Edward Conard

Top Ten New York Times Bestselling Author

  • “Unintended Consequences offers deep and well-argued analyses on almost every issue.” - The New York Times
  • “…challenges misconceptions that distort our economic debates.” - Arthur Brooks, President of the American Enterprise Institute
  • “…a must-read for serious students of economic policy.” - Glenn Hubbard, Dean, Columbia Business School, and former Chairman of the Council of Economic Advisers
  • “…a must-read for serious students of economic policy.” - Glenn Hubbard, Dean, Columbia Business School, and former Chairman of the Council of Economic Advisers
  • “…a fresh argument for the productive value of inequality.” - David Autor, Professor of Economics, Massachusetts Institute of Technology
  • “…challenges misconceptions that distort our economic debates.” - Arthur Brooks, President of the American Enterprise Institute
  • “…reminds us that inequality sends a signal of what society lacks most, in America’s case, entrepreneurship and risk taking.” - Lawrence Lindsey, CEO, The Lindsey Group, former Director of the National Economic Council
  • “Unintended Consequences represents the most cogent and persuasive analysis of the Financial Crisis to date.” - Andrei Shleifer, 1999 John Bates Clark Medal Winner
  • “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
  • “…reminds us that inequality sends a signal of what society lacks most, in America’s case, entrepreneurship and risk taking.” - Lawrence Lindsey, CEO, The Lindsey Group, former Director of the National Economic Council
  • “Unintended Consequences is far smarter and more thought-provoking than most economics written for the general public” - Greg Mankiw, Harvard University, Former Chairman of the Council of Economic Advisors
  • “Unintended Consequences should be read by anyone who takes for granted the superiority of progressive taxation and has not thought carefully about the trade-offs involved.” - The New Republic
Upside of Inequality Oxford Unintended Consequences

Macro Roundup


Wednesday, December 10, 2025

AI Summary. Transformational technologies have historically generated excessive enthusiasm and investment, producing more infrastructure than needed, asset prices that prove too high, and losses for those funding them, while also accelerating adoption. AI enthusiasm is argued to fit this pattern, with debt financing having the potential to magnify the effects.

Howard Marks Oaktree Capital
Date Posted:

Like previous “transformational technologies,” Marks writes, AI is going through a period of excess enthusiasm, valuations, and investment given the unfeasibility of knowing “which [firms] will emerge as winners.” Today, these dynamics are amplified by debt financing.

Does AI enthusiasm follow the historical pattern of technology investment bubbles?

Core argument: Transformational technologies have consistently generated excessive enthusiasm and investment, leaving more infrastructure than needed and asset prices that prove too high, and AI failing to follow that pattern would be a first.

There’s a consistent history of transformational technologies generating excessive enthusiasm and investment, resulting in more infrastructure than is needed and asset prices that prove to have been too high. These excesses accelerate the adoption of the technology in a way that wouldn’t occur in their absence. AI is currently the subject of great enthusiasm. If that enthusiasm doesn’t produce a bubble conforming to the historical pattern, that will be a first. Bubbles created in this process usually end in losses for those who fuel them. The losses stem largely from the fact that the technology’s newness renders the extent and timing of its impact unpredictable. This in turn makes it easy to judge companies too positively amid all the enthusiasm and difficult to know which will emerge as winners when the dust settles. There can be no way to participate fully in the potential benefits from the new technology without being exposed to the losses that will arise if the enthusiasm and thus investors’ behavior prove to have been excessive. The use of debt in this process – which the high level of uncertainty usually precluded in past technological revolutions – has the potential to magnify all of the above this time.

Takeaways by Macro Roundup® AI

  1. Transformational technologies have consistently generated excessive enthusiasm and investment, leaving more infrastructure than needed and asset prices that prove too high, and AI failing to follow that pattern would be a first.
  2. Losses from technology bubbles stem largely from the newness of the technology making the extent and timing of its impact unpredictable, which makes companies easy to judge too positively and eventual winners hard to identify.
  3. The use of debt, which high uncertainty usually precluded in past technological revolutions, has the potential to magnify the excesses and the resulting losses in the case of AI.

