Marginal Revolution

Mexico has quietly become a cornerstone of the AI boom, providing 40 per cent of US imports this year of the computer servers that are widely used in the data centres powering artificial intelligence.

Taiwanese manufacturers are rapidly expanding factories in Mexico to assemble servers, which are now the country’s top US export, overtaking autos that dominated its trade for decades.

Mexico is the second-largest provider of enterprise servers to the US so far this year, with sales reaching $46.9bn, behind Taiwan, which sold $53.5bn, though Mexico became the largest on a monthly basis in May.

The phenomenon is helping prop up Mexico’s limp economy, pushing the country’s exports to record levels. Servers and related hardware made up almost one-fifth of the $317bn of goods Mexico exported between January and May, more than double the same period a year earlier. Taiwan, in turn, is now Mexico’s third-largest trading partner, up from eighth place in 2022. Taiwanese companies have spent more than $1.6bn since 2020 on Mexican factories that offer geographic proximity and tariff-free access to US tech giants spending hundreds of billions on data centres.

“Mexico is of tremendous importance to the way this new AI-powered economy is working,” said Jesse Rogers, head of Latin America economics at Moody’s Analytics. “It is really a new frontier of collaboration and even dependency on the Mexican economy when it comes to AI servers.”

The post Mexico (Taiwan) fact of the day appeared first on Marginal REVOLUTION.

Comments

This should be no surprise to anyone. Mexico has been ... by PaulD
In reply to ElsieT. That was – perhaps – true 50 years ... by Engineer
In reply to dearieme. Well we can't get too enthusiastic. ... by Anti-Gnostic
In reply to steve. I've learned that dimwits gotta be ... by rayward
In reply to steve. Its bias is worse than that. It is also ... by Mitch Berkson
Plus 8 more...

Rothko show in Florence

Tyler Cowen — Sat, 01 Aug 2026 17:50:35 +0000

In Museo San Marco there is an ancillary exhibit, placing some smaller, lighter colored Rothkos next to their Fra Angelico frescoes. Those rooms bowled me over, as for one thing they show how much Rothko was a religious painter at heart.

I also was struck by the high proportion of Chinese visitors at this show, even though they are not especially numerous in Florence right now.

The main Rothko show, at Palazzo Strozzi, had an opening room showing Rothkos tied to Renaissance (!) art, mostly with works from early Rothko. The rest of the exhibit had perfect lighting and was the best Rothko display I have seen.

Ken Griffin has very good taste in Rothko. One of his works is here.

Visiting Florence in August is less unbearable than I was expecting. So many other places have become crowded that the situation here does not phase me. Plus the heat is keeping many away? At dinnertime, there is no crush to get into the restaurants.

Moving on shortly, just here for one day!

The post Rothko show in Florence appeared first on Marginal REVOLUTION.

Comments

Hanging out with Ken Griffin to smooth out what's happened with ... by O.B.
In reply to Moral Panic. For one thing, that's a poor ... by Dzhaughn
It was in Florence in late July one year that I noticed that in ... by Hadur
In reply to mkt42. Very interesting….thank you for ... by Paul Selesko
In reply to Paul Selesko. Portland's public schools are not ... by mkt42
Plus 10 more...

Saturday assorted links

Tyler Cowen — Sat, 01 Aug 2026 16:21:09 +0000

1. What do consultants get paid for?

2. Ugly architecture creates coalitions against development.

3. London is (modestly) degentrifying.

4. Is the “no forward guidance” approach failing? (FT)

5. Chinese interview with Christopher Nolan.

6. “social media use and school grades are unrelated for adolescents…meta-analytic evidence is not in support of dramatic claims relating social media use to mischief.” Link here.

The post Saturday assorted links appeared first on Marginal REVOLUTION.

Comments

In reply to Rebes. I was going to say, consultants get paid ... by Anti-Gnostic
#5: Interesting interview. I certainly think Nolan and the ... by M
In reply to Kaz Stanczak. “But clearly social media also ... by JFA
In reply to Brian Villanueva. Also, the meta-analysis being ... by JFA
In reply to Brian Villanueva. Empirical studies are ... by Kaz Stanczak
Plus 10 more...

More math breakthroughs from GPT models

Tyler Cowen — Sat, 01 Aug 2026 10:53:40 +0000

The cost of generating the proofs for all 10 of these breakthroughs combined was under $2,000 at Sol API prices. We’re excited to see what scientists and researchers are able to create with our upcoming Astra models!

That is from Noam Brown, at the link you will find his list of the ten breakthroughs. Here are some views on the import of these.

The post More math breakthroughs from GPT models appeared first on Marginal REVOLUTION.

