The 9 capacity gaps Belton & Zenick flag

I use slightly different descriptions below (with help from ChatGPT 5 and Grok). The 10th is the meta-capability: general purpose metal-cutting machine tools. 🤘

Let’s dive into the weeds...

Thank you for subscribing. This is public content. If you are not a member, please sign up here.

Proposals to Rebuild Capacity Gaps

1) Special-Purpose Industrial Machinery (HS 8479.10)

What it is
Non-standard machinery for public works, mining, and factory automation; often custom platforms integrating sensors, drives, and controls.

Why it matters
Seeds automation across fabs, EV/battery lines, and energy projects; co-export with a dozen frontier goods.

What blocks it
Lumpy demand; long payback on demo lines; vendor-qualification risk for primes.

12–36-month actions (Owner × Lever × Target)

• DOC + DoD — Demand aggregation: 3-year framework contracts for standardized sub-platforms; include reference factory buys (10–15 lines).

• Treasury/DOE — Targeted Section 48C (Advanced Energy Project) ITC: up to 30% credit (6% base / 30% with prevailing wage & apprenticeship) for domestic assembly and test cells tied to U.S. content in motion control.

• NIST MEP — Vendor-qualification sprints: Fund 50 SME suppliers through OEM qual and ISO upgrades.

Risks & mitigations
Over-customization → require modular SKUs; export retaliation → prioritize non-substitutable niches.

Metrics (by Month 24)
10 domestic lines installed; median lead time to U.S. buyers ↓ 20%; import share in targeted SKUs ↓ 10 pts.

2) Precision Metrology & Alignment Instruments (HS 9017.10; adj. 9031)

What it is
Precision geometric instruments, optical profilers, interferometers, alignment systems used in aerospace and lithography.

Why it matters
Nanometer-scale accuracy underpins chipmaking, optics, and satellite/aero assemblies; metrology is the rate-limiter for yield ramp.

What blocks it
Thin domestic cluster; long standards/certification cycles; shortage of metrology engineers and optics/tooling talent.

12–36-month actions (Owner × Lever × Target)

• NIST — Round-robin calibration program1: Subsidize inter-lab comparisons and national traceability; publish open qualification datasets.

• USCIS/State — Fast-track visas: Dedicated O‑1/H‑1B lane for metrology/optics engineers; employer attestations via MEP.

• DoD/NSF — SBIR/STTR2 set-aside: Alignment sensors and in‑situ metrology for advanced packaging; 50 Phase IIs with matching from primes.

• State/MEP — Cluster grants: Support two metrology hubs co-located with optics and precision machining shops.

Risks & mitigations
Fragmented demand → create pooled buying via reference fabs; IP leakage → export-control aligned NDAs and onshore test.

Metrics (24–36 mo)
Lead times ↓ 25%; 200 new metrology roles filled; 100+ SKUs with NIST-traceable certificates made in U.S.

3) Organo‑Sulfur Specialty Chemicals (HS 2930.90)

What it is
Specialty organo-sulfur intermediates for photoresists, EV battery additives, and pharma.

Why it matters
Feedstocks for semiconductors, clean energy, and critical drugs; chokepoint chemicals can idle billion-dollar lines.

What blocks it
Site permitting, waste treatment, hazmat logistics; off‑take uncertainty; need for rail/steam/utilities co-location.

12–36-month actions (Owner × Lever × Target)

• DOE LPO — Project guarantees: Support on‑site waste‑treatment and utilities at two chemical parks; require domestic offtake.

• GSA/DoD/DOE — Long-term offtakes: 5–7‑year contracts for defined intermediates (photoresist/battery grades).

• EPA — Green-chemistry grants: Process intensification & solvent recovery; expedite permits for low-emission designs.

• States — Industrial sites: Pre‑permitted parcels with rail/steam; tax increment for onsite treatment CAPEX.

Risks & mitigations
Community opposition → transparent monitoring and escrowed remediation funds; demand risk → multi-agency offtake pool.

Metrics (36 mo)
Domestic capacity +30 kt/yr in target SKUs; >60% U.S.‑sourced binders/solvents for selected photoresists; incident rate below industry benchmarks.