Related Articles:

  • Minsky Moments and AI CapEx — Capital-spending forecasts for the four largest cloud providers rose nearly 50% over the year, and AI data centers are increasingly funded through leveraged and off-balance-sheet structures, including $3.2bn in senior secured notes and a roughly $30bn asset-backed financing. The author argues this marks a speculative-finance phase where borrowers must roll over principal.
  • On Benchmark Games, Gemini, and Declining Returns to Scale — Composite benchmarks show Google's Gemini roughly in line with GPT-5-class models, with any gap small, and the author argues it does not lead on strict low-leakage reasoning exams. The pattern reflects growing training costs yielding modest gains concentrated in vision, video, and some coding rather than broad reasoning advances.
  • Why We Are Not In A Bubble….Yet — The 5 largest US technology companies are worth about 16% of the global public equity market, and the top 10 US stocks nearly 25%, almost $25tn. The author argues valuations are high but not bubble-level, with the "Magnificent 7" median forward price-to-earnings ratio at 27x, roughly half that of the late-1990s leaders.

AI Summary. Private economists and Federal Reserve officials disagree about the level of the "neutral" interest rate—the inflation-adjusted short-term rate expected to prevail with the economy at full strength and stable inflation. Averaging model-based and market-based estimates puts the neutral real rate near 1.5%, and with expected inflation above the Federal Reserve's 2% target, the neutral nominal rate is correspondingly higher.

Tom Lawler Calculated Risk
Date Posted:

The average of various models of the real natural short rate and the “market’s” assessment has R* very close to 1.5%. With expected inflation > 2%, today’s 25bp cut will leave the funds rate “very close to the neutral nominal policy rate.”

How high is the neutral interest rate?

Core argument: Averaging model-based and market-based estimates puts the best guess for the neutral real short-term interest rate (R*) at very close to 1.5%, though private economists and Federal Reserve members disagree considerably about its level.

There is a considerable disagreement among both private economists and Federal Reserve members about the level of the so-called “neutral” rate of interest (which some label R*), which is the real (or inflation-adjusted) short-term interest rate expected to prevail if the economy is at full strength and inflation is stable. Some folks look to economic “models” to estimate R*, while other folks look to “the markets” for the “market’s” estimate of R*. If one takes the approach that the “best guess” estimate of R* is found by looking at the average of various models and the “market’s” assessment of R*, one would come to the conclusion that the current “best guess” estimate of the neutral real rate of interest is very close to 1.5%. “Consensus” forecasts as well as “market-based” measures suggest that expected inflation over the next year is above the Federal Reserve’s targeted inflation rate of 2%. As such, the current “neutral” nominal interest rate would be above a neutral real interest rate estimate where expected inflation was at 2%. If that is the case, and if, as expected, the FOMC decides to cut its federal funds rate target by 25 bp tomorrow, then the resulting level of the federal funds rate will be very close to the neutral nominal policy rate.

Takeaways by Macro Roundup® AI

  1. Averaging model-based and market-based estimates puts the best guess for the neutral real short-term interest rate (R*) at very close to 1.5%, though private economists and Federal Reserve members disagree considerably about its level.
  2. Consensus forecasts and market-based measures place expected inflation over the next year above the Federal Reserve's 2% target, which puts the neutral nominal interest rate above an estimate built on 2% expected inflation.
  3. A 25 basis point cut in the federal funds rate target by the FOMC would leave the resulting rate very close to the neutral nominal policy rate.

Related Articles:

  • Trump’s Interest Rate Obstacle Is Bigger Than Powell — Bloomberg Economics estimates the natural interest rate has climbed from a low of 1.7% in 2012 to about 2.5%, and projects 2.8% by 2030 as baby boomers retire, foreign savings gluts fade, and government borrowing rises — keeping ten-year Treasury yields between 4.5% and 5%.
  • The Race Between Asset Supply and Asset Demand — The authors estimate that demand for assets has outpaced supply by roughly 400% of GDP versus 100%, making rising asset demand the leading contributor to falling interest rates. They project that debt sustainability requires fiscal consolidation exceeding 10% of GDP, though debt could reach 250% of GDP without raising interest rates.
  • Data Update 4 for 2025: Interest Rates, Inflation and Central Banks! — Interest rates are ultimately driven by macroeconomic fundamentals, with an "intrinsic risk-free rate" defined as the sum of the inflation rate and real GDP growth tracking the observed 10-year Treasury bond yield closely. On this view, central banks' influence comes from preferential access to data on those fundamentals, their capacity to alter them, and their credibility.

AI Summary. China redirected domestic savings toward de-Westernizing its industrial supply chains through tighter capital controls, suspended stock listings, and concentrated bank lending, which the author argues was deflationary and produced poor investor returns. The author argues that if the US pursues comparable supply-chain separation from China, capital would shift into lower-return activities.

Louis-Vincent Gave Gavekal
Date Posted:

Gave argues that post-2018, China undertook a painful process of de-Westernizing its supply chain. To de-Sinify American supply chains, the US would have to redeploy capital into activities with lower returns, which is politically unpalatable.

