affect output?
Core argument: Before 2010, an adverse oil supply news shock produced a persistent contraction in U.S. industrial production of roughly -0.6 percent after a year, while from 2010 onward the point estimate turns positive over most horizons and peaks near +1 percent.
As Figure 2 shows, U.S. crude oil production rose from roughly 5 million barrels per day in the late 2000s to almost 14 million barrels per day by 2025. Over the same period, net imports of crude oil and petroleum products fell steeply and turned negative in 2019. The United States thereby went from being one of the world’s largest petroleum importers to a net petroleum exporter. We examine whether the U.S. response to oil supply news shocks differs in the post-shale period by allowing the impulse responses to change from January 2010 onward. Figure 7 reports the results. Before 2010, an adverse oil supply news shock produced a persistent contraction in industrial production, reaching roughly -0.6 percent after a year. From 2010 onward, the point estimate instead becomes positive over most horizons and reaches almost +1 percent at its peak. Both the pre-2010 contraction and the post-2010 expansion are economically large and statistically significant over a range of horizons. This contrast mirrors the evidence from the time-varying VAR.
Takeaways by Macro Roundup® AI
Before 2010, an adverse oil supply news shock produced a persistent contraction in U.S. industrial production of roughly -0.6 percent after a year, while from 2010 onward the point estimate turns positive over most horizons and peaks near +1 percent.
U.S. crude oil production rose from roughly 5 million barrels per day in the late 2000s to almost 14 million barrels per day by 2025, with net imports of crude oil and petroleum products turning negative in 2019.
The pre-2010 contraction and the post-2010 expansion in U.S. industrial production following oil supply news shocks are both economically large and statistically significant over a range of horizons, mirroring evidence from a time-varying VAR.
Related Articles:
U.S. Economy Less Vulnerable To Geopolitical Oil Price Shocks Than In The Past— The United States became a net oil exporter in late 2019, and oil and oil product spending fell from near 8% of GDP in 1980 to 3% in 2024. The authors estimate a 15% global oil supply disruption would now cut annualized U.S. real GDP growth by 0.3pp, versus 1.7pp in the rest of the world.
Fighting Words: The Energy Transition in 2026— Estimates based on primary energy overstate how much electricity is needed to replace oil in passenger cars, since electric motors convert 80%-90% of electricity into motion versus about 18% for gasoline cars. Measuring useful final energy net of waste heat avoids this "primary energy fallacy": the US consumed 45 exajoules of useful final energy versus 95 exajoules of primary energy.
Power Politics: Energy Self-Sufficiency in a Modern Mercantilist World— China leads in adding power to the grid, with more than 30,000 miles of ultra-high-voltage transmission lines versus zero in the US and lead times for large transformers of 48 weeks against a US average of 143 weeks. This positions China to accommodate new demand sources such as artificial intelligence.
AI Summary.The author argues that a slowdown in artificial-intelligence capital spending would remove a stimulus that has been masking weakness, colliding with overcapacity, falling prices, refinancing needs and weaker labor income. On this view the main risk is deflationary, transmitting through investment, employment, consumption, credit, housing and asset prices at once.
Paul KedroskyApplied Complexity
Date Posted:
Kedrosky argues that the “inevitable transition” from the current inflationary AI-buildout regime to a new regime in which AI-driven productivity gains exert deflationary pressure is “where the real macro risk lies.”
Would an AI spending slowdown create deflationary rather than inflationary risk?
Core argument: The author argues the dangerous phase arrives when the AI capex boom slows, removing stimulus just as overcapacity, falling prices, refinancing needs and weaker labor income hit together, with a refinancing wall waiting in 2028.
If AI investment has been masking underlying weakness, a slowdown removes the stimulus just as overcapacity, falling prices, refinancing needs and weaker labor income hit together. A refinancing wall sits waiting in 2028. Failure is over-determined. This moment sits at the intersection of the four forces that have produced the most consequential boom-bust cycles in Western economic history: loose credit, government policy, technology, and real estate. There are, as a result, myriad catalysts, each with a low probability of happening, but which collectively guarantee collapse. That is why the principal risk is deflationary rather than inflationary. An AI bust could transmit through capex, employment, consumption, credit, housing, and asset prices simultaneously.
Takeaways by Macro Roundup® AI
The author argues the dangerous phase arrives when the AI capex boom slows, removing stimulus just as overcapacity, falling prices, refinancing needs and weaker labor income hit together, with a refinancing wall waiting in 2028.
The principal risk from an AI bust is deflationary rather than inflationary, according to the author, because it could transmit simultaneously through capex, employment, consumption, credit, housing, and asset prices.