Share

4) Cermet Cutting Inserts & Tips (HS 8209.00)

What it is
Cermet inserts for high-speed machining of nickel superalloys and titanium.

Why it matters
Drives precision and tool life in aerospace/energy parts; uptime and edge retention set cell economics.

What blocks it
Cobalt/nickel powder supply; sintering furnace lead times; know‑how concentrated abroad.

12–36-month actions (Owner × Lever × Target)

• Treasury — 100% bonus depreciation: For domestic powder‑metallurgy lines and sintering furnaces meeting yield KPIs.

• Congress/Treasury — 45X‑style production credit: Per‑kg credit for cermet powder made in U.S.

• DARPA/DOE — Materials program: Cold‑spray/coatings to extend tool life; public testbeds with OEMs.

Risks & mitigations
Powder pricing volatility → indexed offtake; environmental → closed-loop powder recovery standards.

Metrics (24–36 mo)
Tool life +25%; domestic share of aerospace‑grade inserts +15 pts; two new sintering lines commissioned.

5) Carbide & HSS Machine Blades (HS 8208.10)

What it is
Precision blades/knives for milling, slitting, and grinding; core to medical, optics, and packaging lines.

Why it matters
Tolerance stack drives yield in med‑device and optics; downtime costs cascade across production cells.

What blocks it
Aging grinder base, metrology gaps, and a shortage of skilled grinder/operators.

12–36-month actions (Owner × Lever × Target)

• NIST MEP + States — Modernization vouchers: $250k–$1M per shop for in‑line metrology, adaptive controls, and coolant systems.

• Treasury — Temporary 100% bonus depreciation: For grinder retrofits meeting energy/yield KPIs (mirrors 179D logic).

• Community Colleges + OEMs — Grinder academies: 12‑month paid programs with OEM-provided machines; 500 graduates/year.

Risks & mitigations
Quality drift → vouchers tied to SPC adoption and external calibration audits.

Metrics (24–36 mo)
Scrap rates ↓ 25%; domestic output +20%; 300+ new certified grinder techs/year.

6) Forklifts & Material‑Handling Equipment (HS 8427.90)

What it is
Lift trucks, AGVs, batteries, and safety subsystems used in every fab, warehouse, and shipyard.

Why it matters
It’s the transport/logistics layer; delays ripple through construction and fab ramp.

What blocks it
Thin final‑assembly footprint, local battery packs, and safety‑certified electronics BOMs.

12–36-month actions (Owner × Lever × Target)

• GSA + USPS + DoD — Federal fleet pull‑through: Commit to U.S.-assembled lift trucks; publish a 3‑year demand schedule.

• DOE LPO + IRA domestic‑content adders: Tie battery subsidies to U.S. pack assembly for MHE; supply‑chain guarantees for motors/controllers.

• OSHA + NIST — Raise standards, don’t “preference”: Update safety rules to require advanced sensing/telematics; fund U.S. vendors to certify.

Risks & mitigations
Price delta vs imports → TCO pilots (charger/network + uptime); workforce → OEM‑run technician training.

Metrics (24 mo)
30% of federal MHE purchases U.S.-assembled; 2 new pack plants qualified; downtime/incident rates ↓ 15%.

7) Roll‑to‑Roll Patterning & Registration (HS 8442.30; adj. 8443)

What it is
Pre‑press/phototypesetting lineage tools for flexible electronics and security printing—high-precision web handling, patterning, and registration.

Why it matters
Enables flexible displays, printed sensors, security features; a bridge between lab recipes and scaled production.

What blocks it
Gaps in pilot‑line access; fragmented IP across web handling, inks, and curing; few domestic OEMs.

12–36-month actions (Owner × Lever × Target)

• NSF Manufacturing USA — New node: Shared pilot coat/dry/inspect lines with cleanroom access; 100 SME runs/year.

• DoD/Treasury — Tooling ITC: 20% credit for precision web-handling platforms meeting registration specs.

• SBA — SBIR set‑aside: Retrofitting legacy printers for functional layers; 40 Phase IIs.

Risks & mitigations
Under‑utilization → anchor tenants (defense/security printing); IP friction → standardized test licenses.