How does redirecting national savings toward supply-chain separation affect investor returns?

Core argument: China spent seven years capturing domestic savings through tighter capital controls, suspended IPOs, and concentrated bank lending, redirecting the money toward de-Westernizing its industrial supply chains.

For China, saving the economy’s long term prospects meant rapidly de-Westernizing its supply chains. Concretely, this meant capturing the massive pool of Chinese savings and redirecting the money away from everything but the single goal of de-Westernizing supply chains. This was massively deflationary for China, and for the world. Essentially, for seven long years China went on a diet, and went to the CrossFit gym. And as all of China’s savings were captured—through tighter capital controls, the suspension of IPOs, and the concentration of bank lending—and redirected towards China’s industrial supply chains, returns for investors were dreadful. The capital markets serving the world’s second largest economy became “uninvestible.” Meanwhile, over the same period, the US did absolutely nothing to de-Sinify its own supply chain. While China hit the gym, the US partied. If US policymakers really are serious about de-Sinifying supply chains, will America’s elites accept the redeployment of capital into activities with lower returns on invested capital, and that in consequence the stock market will struggle and the US dollar roll over?

Takeaways by Macro Roundup® AI

  1. China spent seven years capturing domestic savings through tighter capital controls, suspended IPOs, and concentrated bank lending, redirecting the money toward de-Westernizing its industrial supply chains.
  2. The redirection of Chinese savings into supply-chain de-Westernization was deflationary for China and the world, leaving investor returns dreadful and the capital markets of the world's second-largest economy "uninvestible.".
  3. The United States did nothing to de-Sinify its own supply chains, raising the question of whether American elites would accept capital redeployed into activities with lower returns on invested capital, a struggling stock market, and a falling dollar.

Related Articles:

  • China’s Trade Surplus Tops $1 Trillion After Exports Rebound — China's trade surplus reached $1.1tn in 11 months, exceeding its previous full-year record, as strong export growth to the European Union and Africa more than offset a 29% drop in shipments to the US. France's president has warned the EU may impose tariffs if Beijing does not address the imbalance.
  • The Fracturing Of The World Economy — The US would gain relative to China from a fracturing world economy because its bloc of allies includes nearly all advanced countries, accounting for 68% of world GDP at market prices versus 26% for China's bloc. China's structural current account surplus also leaves it reliant on the US and its allies as safe places to invest those funds.
  • Sticking with What Works: Raising GDP Forecast on China’s Manufacturing Push — Stronger Chinese growth driven by export-oriented manufacturing may not produce net positive spillovers: for every 1pp export-driven boost to China's GDP, other economies may see a 0.1–0.3pp drag, with high-tech producers such as Europe and Japan facing greater pressure. Cheaper Chinese goods continue to exert a disinflationary impulse.

AI Summary. Goldman Sachs estimates the Chinese yuan is about 25% undervalued on a weighted average of two of its models: one based on productivity and terms of trade puts fair value at 5 per US dollar versus roughly 7.06 in the market, while a current-account-based model finds it 12% undervalued.

Srinivasan Sivabalan Bloomberg
Date Posted:

The weighted mean of two GS models, one based on terms of trade and productivity, the other on the current account, have the yuan 25% undervalued. “The strengthening we project would still leave the currency comfortably in inexpensive territory.”

Is the yuan undervalued relative to its economic fundamentals?

Core argument: Goldman Sachs puts the yuan 25% below fair value on a weighted average of two models, against a traded level of about 7.06 per US dollar on Wednesday.

The Goldman Sachs Dynamic Equilibrium Exchange Rate (GSDEER), which posits that exchange rates will eventually reflect differences in productivity and terms of trade, has the yuan’s fair value pegged at 5 per US dollar. The currency traded at around 7.06 on Wednesday. Meanwhile, the Goldman Sachs Fundamental Effective Exchange Rate (GSFEER) framework, which tracks the current account’s influence on exchange rates, finds the yuan 12% undervalued. The weighted average of the two models has the currency pegged at 25% below its fair value. Forwards contracts for the fourth quarter of 2026 price the currency offshore at 6.91 per dollar, or about 2% stronger than its current level, according to data compiled by Bloomberg.