The author places this moment at the intersection of loose credit, government policy, technology, and real estate — the four forces behind the most consequential boom-bust cycles in Western economic history — and sees a plausible decade of Japan-style balance recession with fiscal stimulus constrained by massive deficits.
Related Articles:
Financing the AI Buildout— The authors estimate roughly 57 gigawatts of operating US data center capacity against a planned development pipeline of about 509 gigawatts, which is not a completion forecast since projects may be delayed, scaled back, or abandoned. Their central scenario has 182.8 gigawatts of added capacity becoming operational through 2032, 117.2 gigawatts completed later, and 226.9 gigawatts never completed.
A Catalyst For The AI Bubble Break— Historically, controlling for valuations, a 1% rise in the share of the market represented by initial public offerings has been associated with roughly a 4% decline in prices. With large artificial-intelligence company listings expected and share issuance exceeding buybacks, equity supply could shift from shrinking about 1% a year to growing near 5%, implying returns roughly 20% below average.
AI Set to be Largest CapEx Cycle Ever … and Soon Majority Externally Financed— Hyperscaler and data-center investment is tracking toward $1.5tn annually by 2030, which in real terms would exceed the 2005-06 housing boom, the 1990s fiber peak, and the 1882 railroad peak combined. External debt and equity are projected to fund close to 90% of the buildout, making it dependent on credit markets.
AI Summary.Historically, controlling for valuations, a 1% rise in the share of the market represented by initial public offerings has been associated with roughly a 4% decline in prices. With large artificial-intelligence company listings expected and share issuance exceeding buybacks, equity supply could shift from shrinking about 1% a year to growing near 5%, implying returns roughly 20% below average.
Ben Inker and John PeaseGMO
Date Posted:
GMO estimates net US equity supply could increase by ~5% or more of total market capitalization over the next year. Inker and Pease note that in 1999–2000, “IPOs peaked at ~5% of total market cap. It doesn’t take much new supply to overwhelm the market.”
How does rising equity supply from new listings affect stock returns?
Core argument: Controlling for valuations, a 1% increase in the market share of IPOs has historically led to an approximate 4% decrease in prices, a multiplier the author applies to a coming surge in equity supply.
Controlling for valuations, a 1% increase in the market share of IPOs has historically led to an approximate 4% decrease in prices [Exhibit 2 in gallery]. Consider what is happening to equity supply. SpaceX’s non-Elon sales alone exceed 1% of the U.S. stock market’s value. We are likely to see IPOs from both Anthropic and OpenAI—together valued at about 5% of the U.S. stock market’s total investable market cap —over the next 12 months. For the first time in twenty years, with the brief exception of the GFC, we are seeing net dilution in the U.S. stock market (i.e., share issuance exceeding share buybacks), as the hyperscalers continue their capital expenditure spree. In aggregate, we are going from a world where equity supply was shrinking by roughly 1% per annum (with buybacks exceeding primary or secondary issuance) to one in which equity supply will be increasing by close to 5%, if not more. If you believe the 4x multiplier [implied by the regression line in Exhibit 2], this means we should expect returns that are some 20% below average (roughly 6% real) over the next 12–18 months from the headwind of rising equity supply alone.
Takeaways by Macro Roundup® AI
Controlling for valuations, a 1% increase in the market share of IPOs has historically led to an approximate 4% decrease in prices, a multiplier the author applies to a coming surge in equity supply.
U.S. equity supply is shifting from shrinking roughly 1% a year, with buybacks exceeding issuance, to growing close to 5% or more, the first net dilution in twenty years apart from the Global Financial Crisis.
Anthropic and OpenAI, together valued at about 5% of U.S. investable market cap, are likely to go public within 12 months, while SpaceX’s non-Elon sales alone exceed 1% of the U.S. stock market’s value.
Related Articles:
Momentum, Rotation and the Value in Growth— After a fifteen-year stretch in which US equities, technology and growth stocks led, equity gains have broadened geographically, with the US the weakest of the major regions. The five largest US stocks now trade at a price-to-earnings ratio only marginally above the other 495, after holding a consistent premium since 2017.
SpaceX and OpenAI Are Ending Wall Street’s Era of Stock Scarcity— JPMorgan projects that share sales will add roughly $1.5tn of stock to the US equity market over the next two years even after buybacks, which would be the strongest net equity issuance since at least the late 1990s. Equity has grown cheaper relative to debt as a funding source.
One Hundred Years in the U.S. Stock Markets— Over a century, U.S. publicly traded stocks generated roughly $91tn in shareholder wealth enhancement, with returns strongly positively skewed and dominated by a relatively small number of high-performing firms. Because most possible outcomes fall below the mean in a skewed distribution, the authors argue financial planning should focus on median outcomes or the full distribution.