Metrics (36 mo)
50 qualified process “recipes”; uptime ≥ 85%; 10 U.S. OEM platforms in production.

8) High‑Performance Printing Inks (HS 3215.90; adj. pigments 3206; resins 3909/3911)

What it is
Functional inks for sensors and additive electronics; depends on specialty pigments/resins.

Why it matters
Critical to wearables, IoT, and printed electronics; supply determines device yield and reliability.

What blocks it
Pilot‑scale bottlenecks; dependence on imported pigments/resins; EHS compliance costs.

12–36-month actions (Owner × Lever × Target)

• UK‑Catapult‑style pilot lines (DOC/NSF): Shared mixing/dispersion/print test lines; fee-for-service for SMEs.

• Grant priority — Domestic inputs: Scoring boost for plants using U.S. pigments/resins; small grants to reshore key pigments.

• Defense/NIH — Targeted offtake: Flexible sensors for training gear and medical wearables; multi‑year buys.

Risks & mitigations
Spec drift across batches → lot certification & digital batch passports; environmental → solvent recovery incentives.

Metrics (24–36 mo)
25 new U.S. ink SKUs qualified; domestic input share >50% for pilot lines; yield variance ↓ 30%.

9) Industrial Valves & Precision Parts (HS 8481.90; adj. forgings 7325/7326; seals 4016)

What it is
Valves and precision components for hydrogen, CCUS, LNG, and semi fabs.

Why it matters
Capex schedule keeper for energy/semiconductor projects; certification gates project risk.

What blocks it
ASME/API certification backlog; forging/casting and elastomer bottlenecks; working‑capital strain for long-cycle orders.

12–36-month actions (Owner × Lever × Target)

• ASME/API + DOE — Accelerated cert program: Funded test capacity and fast‑track audits for U.S. shops.

• DOE LPO — Buy America clauses: Tie LPO loan guarantees to U.S.-made valves/parts where feasible.

• NIST MEP — Die‑shop vouchers: Modernize forging/casting die makers and QA; supplier development for elastomer seals.

Risks & mitigations
Cost spikes → multi‑year pricing with escalation clauses; QC risk → third‑party inspection pools.

Metrics (24–36 mo)
Lead times ↓ 20%; 500 additional certified SKUs; domestic share in DOE‑backed projects +15 pts.

10) General‑Purpose Metal‑Cutting Machine Tools (HS 8457.10/.30; adj. 8456/8459/8460)

What it is
Five‑axis machining centers, EDM/laser, precision grinders—the meta‑capability that makes other tools.

Why it matters
Governs ~30% of frontier goods; concentrates know‑how in fixtures, metrology, motion control.

What blocks it
Export‑finance gap vs peers; demo‑line CAPEX; fragmented SME demand; thin grinder/toolmaker pipeline.

12–36-month actions (Owner × Lever × Target)

• Ex‑Im + DFC — Export scale window: Facility for machine‑tool OEMs; blended finance for overseas buyers of U.S. five‑axis/EDM.

• Treasury/DOE — Targeted 48C: ITC for ultra‑precision platforms tied to kWh/part efficiency.

• DoD/DOC — Reference factories: Two lines each for aerospace/medical demo parts with multi‑year parts orders.

• MEP + Colleges — Tool/grind academies: Paid residencies; 500 graduates/yr by Year 3; fast‑track visas for master grinders.

Risks & mitigations
Commodity creep → restrict to ultra‑precision niches; energy draw → efficiency‑linked credits.

Metrics (36 mo)
10 new domestic lines; 20% shorter lead times; U.S. share of domestic five‑axis purchases +10–15 pts.

If you read this far, thank you—we clearly have a shared interest. Please consider hitting the like button, it helps others discover us, or consider subscribing to a paid tier to support my work. Comments are only open to premium members, but feel free to DM me with any thoughts.

Subscribe now

A few years ago, I visited a U.S. injection molding facility and asked a simple question: “Can you make these parts?” They said yes. Then I asked if we could also make the molds there — the high-precision tooling that enables the molding process. Their answer: “Oh no, we don’t make molds here. We send the CAD files to China. They make the molds and send them back.”