Comments

In reply to Brent Buckner. One stone to rule them all! by Stickler Boy
In reply to Bill Benzon. That AIs are now solving Erdős ... by Bill Benzon
In reply to Lord Hee Haw. That's a domain where you can ... by M
In reply to RAD. You forgot an entire massive domain of ... by Lord Hee Haw
In reply to white fog gannet. That seems a bit complicated ... by M
Plus 10 more...

Immigration and Macroeconomic Outcomes in OECD Countries

Tyler Cowen — Sat, 01 Aug 2026 06:48:26 +0000

OECD countries experienced declining native population growth and rising net immigration over 1990-2024. We compile a new dataset of net immigration rates to OECD countries from all origins and show that most of the increase came from non-OECD countries and was predominantly high-skilled. Push factors, network effects, and policy indices explain little of the large cross-country heterogeneity in immigration dynamics; unexpected shocks and surges were common. Using local projections and several sources of identifying variation, we then estimate the relationship between immigration and growth in GDP per capita, labor productivity, capital investment, and total factor productivity (TFP). Immigration from non-OECD countries was a significant predictor of GDP per worker growth, primarily through higher investment. High-skilled immigration, in particular, was associated with stronger human capital accumulation, faster TFP growth, and greater capital deepening. Native population growth, by contrast, had no or weakly negative effects on GDP per capita and productivity. These results are consistent with a large literature documenting the positive productivity and growth effects of immigration, especially high-skilled immigration.

That is from a new NBER working paper from Gaetano Basso, Mitali R. Mathur & Giovanni Peri. The natives are just not that impressive!

The post Immigration and Macroeconomic Outcomes in OECD Countries appeared first on Marginal REVOLUTION.

Comments

Does a nation exist only to produce GDP or to preserve a ... by Benjamin Cole
This strikes me as a “wet pavement causes rain” argument. ... by Horhe
There is more to a person (and a family) than their income. ... by Jon B
In reply to Naveen_K. Equally though (and though I understand ... by M
In reply to Kaz Stanczak. Thanks Kaz, many great comments. by Bubbles
Plus 10 more...

Emergent Ventures India, 18th cohort

Tyler Cowen — Sat, 01 Aug 2026 04:24:42 +0000

Adithyan Madhu, 14, an independent researcher, received his grant to study early writing systems, applying computational linguistics, pattern analysis, and structural methods to the Indus Valley script.

Aparajeet Shadangi, 17, received his grant to explore conceptual approaches to cosmic expansion and dark energy while continuing his formal academic preparation in cosmology.

Kevin Gibson, a student developer, received his grant for Crsynk OS, a Python-based operating system integrated with an AI assistant and focused on lightweight user interfaces and accessible computing. He is also developing WiseChat AI and an AI Voice Engine.

Vaishnavi Dilip, 13, received her grant to identify fire-resistant plants native to Indian ecosystems and plant them strategically as natural firebreaks in fire-prone areas.

Prince T. Philip, an independent cybersecurity researcher and product architect focused on high-impact, future-facing risks in modern software systems, received his grant to build a security audit tool that surfaces vulnerabilities and design assumptions missed by traditional scanners.

Michaela Cross, a writer and content creator, received her grant for Rose Chasm, a video project on the history of India and its relationship to America and the wider world.

Vivekanand Bhat and Rijesh Panicker received their grant for Pathika, to create immersive, augmented, multilingual onsite guided tours for heritage monuments across India.

Devanarayan, 15, received his grant for Open Horizon Observatory, a small open-access observing setup that produces and freely shares real astronomical data. He is developing low-cost workflows for time-series photometry and open data sharing for students and the public.

Roshan Kumar Gupta, 19, a computer science undergraduate, received his grant for TARS, a deterministic, physics-aware machine-learning pipeline for exoplanet discovery using TESS data. He aims to reduce false positives and identify physically credible habitable-world candidates in noisy observations.

Yajan Adhikari, 19, a materials researcher, received his grant to build an operating system for digitizing and scaling the management of recyclable materials.

Pranet Khetan, 17, received his grant to develop a novel, low-cost architecture for refreshable tactile graphics supporting STEM education for visually impaired students.

Karthik, 18, a space engineer, received his grant to research the use of remote sensing in agricultural insurance.

Alexander Hogeveen Rutter, an energy expert, received his grant for analysis and advocacy of least-cost resource adequacy planning and other market-based measures in India’s power sector.

Akshit Arora, an Engineering Physics student at IIT Delhi, and Bhavey Sehgal, a BITS Pilani student, received their grant for NeuroBridge. It uses eye tracking, computer vision, and AI to predict intent, giving people with ALS fast, confirmation-based speech instead of slow letter-by-letter eye typing, at a fraction of existing device costs.