Takeaways by Macro Roundup® AI

  1. Goldman Sachs puts the yuan 25% below fair value on a weighted average of two models, against a traded level of about 7.06 per US dollar on Wednesday.
  2. Goldman Sachs's GSDEER model, which posits exchange rates eventually reflect differences in productivity and terms of trade, pegs the yuan's fair value at 5 per US dollar, while its current-account-based GSFEER framework finds the currency 12% undervalued.
  3. Forward contracts for the fourth quarter of 2026 price the offshore yuan at 6.91 per dollar, about 2% stronger than its current level, according to data compiled by Bloomberg.

Related Articles:

  • The Case that China is Now Actively Resisting Pressure on the Yuan to Appreciate — The author argues China has been resisting pressure for the yuan to appreciate against the dollar, citing state bank foreign asset accumulation, balance of payments data, a yuan trading strong relative to its official fix, and positive foreign exchange settlement. Inflows from China's large trade surplus exceed private capital outflows.
  • China’s “Anti-Involution” Campaign: Global Implications — China's economy has slowed as domestic investment stagnates and consumer spending stays weak, with the shortfall partly offset by rising net exports. The author argues these internal problems are being passed to the rest of the world, since export values are rising about 6% a year while import values are down 2.2%.
  • China Chooses Deflation — China's headline consumer price inflation has fallen below zero and core inflation is little better, and the author argues US tariffs will hasten this deflationary dynamic. Because easing monetary policy would weaken the Yuan against the Dollar, China is accepting deflation instead, which the author argues means rising real interest rates.

AI Summary. Chinese AI developer DeepSeek used Nvidia Blackwell chips that were smuggled into China through countries where their sale was permitted, according to a report citing unnamed sources. The chips were installed in data centers abroad, then dismantled and shipped to China after clearing inspection by server-equipment companies.

Mark Bergen Bloomberg
Date Posted:

DeepSeek has been developing its upcoming model with Nvidia’s Blackwell chips that were smuggled into the country after being removed from datacenters in third countries.

Are Chinese AI developers obtaining banned chips through smuggling routes?

Core argument: DeepSeek used Nvidia Blackwell chips barred from export to China, which were smuggled in through countries where their sale was permitted, The Information reported, citing unnamed sources.

Nvidia’s Blackwell chips were smuggled into China through countries that permitted their sale, The Information reported, citing unnamed sources. More specifically, DeepSeek tapped chips that were installed in data centers in unspecified countries, then dismantled and shipped to China after clearing inspection by companies developing server equipment. A representative for DeepSeek didn’t immediately respond to a request for comment.

Takeaways by Macro Roundup® AI

  1. DeepSeek used Nvidia Blackwell chips barred from export to China, which were smuggled in through countries where their sale was permitted, The Information reported, citing unnamed sources.
  2. The chips were installed in data centers in unspecified countries, then dismantled and shipped to China after clearing inspection by companies developing server equipment, according to The Information.
  3. A representative for DeepSeek did not immediately respond to a request for comment on the report.

Related Articles:

  • America’s Chip Restrictions Are Biting in China — Shortages of advanced semiconductors in China are acute enough that authorities have begun intervening in how output from Semiconductor Manufacturing International, the country's largest contract chip maker, is distributed, prioritizing Huawei's artificial-intelligence chip production. Chinese firms also smuggle Nvidia chips or access them remotely through cloud computing abroad.
  • The Geopolitics of AI: Decoding the New Global Operating System — The performance gap between the top American and Chinese AI models narrowed from 9.3% to 1.7%, with Chinese firms relying on cheaper open-source models—DeepSeek prices usage 20-40x below OpenAI's ChatGPT. This price advantage could win adoption in developing countries and build dependency on Chinese AI ecosystems.
  • DeepSeek Debates: Chinese Leadership On Cost, True Training Cost, Closed Model Margin Impacts — Estimates put algorithmic progress in artificial intelligence at roughly 4x per year, meaning the compute needed for a given capability level falls 4x annually, with some industry leaders arguing improvements run closer to 10x. The Bank of China announced a $140bn (1tn yuan) five-year subsidy for the AI industry chain aimed at making China self-reliant in science and technology.

Thursday, December 11, 2025

AI Summary. Construction labor productivity has been roughly flat or declining since the 1990s across most of Western Europe, Japan, and Korea, a pattern consistent with the flat US trend. Countries often held up as models — Sweden and Japan for prefabrication, China for infrastructure — show little or unremarkable construction productivity improvement.

Brian Potter Construction Physics
Date Posted:

The sharp contrast between soaring productivity in manufacturing and limited or no such growth in construction since the 1990s is not limited to the US, but largely holds across advanced countries, even those the US is “encouraged to emulate.”

Is flat construction productivity a problem across wealthy countries?