AI Summary.An investment argument holds that long-term interest rates fall under either artificial intelligence outcome: success would generate trillions in technology revenue and prove deflationary, while failure would burst the bubble, cut the Nasdaq in half, and send investors from stocks into Treasury bonds.
Torsten SløkApollo
Date Posted:
Sløk argues that the next six months will see either a large increase in revenue associated with AI or the start of a major market correction. He argues either scenario will bring long rates down.
Do long-term rates fall whether or not AI succeeds?
Core argument: If AI fails to deliver, the argument runs, the bubble bursts and the Nasdaq falls 50% as investors rotate out of equities into Treasuries, pushing long rates dramatically lower.
If AI succeeds and tech companies generate trillions in revenue, AI will be massively deflationary and push rates lower. If AI does not work out, the bubble bursts and the Nasdaq is down 50% as investors rotate out of equities into Treasuries and long rates fall dramatically. Over the next six months, the market will make up its mind about which AI scenario is playing out. The narrative in rates today is all about inflation and fiscal problems. But the narrative going into 2027 is going to be all about either the success or failure of AI. And in both scenarios, long rates are going to be lower.
Takeaways by Macro Roundup® AI
If AI fails to deliver, the argument runs, the bubble bursts and the Nasdaq falls 50% as investors rotate out of equities into Treasuries, pushing long rates dramatically lower.
If AI succeeds and tech companies generate trillions in revenue, the argument runs, AI proves massively deflationary and pushes rates lower, leaving long rates lower in both scenarios.
The rates narrative is expected to shift from inflation and fiscal problems to the success or failure of AI going into 2027, with the market making up its mind over the next six months.
Related Articles:
The Summer I Turned Pretty— A warning sign in a boom cycle is when companies closest to final demand weaken while capital spending beneficiaries keep thriving, as when telecom operators' shares rolled over while communications equipment stocks kept rising. Semiconductor stock prices and free cash flow are soaring while large cloud providers' shares stagnate and their free cash flow plummets.
The Buyers of AI Are Still Waiting for the Payoff— Companies in the S&P 500 outside the technology sector are spending heavily on artificial intelligence without showing gains in profit margins, with health care margins halved since 2015, consumer staples near 6%, and consumer discretionary near 8%. So far the artificial intelligence capital spending boom appears in sellers' margins rather than buyers'.
The Market Is Asking Questions— Markets are weighing whether the trillions committed upfront to AI data centers, chips, and power will generate returns above the cost of capital before those assets depreciate. With the buildout increasingly funded by debt rather than internal cash flow and borrowing spreads widening, financing costs could rise until new data-center spending no longer clears its return hurdle.
AI Summary.Artificial intelligence models are converging in capability and price, making switching easier and margins harder to defend, with US labs cutting prices on lower- and mid-tier offerings while keeping frontier models expensive. An 80% annual price decline requires 400% unit growth just to hold revenue flat, straining firms financing enormous capital spending.
Paul KedroskyApplied Complexity
Date Posted:
US labs have “conceded commoditization” for models behind the frontier, while still trying to “preserve scarcity rents” for their frontier models. Kedrosky warns “this market-ceding tactic has failed in most markets historically, especially when competing with China.”
Can AI firms defend margins as model prices converge?
Core argument: An 80% year-over-year decline in AI model prices requires 400% unit growth simply to hold revenue flat, and considerably more to generate growth.
Good-enough models are converging in capability and price. That makes model switching easier and durable margins harder to defend. US labs are cutting lower- and mid-tier offerings while keeping their best models expensive. They are conceding commoditization below the frontier while trying to preserve scarcity rents at the top. This market-ceding tactic has failed in most markets historically, especially when competing with China. This is irreconcilable with trillion-dollar valuations and high & growing debt loads. OpenAI and Anthropic must finance enormous capital spending while competition pushes the models likely to generate the most volume toward commodity pricing. An 80% year-over-year price decline requires 400% unit growth just to stand still, and much more to deliver overall growth.
Takeaways by Macro Roundup® AI
An 80% year-over-year decline in AI model prices requires 400% unit growth simply to hold revenue flat, and considerably more to generate growth.
US AI labs are cutting prices on lower- and mid-tier models while keeping frontier models expensive, conceding commoditization below the frontier while trying to preserve scarcity rents at the top.
OpenAI and Anthropic must finance enormous capital spending as competition pushes the models likely to generate the most volume toward commodity pricing, a position irreconcilable with trillion-dollar valuations and high and growing debt loads.