When I told them I wanted to keep the work, and the intellectual property, in the U.S., they just said, “Good luck.”

That moment snapped everything into focus. We’re no longer the nation that builds the machines. We’ve become the operators of someone else’s machines. Once, America exported intelligence and capability — the “hard stuff,” like tool-and-die making — and let others do the routine labor. Now the reverse is true: the advanced capabilities are offshore. We push the buttons.
— Jeremy Fielding1

This story comes from a video by engineer Jeremy Fielding, where he describes the moment he realized just how hollowed out America’s tooling capacity had become.

Tim Cook, Apple’s CEO, once put it bluntly. Asked why Apple builds in China, he replied:

“The products we make require really advanced tooling. The precision... is state-of-the-art. In the U.S., you could have a meeting of tooling engineers and I’m not sure you could fill the room. In China, you could fill multiple football fields.”

We’ve lost something foundational. But can we measure what exactly?

Thank you for subscribing. This is premium member content. Occasional forwarding is okay. If you are not a member, please sign up here.

A Framework for Rebuilding

A new paper by Keith Belton and Joseph Zenick offers one way to do just that. Their approach uses economic complexity theory (ECT) to guide industrial policy with data rather than guesswork. They focus on three key metrics:

1. Revealed Comparative Advantage (RCA) – Are we globally competitive at producing this good?

2. Revealed Comparative Dependence (RCD) – Are we overly reliant on imports for this good?

3. Relatedness – How closely connected is this good to what we already know how to make?

Read more

“Now, boys, we have got her done, let’s start her up and see why she doesn’t work.”1

That wasn’t a joke. It was a method.

Fritz helped define the spirit of American manufacturing: build something, test it, learn from what breaks, and improve it. It was how the steel industry was born. It’s how modern aviation took off. It’s how SpaceX developed rockets. And oddly enough—it’s not far from how China built its industrial strength.

As economist Arthur Kroeber put it in his recent interview with :

“In terms of how they got here... this was not the result of some carefully crafted master plan... It was a lot of groping and a lot of random stuff... Then through this random process, it all coalesced in the end... A lot of it was essentially a random walk where a lot of things that were enabled then started to interact with each other... They got a few big directional things correct. They stuck with them... They were also willing to adapt when things were obviously not working.” — Arthur Kroeber on the Dwarkesh Podcast

That’s not inefficient—it’s emergent. It’s industrial evolution, not industrial design.

In modern language, you could say China seeded its industrial commons: creating supply chains, clusters, and know-how by trying lots of things, tolerating misfires, and doubling down where momentum appeared. The results couldn’t have been pre-planned—but they were made possible by persistent experimentation.

Thank you for subscribing. This is premium member content. Occasional forwarding is okay. If you are not a member, please sign up here.

Read more about the industrial commons:

Relearning How to Build

China tolerates “inefficiencies”

“They are willing to lose huge amounts of money for a few of their bets to pay off: at China’s scale, effectiveness matters more than efficiency.”
— Arthur Kroeber

This is a deep contrast with the Western mindset. In the U.S., we gave one loan to one solar company—Solyndra—and when it failed, declared industrial policy a bust. Meanwhile, China was subsidzing every step of the solar supply chain, tolerating years of losses across dozens of firms, until they controlled the entire sector.

Read more

No other Western company has invested more heavily in Chinese manufacturing than Apple—about $55 billion a year. And no company has helped train more Chinese workers: estimates put the number around 26 million. Apple doesn’t just manufacture in China. It helped build China’s industrial capacity.

Patrick McGee's book, Apple in China, paints the picture. It's a brilliant investigative work. It's *not* blessed by Apple, which raises its truthiness value to me. It's a complicated story and Patrick comes at it from many angles. I highly recommend it. Here’s Patrick’s X account.

Today I'm going to explore what the book says about the industrial commons in China, and lack there of elsewhere. The strengths of nurturing an industrial commons come through, and maybe we can find some lessons for America. Let's dive in.

Thank you for subscribing. Occasional forwarding is okay. If you are not a member, please sign up here.

1. China’s Industrial Commons: Deep, Dense, and Responsive

Dense supplier networks:

“The Foxconn hubs in China are surrounded by hundreds of sub-suppliers all ready to compete for the next major order.”