Core argument: Most Western European countries, along with Japan and Korea, have had flat or declining construction labor productivity growth since the 1990s, and the roughly flat US trend is fairly consistent with trends in other large, wealthy countries.

Most Western European countries, along with Japan and Korea, have had flat or declining labor productivity growth since the 1990s. The recent US trend in labor productivity — it staying roughly flat — is fairly consistent with trends in other large, wealthy countries. This includes countries whose construction practices the US is often encouraged to emulate. Sweden, for instance, is often praised for its high adoption of prefabricated construction, but since the 1990s Sweden has had roughly flat construction labor productivity, even as it has made more extensive use of prefabrication. Japan similarly makes wider use of prefabrication than the US does, and Japan has proved willing to experiment with technologies like automated skyscraper construction, but has seen virtually no construction productivity improvement since the 1970s. China is held up as an example of an infrastructure-focused nation that the US should emulate, but its construction productivity gains, while positive, don’t appear to be particularly impressive.

Takeaways by Macro Roundup® AI

  1. Most Western European countries, along with Japan and Korea, have had flat or declining construction labor productivity growth since the 1990s, and the roughly flat US trend is fairly consistent with trends in other large, wealthy countries.
  2. Sweden, often praised for its high adoption of prefabricated construction, has had roughly flat construction labor productivity since the 1990s even as its use of prefabrication became more extensive.
  3. Japan uses prefabrication more widely than the US and has experimented with technologies such as automated skyscraper construction, yet has seen virtually no construction productivity improvement since the 1970s.

Related Articles:

  • Five Decades of Decline: U.S. Construction Sector Productivity — U.S. construction labor productivity in 2023 was essentially unchanged from 1948, with real output per worker in 2020 more than 30% below its 1970 level, even as aggregate U.S. labor productivity tripled between 1950 and 2020. Construction accounted for about 4.5% of U.S. GDP in 2024.
  • The Strange and Awful Path of Productivity in the U.S. Construction Sector — U.S. construction sector labor productivity and total factor productivity grew faster than the whole economy in the 1950s and 1960s, then fell, ending 2020 below their 1950 levels while economywide labor productivity and total factor productivity were 290% and 230% higher. Construction labor productivity fell about 1% a year from 1970 to 2020.
  • Why Has Construction Productivity Stagnated? The Role of Land-Use Regulation — Homes built per construction worker boomed after World War II and fell after 1970, a decline the authors link to local land-use regulation that limits project size and keeps construction firms small and less productive. They estimate US residential builders would be roughly 60% more productive if their size distribution matched manufacturing.

AI Summary. South Korea's low fertility rate reflects in part decades of government antinatalist policy: family planning centered on contraceptive distribution is estimated to have averted 1.8m–2.1m births by 1976 at about $103 per prevented birth in today's money. Fertility fell from six children per woman in 1960 to three by 1978.

Phoebe Arslanagic-Little Works In Progress
Date Posted:

Given current fertility trends, every 100 South Koreans today will have 6 great-grandchildren. ROK’s fertility drop began earlier than the global decline, dating from 1961 following a sustained government effort to reduce fertility, which ended in 1994.

Did decades of government birth-control policy drive South Korea's low fertility?

Core argument: South Korea's total fertility rate fell from six children per woman to three between 1960 and 1978, a drop that took 96 years in the UK and 82 years in the US.

South Korea’s fertility crisis sounds similar to that in the rest of the world – just a much more extreme version. But one cause of its low fertility rate is unusual, especially compared to the Western world. This is the legacy of decades of sustained government action to reduce fertility. By 1976, Korea’s fam­ily planning policy, centering on the promotion and distribution of contraceptives, is estimated to have averted between 1.8 million and 2.1 million births. And it did so remarkably cost-effectively, at a cost of just $103 per prevented birth in today’s money. Between 1960 and 1978, South Korea’s total fertility rate fell from six children per woman to three. Comparable drops took 96 years in the UK and 82 in the US. These antinatalist policies survived the end of Park’s dictatorship in 1979. The fertility rate continued to plummet, falling below replacement rate (2.1 children per woman) in 1984. In 1989 the government stopped giving out free contraceptives and relaxed the sterilization drive. By 1990, South Korea was experiencing the consequences of these policies: the average age of the country was now rising, its working age population had begun to fall, and the country now had an imbalanced sex ratio caused by sex-selective abortions. In 1994, the government officially abandoned its population suppression targets.