Related Articles:
OpenAI and Anthropic In Price War as Chinese AI Rivals Gain Ground— OpenAI cut its GPT-5.6 Luna prices by 80% and Anthropic launched Opus 5 at half the price of its Fable 5 model while cancelling a planned Sonnet 5 price increase. Headline token prices are not directly comparable, since more capable models can finish tasks using fewer tokens or attempts.
Who’s Afraid of Chinese Models?— Demand for frontier artificial intelligence models exceeds supply because computing capacity is scarce, allowing chip suppliers and their customers to resell compute at high margins. The author argues alarm over Chinese models is overblown, expecting the inference market to grow faster than training costs so labs can thrive at lower prices.
The AI Trade Is Losing One of Its Key Signals— An index blending artificial intelligence token prices and usage is ambiguous when it softens, since token prices have fallen more than 90% since 2023 while total spending has roughly doubled over the past year. Allianz Research measures a nearly 46% growth gap between artificial intelligence investment and sales, wider than the 32% divergence during the 2001 telecom bust.
AI Summary.Anthropic, an artificial-intelligence company valued at $965bn, faces investors taking a more critical look at its business as it approaches a public listing. Company executives have played down Chinese competition, arguing that most users want the most capable models available and that Chinese systems trail top models by at least a few months.
Berber Jin, Corrie Driebusch and Kate ClarkWall Street Journal
Date Posted:
Anthropic is meeting with investors ahead of a planned IPO in September or early October. The firm has downplayed the potential impact of Chinese competition, arguing “most users want the most intelligent AI systems available at any given time.”
Does Chinese competition threaten a leading AI company's valuation?
Core argument: Anthropic, valued at $965 billion, faces fresh challenges ahead of its planned IPO that are leading investors to take a more critical look at its business.
The $965 billion artificial-intelligence juggernaut is staring down a host of fresh challenges that are leading investors to take a more critical look at the company’s business. Company executives have played down the impact of Chinese competition in meetings, telling investors that Anthropic is hyperfocused on offering cutting-edge AI models, the people said. Chief Executive Dario Amodei and other top U.S. AI leaders have publicly said that most users want the most intelligent AI systems available at any given time, suggesting that Chinese systems are less of a threat since their capabilities generally trail those of top AI models by at least a few months.
Takeaways by Macro Roundup® AI
Anthropic, valued at $965 billion, faces fresh challenges ahead of its planned IPO that are leading investors to take a more critical look at its business.
Anthropic executives have played down the impact of Chinese competition in investor meetings, saying the company is hyperfocused on offering cutting-edge AI models.
Anthropic Chief Executive Dario Amodei and other top U.S. AI leaders have publicly said most users want the most intelligent AI systems available at any given time, with Chinese systems' capabilities generally trailing top AI models by at least a few months.
Related Articles:
They’re Shooting at Everyone!: The Yossarian Effect in Capital Markets— A concentrated set of highly correlated assets financed with 3-4x leverage meant a large move in one part of the market forced selling elsewhere. Funds holding no idle cash sold liquid winners such as semiconductors and other artificial intelligence names to fund SpaceX purchases, with index inclusion triggering further forced reallocation.
The Summer I Turned Pretty— A warning sign in a boom cycle is when companies closest to final demand weaken while capital spending beneficiaries keep thriving, as when telecom operators' shares rolled over while communications equipment stocks kept rising. Semiconductor stock prices and free cash flow are soaring while large cloud providers' shares stagnate and their free cash flow plummets.
OpenAI Leans Toward Waiting Until Next Year for I.P.O.— OpenAI is leaning toward delaying its initial public offering until next year, according to people involved in the company's deliberations. Advisers have cautioned that retail investor enthusiasm may be limited, following a decline in SpaceX's stock after its public offering and choppy markets in which technology stocks have dragged down indexes.
AI Summary.Free cash flow at large technology companies is deteriorating rapidly as data-center spending absorbs it, even as the rest of the corporate sector generates more cash. Across the S&P 500, earnings per share have risen much faster than free cash flow over a protracted stretch, raising questions about how quickly artificial-intelligence capital spending will turn into profits.
John AuthersBloomberg
Date Posted:
As the Mag 7’s free cash flow has turned negative, that of the S&P 500 net the Mag 7 has ticked up since the start of the year.
Is data-center spending eroding big tech's free cash flow?
Core argument: Tech companies' free cash flow is deteriorating at an unprecedented pace as data centre spending consumes it, even as the rest of the corporate sector generates more cash.