“If Foxconn needs to install sonic welders... it can call up any number of firms to run the line and hire the labor.”

Ecosystem depth:

“What China offers... is not simply labor, but an entire ecosystem of processes developed over more than two decades.”

Localized manufacturing clusters:

Apple seeks “next-door” suppliers in India—“industrial clusters comprising operations from a diverse set of vendors”—but these clusters are not yet comparable to China’s.

Apple benefited from the proximity of suppliers, if they needed a part cut at a different length, the change request could be given and tooling reconfigured the same day. Much like America had in its industrial clusters as depicted in Freedom’s Forge.

Technological integration:

“Apple engineers were the general managers for this manufacturing supply base. But there wasn’t any way to do it, anywhere, but in China.”

“If you’re shipping a million of anything a day, there are only a handful of companies on the planet that can do it, and Apple found itself in that place with many technologies.”

2. The Manufacturing Feedback Loop

A manufacturing process is highly iterative before scaling, especially for Apple, where millions of iPhones would be produced before launch date. Once the right process and procedure is discovered, it’s scaled up. China’s industrial commons grew to support exactly this kind of quick scale up.

Iteration speed and agility:

There was a bunch of stuff that we at Apple were very used to doing, that just didn't work anymore. Like, we're very fond of making custom fastening hardware, custom screws—you know, little nuts and stuff like that. Well, in China, if you're building and it's like, "Oh, shit, the screw is too short," and like, "I need a longer one," you call someone on a cellphone and 1,000 are at the factory tomorrow. That was not a thing in Texas. It would take two months. It was absurd.

“All the engineers are so frustrated with us because we don’t move at ‘China speed.’” - a Flextronics worker in Texas, "described as an overweight fifty-year-old white guy."

US production experience WITHOUT a local Industrial Commons

Apple’s industrial commons was, and remains, in China and that caused countless issues when Cook placated Trump by building the Mac Pro in Texas.

“Mac Pros were indeed shipped from Texas... We flew people from China to get it fixed... People working for Foxconn.”

Even working with local [U.S.] supplier Texas Instruments was difficult, because the semiconductor maker was producing parts overseas, testing the parts overseas, and then reimporting them.

Grove's Innovation <-> Manufacturing connection

Grove’s lamentation: “Without scaling, we don’t just lose jobs—we lose our hold on new technologies. Losing the ability to scale will ultimately damage our capacity to innovate.”

"Just a few years earlier, Intel cofounder Andy Grove had warned that without manufacturing, America might lose the ability to innovate. Certainly that was the case for any number of hardware companies that sent blueprints off to Asia for someone else to turn into gleaming products. In Apple's case, Grove's worry was only partially true, but in revealing ways. Apple still knew how to manufacture; the problem was that it couldn't execute these plans without China."

I’ve mentioned Grove’s lamentation here, here, here and here.

3. The Limits of Rebuilding Elsewhere

India’s slower progress:

“India is taking on iPhone orders at one-tenth the rate China did a decade earlier.”

“Apple’s India operations are only in the early stages... and it’s likely to take a minimum of five to ten years.”

Troubled U.S. reshoring:

“Rather than representing some milestone, the [Texas] factory had been demonstrating just how difficult it was to make computers in America.”

“The worst project I ever worked on at Apple was bringing up Flextronics for the Mac Pro in Texas.”

Bonus: Risks & “Benefit” in China

Compliance costs:

“You don’t get to do business in China today without doing exactly what the Chinese government wants you to do. Period. No one is immune.”

Apple’s survival strategy:

"Politically, Cook's move was a masterstroke. Apple survived a trade war between the United States and China by gaining exemptions on tariffs from the Trump administration. And even though Washington kneecapped Huawei, Beijing didn't retaliate, because Cook had already made it clear just how much Apple was investing in the country, raising quality standards across its electronics supply chain."

Contrast with democratic systems (the “Benefit”):

Samsung operates at a disadvantage because South Korea is a democracy...