Takeaways by Macro Roundup® AI

  1. South Korea's total fertility rate fell from six children per woman to three between 1960 and 1978, a drop that took 96 years in the UK and 82 years in the US.
  2. South Korea's family planning policy, centered on promoting and distributing contraceptives, is estimated to have averted 1.8 million to 2.1 million births by 1976, at a cost of $103 per prevented birth in today's money.
  3. South Korea's antinatalist policies outlasted Park's dictatorship, which ended in 1979, with fertility falling below the 2.1 replacement rate in 1984, free contraceptives discontinued in 1989, and population suppression targets abandoned in 1994.

Related Articles:

  • Depopulation Globally and in the Asia-Pacific: The Shape of Things to Come — Population aging and fertility below replacement levels are connected to a revolution in family structure in which families grow smaller, fewer people marry, and voluntary childlessness becomes common across countries. As a result, families become less able to bear the demands that may be placed on them.
  • The Power of Demography and the Fate of East Asia — East Asia's total population has peaked and is projected to decline for at least a generation, with the pace of decline likely accelerating each decade, while US population is projected to keep growing for decades. The author argues such sustained divergence can shift the global balance of power.
  • The Demographic Future of Humanity: Facts and Consequences — The authors estimate the world total fertility rate is around 2.17, below the global replacement rate of roughly 2.21, meaning humanity is already below replacement, with fertility and births falling faster than anticipated. They project middle-income countries' fertility may fall below that of rich countries outside East Asia, with emigration compounding population decline.

AI Summary. Long-term interest rates have drifted higher than short-term rates and oil prices would predict, an unusual pattern during Federal Reserve rate-cutting cycles. Proposed explanations include investor concern about growing Treasury debt issuance or an effective increase in the Federal Reserve's 2% inflation target.

Torsten Sløk Apollo
Date Posted:

U.S. 10-year yields are higher today than at the start of the Fed’s cutting cycle in September 2024. Sløk notes that the “pattern of rising long-term interest rates is highly unusual when we look at the historical reaction during Fed cutting cycles.”

Why are long-term interest rates rising while short-term rates fall?

Core argument: Long-term interest rates began drifting about a year ago above levels predicted by short-term rates and oil prices, a pattern described as highly unusual relative to the historical reaction during Fed cutting cycles.

Why are long-term interest rates going up when the Fed is cutting rates? Is the market worried about growing Treasury issuance, or about a new Fed leadership effectively raising the inflation target from 2% to, say, 4%? The reality is that about a year ago, long-term interest rates started drifting higher than what would have been predicted by short-term interest rates and oil prices. This pattern of rising long-term interest rates is highly unusual when we look at the historical reaction during Fed cutting cycles. The bottom line is that the yield curve continues to steepen, and investors across all asset classes need to think about why.

Takeaways by Macro Roundup® AI

  1. Long-term interest rates began drifting about a year ago above levels predicted by short-term rates and oil prices, a pattern described as highly unusual relative to the historical reaction during Fed cutting cycles.
  2. Two candidate explanations for rising long-term yields while the Fed cuts are market concern over growing Treasury issuance and the prospect of new Fed leadership effectively raising the inflation target from 2% to around 4%.
  3. The yield curve continues to steepen, a development investors across all asset classes need to think about.

Related Articles:

  • Trump’s Interest Rate Obstacle Is Bigger Than Powell — Bloomberg Economics estimates the natural interest rate has climbed from a low of 1.7% in 2012 to about 2.5%, and projects 2.8% by 2030 as baby boomers retire, foreign savings gluts fade, and government borrowing rises — keeping ten-year Treasury yields between 4.5% and 5%.
  • The Race Between Asset Supply and Asset Demand — The authors estimate that demand for assets has outpaced supply by roughly 400% of GDP versus 100%, making rising asset demand the leading contributor to falling interest rates. They project that debt sustainability requires fiscal consolidation exceeding 10% of GDP, though debt could reach 250% of GDP without raising interest rates.
  • Data Update 4 for 2025: Interest Rates, Inflation and Central Banks! — Interest rates are ultimately driven by macroeconomic fundamentals, with an "intrinsic risk-free rate" defined as the sum of the inflation rate and real GDP growth tracking the observed 10-year Treasury bond yield closely. On this view, central banks' influence comes from preferential access to data on those fundamentals, their capacity to alter them, and their credibility.

AI Summary. China generates more than twice as much electricity as the US, and Goldman Sachs projects it will have roughly 400 gigawatts of spare capacity by 2030 — about 3x expected global data-center power demand. Morgan Stanley forecasts US data centers could face a 44-gigawatt electricity shortfall within three years.