Cartoonist Sumit Kumar and Vivekananda Roy Ghatak received their grant to advance their animated series Aaapki Poojita and produce short, animated sketches and redesign and relaunch the famous Bakarmax webtoon platform.

Progyan Das, 24, received his grant to develop fault-tolerant systems capable of using consumer devices for large-scale foundation model training.

Yash Chitte, an MBA student, received his grant to build a wall-climbing robot that autonomously cleans glass facades.

Khushi Jain, 23, an aerospace engineer, received her grant for NadirX Space, to build space-grade test infrastructure and hardware-in-the-loop systems that enable reliable testing of advanced space technologies and support future interplanetary missions.

Rishit Kapoor, an AI/ML researcher and engineer, received his grant for Periscopic Labs, which turns systematic reviews into living, AI-native infrastructure that keeps scientific claims traceable, verifiable, and continuously updated.

Koushik Das received his grant to build an explainable clinical decision-support system that helps primary-care doctors interpret complete blood count reports and detect anemia patterns more quickly and safely.

Shashank Sanjay, a medical student, received his grant to develop an affordable coronary stent.

Abrar Shahid, a systems designer using AI to reduce accessibility gaps, received his grant for Project Lumen AI, to build an AI translation layer connecting external eye photographs with retinal fundus scans and make eye-disease diagnosis easier.

Those unfamiliar with Emergent Ventures can learn more here and here. The EV India announcement is here. More about the winners of EV India second, third, fourth, fifth, sixth, seventh, eighth, ninth, tenth, eleventh, twelfth, thirteenth, fourteenth, and fifteenth , sixteenth, and seventeenth cohorts. To apply for EV India, use the EV application, click the “Apply Now” button and select India from the “My Project Will Affect” drop-down menu.

And here is Nabeel’s AI engine for other EV winners. Here are the other EV cohorts.

If you are interested in supporting the India tranche of Emergent Ventures, please write to me or to Shruti at srajagopalan@mercatus.gmu.edu.

The post Emergent Ventures India, 18th cohort appeared first on Marginal REVOLUTION.

Comments

In reply to Florida Man. Not one western name on that list, ... by Florida Man
So an Indian can get the EV the normal way, plus through their ... by Florida Man
In reply to But NSF is broken!. I too would be interested to ... by Marie
In reply to Rahul. It does seem strange to me that MR spends ... by But NSF is broken!
In reply to Bob Ewing. I know nothing about the subjects most ... by Dismalist
Plus 4 more...

Redefining Global Health in the 21st Century

Tyler Cowen — Fri, 31 Jul 2026 17:54:16 +0000

While donor-driven programs undoubtedly saved millions of lives, they also created unintended distortions in national health priorities. Many governments in sub-Saharan Africa and parts of Asia actually scaled back domestic health investments, as donor program filled key gaps in HIV/AIDS, maternal and child health, and infectious disease control. In some cases, domestic health budgets shrank in real terms, even as external funding increased. This phenomenon, often referred to as “fiscal substitution,” led to national health systems that were heavily donor-dependent, externally managed, and vulnerable to funding shocks. Notably, this reliance emerged despite countries pledged to allocate at least 15% of their national budgets to health. More than two decades later, only a handful has met this target.

That is from the new and useful book by Michael John Alastair Reid and Eric Paul Goosby. But $150 for a not so thick volume!? If the authors believe in aid, as they should, perhaps they could consider giving this book away for free rather than turning it over to Elsevier…?

The post *Redefining Global Health in the 21st Century* appeared first on Marginal REVOLUTION.

Comments

In reply to Moo cow. Ezra Klein in 2025. Like, seriously? ... by Dispatching from the Global South
In reply to Anonymous Bob. In a conference I attended as ... by Dispatching from the Global South
“While donor-driven programs undoubtedly saved millions of ... by German_doctor
In reply to Ricardo. But I think the point is that if PEPFAR ... by engineer with only 3 kids
In reply to Vivian Darkbloom. > If you don't demand that the ... by engineer with only 3 kids
Plus 10 more...

Friday assorted links

Tyler Cowen — Fri, 31 Jul 2026 16:05:45 +0000

1. Tokenizing the value from compute?

2. Will they ruin the look of Dulles airport?

3. Pathways into homelessness.

4. Defensible valuations for AI companies.

5. Further reasons why very rapid economic growth is very difficult to achieve.

6. Over 20,000 precolonial earthworks in the Southwest Amazonia.

The post Friday assorted links appeared first on Marginal REVOLUTION.

Comments

In reply to Kaleberg. Fair. I was thinking about Western ... by Peppa
About one quarter of kids aging out of the foster care system ... by So...
In reply to Hazel Meade. It's not nuts, but I think the ... by Lord Action
In reply to Lord Action. Maybe what we need is something like ... by Hazel Meade
In reply to Kaleberg. If you haven't seen it – you might ... by Mitchell
Plus 10 more...