Questions are mounting over how quickly the capex ploughing into AI will translate into profitability, but tech’s earnings growth is still impressive — it’s just that the cost of that investment is increasingly hard to ignore. Their free cash flow is deteriorating at an unprecedented pace as spending on data centres gobbles it up, even as the rest of the corporate sector generates more cash. The numbers for the overall S&P 500, as shown by Bloomberg’s Graph Fundamentals function, confirm a protracted run in which earnings per share have risen much faster than free cash flow. This started after companies had made big pandemic cost cuts and then went into overdrive after the arrival of ChatGPT.
Takeaways by Macro Roundup® AI
Tech companies' free cash flow is deteriorating at an unprecedented pace as data centre spending consumes it, even as the rest of the corporate sector generates more cash.
S&P 500 earnings per share have risen much faster than free cash flow in a protracted run, as shown by Bloomberg's Graph Fundamentals function.
The divergence between S&P 500 earnings per share and free cash flow began after companies made big pandemic cost cuts and went into overdrive after the arrival of ChatGPT.
Related Articles:
The Buyers of AI Are Still Waiting for the Payoff— Companies in the S&P 500 outside the technology sector are spending heavily on artificial intelligence without showing gains in profit margins, with health care margins halved since 2015, consumer staples near 6%, and consumer discretionary near 8%. So far the artificial intelligence capital spending boom appears in sellers' margins rather than buyers'.
The Summer I Turned Pretty— A warning sign in a boom cycle is when companies closest to final demand weaken while capital spending beneficiaries keep thriving, as when telecom operators' shares rolled over while communications equipment stocks kept rising. Semiconductor stock prices and free cash flow are soaring while large cloud providers' shares stagnate and their free cash flow plummets.
A Slower AI Payoff Would Be Everyone’s Problem— If artificial intelligence profits arrive later than expected, committed capital spending and depreciation would still hit on schedule, squeezing margins, stretching balance sheets toward more debt and possible ratings downgrades, and repricing equities. Because a handful of large technology firms dominate the indices, the author argues the damage would spread to chips, power, data centers and the broader market.
AI Summary.In artificial intelligence, profit margins rise the further a company sits from the end customer, reversing the usual pattern in which owning the customer relationship commands the highest margins. The author argues the upstream profits are funded by capital raised by the loss-making consumer-facing layer rather than by cash generated from end demand.
Torsten SløkApollo
Date Posted:
AI profits have so far come only from the buildout itself, while the end-product generates losses. Slok notes that in the absence of revenue from models and applications, raising capital “can bridge the gap for a while, but not indefinitely.”
Are AI profit margins higher farther from the end customer?
Core argument: In the AI value chain, profit margins rise the further a firm sits from the end user, reversing the usual pattern in which the owner of the end-customer relationship earns the higher margin.
In business, profit margins are frequently higher for the owner of the end-customer relationship. But that is not the case for AI. In AI, profit margins are higher the further you get from the end user. This is important because it means the AI boom's profits are currently being funded by investors rather than earned from customers. The upstream margins are real, but they are paid for out of capital raised by the layer losing money, not out of cash generated by end demand. That makes the 41% contingent on the -59% continuing to be financeable. The bottom line is that the most profitable part of the AI value chain depends on the least profitable part continuing to grow revenue or raise capital.
Takeaways by Macro Roundup® AI
In the AI value chain, profit margins rise the further a firm sits from the end user, reversing the usual pattern in which the owner of the end-customer relationship earns the higher margin.
The AI boom's upstream 41% margins are paid out of capital raised by the layer losing money at -59%, not out of cash generated by end demand.
The most profitable part of the AI value chain depends on the least profitable part continuing to grow revenue or raise capital.
Related Articles:
The State of the AI Economy— An analysis estimates the AI ecosystem generated $110bn in revenue over 12 months excluding double-counting, with the most recent month annualizing to a $175bn run rate. After separating AI-specific capital spending from the roughly $120bn hyperscalers already spent annually and depreciating compute over 6 years, revenues attributable to hyperscalers roughly cover depreciation.
AI: The ROI Runway Could Be Long Outside the Tech Sector— AI company valuations rest on the assumption that profit margins across non-tech firms will eventually rise, but outside a handful of sectors adoption is slow, creating a divergence between front-loaded valuations and slower cash flows. If the per-use costs of AI models converge toward zero, revenue may be insufficient for all major cloud providers.
A Slower AI Payoff Would Be Everyone’s Problem— If artificial intelligence profits arrive later than expected, committed capital spending and depreciation would still hit on schedule, squeezing margins, stretching balance sheets toward more debt and possible ratings downgrades, and repricing equities. Because a handful of large technology firms dominate the indices, the author argues the damage would spread to chips, power, data centers and the broader market.