Raffaele Huang and Brian Spegele Wall Street Journal
Date Posted:

Chinese datacenters can secure power for ~3 cents a kilowatt-hour using long-term purchase agreements. This is 57–67% less than the 7 to 9 cents a kilowatt-hour that datacenters in northern Virginia pay.

Does China's electricity advantage give it an edge in AI?

Core argument: China generated more than twice as much electricity as the U.S. last year, and its power production rose between 2010 and 2024 by more than the rest of the world combined.

Between 2010 and 2024, [China's] power production increased by more than the rest of the world combined. Last year, China generated more than twice as much electricity as the U.S. Morgan Stanley forecasts that China will spend some $560 billion on grid projects in the five years through 2030, up 45% from the previous five years. Goldman Sachs predicts that by 2030, China will have about 400 gigawatts of spare capacity, about three times the world’s expected data-center power demand at that time. The U.S.-China “electron gap,” as OpenAI now calls it, has become a major preoccupation for American tech leaders. Microsoft Chief Executive Satya Nadella has said his company is worried it won’t have enough power to run the enormous number of chips it is buying. In the next three years, U.S. data centers could face an electricity shortfall of 44 gigawatts, the equivalent of New York state’s summertime capacity, Morgan Stanley has forecast, posing a “daunting challenge” for the nation’s AI ambitions.

Takeaways by Macro Roundup® AI

  1. China generated more than twice as much electricity as the U.S. last year, and its power production rose between 2010 and 2024 by more than the rest of the world combined.
  2. Morgan Stanley forecasts China will spend some $560 billion on grid projects in the five years through 2030, up 45% from the prior five years, while Goldman Sachs predicts about 400 gigawatts of Chinese spare capacity by 2030, roughly three times expected global data-center power demand.
  3. Morgan Stanley forecasts U.S. data centers could face a 44-gigawatt electricity shortfall within three years, equivalent to New York state's summertime capacity, posing a "daunting challenge" for U.S. AI ambitions.

Related Articles:

  • 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.
  • What’s Happening to Wholesale Electricity Prices? — Average US retail electricity prices have risen about 35% since 2020, while average wholesale electricity costs have nearly doubled, with wholesale prices far more volatile. Wholesale location marginal prices consist of energy generation costs (typically 80–90% of the total), transmission congestion, and transmission losses, with a price spike partly driven by a large rise in natural gas prices.
  • Europe’s Green Energy Rush Slashed Emissions—and Crippled the Economy — Europe has cut carbon emissions 30% from 2005 levels versus 17% in the US, while the shift to renewables has helped push electricity prices higher across much of the continent. Average electricity prices for heavy industry in the European Union are roughly 2x those in the US and 50% above China.

AI Summary. A US decision to permit Nvidia H200 sales was shaped by an assessment that Huawei competes with Nvidia more closely than acknowledged, with officials finding Huawei's CloudMatrix 384 platform performed as well as Nvidia's most advanced NVL72 system. Officials also concluded Huawei could produce a few million Ascend 910C accelerators annually.

Edward Ludlow and Maggie Eastland Bloomberg
Date Posted:

One driver of Trump’s decision to allow H200 sales to China was an assessment that Huawei will be able to produce Ascend 910C chips, which were designed to compete with Nvidia products, at scale in 2026.

Did Chinese AI chip progress drive the easing of US export limits?

Core argument: US officials concluded Huawei will be capable of producing a few million Ascend 910C accelerators in 2026, against a June US estimate that the Shenzhen-based company could make just 200,000 Ascend chips this year.

Underpinning the move was an assessment that Huawei can compete far more closely with Nvidia than the US has acknowledged. White House officials focused on a Huawei AI platform known as CloudMatrix 384 that relies on the company’s newer Ascend chips, the person said. Officials found that CloudMatrix 384 performed as well as a similar Nvidia system known as NVL72 that uses the US company’s most advanced Blackwell-design chips, according to the person, who spoke on condition of anonymity to discuss sensitive matters. Adding a sense of urgency, the person said, was a conclusion by US officials that Huawei would be capable in 2026 of producing a few million of its Ascend 910C accelerators, a chip designed specifically to compete with Nvidia’s product line. That compares with a US estimate, given in June, that the Shenzhen-based company would be able to make just 200,000 of the Ascend line this year.