An unprecedented weapon?

Tyler Cowen — Fri, 31 Jul 2026 07:03:26 +0000

I wonder if England had a “pause” movement back then:

Trebuchets reached the peak of their development in medieval times, probably the most famous example being Edward I’s massive machine, Warwolf. This terrifying device was over sixty feet (18 meters) high and was capable of throwing a three-hundred-pound (140-kilogram) rock several hundred yards, projecting some five hundred kilojoules of energy, a thousand times the power of Roman ballistas and five thousand times as powerful as a single person. Edward used it just once, in 1304 during his wars of conquest against the Scots, to attack Stirling Castle, the last stronghold of the Scottish armies. The defenders were so petrified by the prospect of the attack that they sued for peace before Warwolf could be used, but Edward refused to let them surrender until he had tested the machine; in four days it totally destroyed one of the castle’s curtain walls.

That is from the new and interesting The Powerful Primate: How Controlling Energy Enabled us to Build Civilization, by Roland Ennos.

The post An unprecedented weapon? appeared first on Marginal REVOLUTION.

Comments

In reply to Andy B. “what does and does not usefully ... by Wayfaring Stranger
In reply to Hazel Meade. Curiously, war can also slow down ... by mdap
In reply to Wayfaring Stranger. Mr. Stranger, that's now ... by Andy B
In reply to Bill Allen. I'll recommend Gillian Bradshaw's ... by Kaleberg
In reply to Quantitative Sneezing. Isn't that 1.21 ... by Kaleberg
Plus 10 more...

The wisdom of Eli Dourado

Tyler Cowen — Fri, 31 Jul 2026 05:02:53 +0000

How do “international efforts” work? Unlike most of the signatories of the letter, I have been a State Department advisor and have participated in multiple treaty negotiations. I have also been a member of technical committees for UN technical agencies.

The international sector is unbelievably dysfunctional. Every single treaty or international agreement is an opportunity for every participant to manipulate the much broader policy environment. Often the participants don’t care about the object of the agreement, and are instead trying to use the agreement as leverage for something else. I have seen countries use their limited leverage in multilateral agreements to try to kneecap American industry, get around sanctions, create a pretense of international justification for domestic illiberalism, or steer technology in an authoritarian direction.

The damage from this dysfunction is limited by the fact that there are zero major countries (and not that many smaller countries) who will actually bind themselves in meaningful ways that they don’t narrowly want. If a previously signed treaty turns out to be inconvenient, it is often subtly ignored or reinterpreted.

In addition, treaty delegations working on industrial issues generally reflect the full spectrum of special interests involved in an issue. A US AI treaty delegation might be led by a State Department ambassador, but he will be advised by representatives of the interagency, major labs, other major tech companies, tech investors, civil society, etc.

“International efforts,” therefore, is not a reassuring answer to the governance problem; it is the name of another enormous, unresolved governance problem.

Here is the full tweet.

The post The wisdom of Eli Dourado appeared first on Marginal REVOLUTION.

Comments

The wisdom of David Friedman: “Markets, property rights, ... by Nathan Braun
In reply to Keith. Yes, you could make the choice to have no ... by ElsieT
In reply to ElsieT. Like literally you missed the context. ... by Keith
In reply to Wayfaring Stranger. If it's few countries, we ... by r_r
In reply to Keith. Please illuminate us on what the choices ... by ElsieT
Plus 10 more...

The Anatomy of Cyber Risk

Tyler Cowen — Thu, 30 Jul 2026 18:06:08 +0000

This paper uses computational linguistics to introduce a novel measure of firm-level cyber risk exposure based on the quarterly earnings calls of listed firms. Our measure covers more than 14,000 firms from over 90 countries between 2003 and 2025. The measure is validated using human auditors and a large language model. We show that cyber risk exposure affects stock returns and profits, is priced in the options market, predicts actual cyberattacks, and propagates from firm to sector level. Back-of-the-envelope estimates suggest that the global cost of cyber risk exposure is around $1 trillion per year.

That is from a recent paper by Rustam Jamilov, Hélène Rey, and Ahmed Tahoun, forthcoming in the Journal of Finance.

That the market prices cyber risk is a point we should not so readily forget. Is anyone arguing that recent market declines are due to higher perceived cyber risk? Does the distribution of these declines match that hypothesis? (Are we possibly seeing the opposite of what that hypothesis might predict, namely that those companies with the best potential defenses are suffering the most?) If the cyber risk cost per year had been about $1 trillion a year, what do we think it is now? Surely members of the rationality community will defer to the scientific methods of investigating these claims…

The post The Anatomy of Cyber Risk appeared first on Marginal REVOLUTION.