Takeaways by Macro Roundup® AI

  1. US officials concluded Huawei will be capable of producing a few million Ascend 910C accelerators in 2026, against a June US estimate that the Shenzhen-based company could make just 200,000 Ascend chips this year.
  2. White House officials found Huawei's CloudMatrix 384 platform, built on the company's newer Ascend chips, performed as well as Nvidia's NVL72 system, which uses Nvidia's most advanced Blackwell-design chips.
  3. Trump's H200 reprieve for Nvidia rested on an assessment that Huawei can compete far more closely with Nvidia than the US has acknowledged.

Related Articles:

  • Should the US Sell Hopper Chips to China? — The H200 chip is roughly 6x as powerful as the H20, which requires an export license for China and has been approved only in limited quantities. The authors estimate the US would hold a 21–49x advantage in AI compute produced in a coming year absent chip exports to China, shrinking to between 6.7x and 1.2x with unrestricted H200 exports.
  • The Geopolitics of AI: Decoding the New Global Operating System — The performance gap between the top American and Chinese AI models narrowed from 9.3% to 1.7%, with Chinese firms relying on cheaper open-source models—DeepSeek prices usage 20-40x below OpenAI's ChatGPT. This price advantage could win adoption in developing countries and build dependency on Chinese AI ecosystems.
  • China’s DeepSeek Uses Banned Nvidia Chips for AI Model, Report Says — Chinese AI developer DeepSeek used Nvidia Blackwell chips that were smuggled into China through countries where their sale was permitted, according to a report citing unnamed sources. The chips were installed in data centers abroad, then dismantled and shipped to China after clearing inspection by server-equipment companies.

AI Summary. Defense startup Anduril has hit setbacks across several weapons programs, including engine damage to its unmanned Fury jet fighter in a ground test, a 22-acre fire caused by a counterdrone system test, and software that struggled to command unmanned boats. Former employees report its systems in Ukraine proved vulnerable to jamming.

Shelby Holliday, Heather Somerville, Alistair MacDonald and Emily Glazer Wall Street Journal
Date Posted:

Tests in the US and on the battlefield in Ukraine suggest that Anduril, a leading new entrant into the defense industry, is facing challenges in developing and fielding attritable mass weapons systems.

Are defense startups hitting setbacks in developing new weapons systems?

Core argument: A mechanical issue damaged the engine of Anduril's unmanned jet fighter Fury during a summer ground test in California, ahead of a first flight critical to the Air Force.

In California, a mechanical issue damaged the engine in Anduril’s unmanned jet fighter Fury in a ground test over the summer ahead of a critical first flight for the Air Force. In August, a test involving its Anvil counterdrone system caused a 22-acre fire in Oregon. And in the exercises with unmanned boats over the summer off the coast of California, Anduril’s Lattice software struggled to command and control vessels. Anduril’s only real battlefield experience—in Ukraine—has been marred by problems as well, including vulnerability to enemy jamming, according to former employees and others familiar with the systems in Ukraine. Some front-line soldiers of Ukraine’s SBU security service, for instance, found that their Altius loitering drones crashed and failed to hit their targets. The drones were so problematic that they stopped using them in 2024 and haven’t fielded them since, according to people familiar with the matter.

Takeaways by Macro Roundup® AI

  1. A mechanical issue damaged the engine of Anduril's unmanned jet fighter Fury during a summer ground test in California, ahead of a first flight critical to the Air Force.
  2. A test of Anduril's Anvil counterdrone system caused a 22-acre fire in Oregon in August, and the company's Lattice software struggled to command and control unmanned boats in summer exercises off California.
  3. Ukraine's SBU security service stopped using Anduril's Altius loitering drones in 2024 after the drones crashed and failed to hit targets, and has not fielded them since, according to people familiar with the matter.

Related Articles:

  • Drones Are Key to Winning Wars Now. The U.S. Makes Hardly Any — China's DJI accounts for about 70% of commercial drones sold globally, and industry experts estimate its output far exceeds that of any other manufacturer, with capacity in the millions of drones a year. About 500 US companies manufacture drones, producing fewer than 100,000 annually, according to an industry association.
  • Why Every Country Needs To Master The Electric Tech Stack — Drones are the foundation of hard power in the early 21st century, so a country unable to produce them domestically is at a disadvantage. As electric motors overtake combustion engines as the core technology that makes machines move, the same supply chains and production processes used for electronics can also produce cars, drones, robots, and appliances.
  • U.S. Military Is Struggling to Deploy AI Weapons — A U.S. military showcase of autonomous weapons acquired under the Pentagon's Replicator program revealed systems that were not ready for operational use. An unmanned boat suffered a rudder failure and drifted away, an aerial drone launch was delayed over a possible problem with its launch tubes, and onboard software misidentified or failed to identify objects.