"In her view, Samsung operates at a considerable business disadvantage because South Korea is a democracy with NGOs, trade unions, and a vibrant press that, for instance, has interviewed grieving parents after some workers developed leukemia from working in factories. "Apple doesn't have any of that possible pressure," White says. "Apple actually has a government [China] that prevents all of those key stakeholders in society from writing an article or appearing on television. They can't even protest."

Summary: What This Says About the Industrial Commons

China’s industrial commons is a deeply integrated ecosystem of skills, tools, suppliers, and institutional knowledge. It thrives on:

• Proximity-based collaboration

• Speed of iteration

• Technically trained labor at scale

• A dense web of subcontractors and specialty firms

These aren’t things you can buy off the shelf. They grow out of decades of cumulative investment—public and private. That’s why Apple can’t just move its supply chain to Texas or Tamil Nadu. It’s not just about cost. It’s about capacity.

Lessons for America

If the U.S. wants to rebuild an industrial commons, here are a few takeaways:

• Make it easy. Apple built its industrial base in China because it was easy. Foxconn knew how to goad local governments into supplying land and labor for their operations. Apple committed significant capital and resources to new production lines: training workers and developing new technologies. American firms and government alike need to make bold bets that justify new industrial investment.

• Co-location matters. Speed, agility, and innovation all thrive on proximity. Rebuilding manufacturing means rebuilding clusters—of suppliers, tools, testers, and talent—not just isolated factories.

• Technical training is infrastructure. 26 million trained workers didn’t appear overnight. China’s vocational and technical pipeline deserves as much credit as its subsidies. The U.S. must re-invest in technical education at scale.

• Time horizons must be long. China’s commons took 20+ years to build. A few tax credits and ribbon-cuttings won’t cut it. Industrial strategy is generational work.

If you read this far, thank you—we clearly have a shared interest. Please consider hitting the like button, it helps others discover us, or consider subscribing to a paid tier to support my work. Comments are only open to premium members, but feel free to DM me with any thoughts.

Subscribe now

—Goethe

What does Goethe mean here? To me, it’s a recognition that beliefs—often formed in a world that no longer exists—can blind us to what’s happening right in front of us. We assume we’re seeing reality clearly, but more often we’re filtering it through outdated models and assumptions.

To see the present clearly, imagination is essential. That sounds counterintuitive. But the world doesn’t always announce when it changes. Reality shifts slowly, then all at once. If you’re not actively updating your mental map, you’ll miss it.

It’s like the parable of the blind men and the elephant: each touches a different part and jumps to a conclusion ("It’s a tree!" "It’s a snake!") without grasping the whole. They weren’t lacking data. They were lacking imagination.

William Gibson put it this way: “The future is already here—it’s just unevenly distributed.” That’s not just about sci-fi and tech startups. It applies to every domain where new realities are quietly taking hold before the rest of the world catches up.

The adoption curve of innovation is a helpful visual here. Most people don’t engage with new realities until they’re in the majority. But some live in the future early because they can see it or can build it. They're not delusional. They're imaginative about reality.

Which brings us to American manufacturing—and to people like Chris Power.

Thank you for subscribing. Occasional forwarding is okay. If you are not a member, please sign up here.

Meet Someone Living in the Future

Chris Power, founder and CEO of Hadrian, isn’t predicting the future of manufacturing—he’s already building it. And what he’s building should feel shocking. Not because it’s sci-fi, but because it already exists, and most people still think it's impossible.

Here’s the punchline: Hadrian can manufacture in the U.S. at only a ~30% premium to China—today. That number is falling. Soon, they expect to reach parity. That’s not theory. That’s happening.

“In some cutting-edge sectors, U.S. factories like Hadrian are only about 30% more expensive than China now — a major technological leap forward, and we could reach parity or better soon.”

They’re achieving this by reimagining what factory work looks like. Forget the smokestacks and deafening shop floors of the past. This is something else entirely.

“Factories like SpaceX, Hadrian, Anduril — those are awesome jobs… The mindset just has to shift.”

Hadrian uses automation and technology to give their workers “Iron Man suits”—not literally, but close. With robotics, software, and high-leverage systems, one person can now run ten machines. The goal isn’t to eliminate workers—it’s to multiply their impact.

“Unless we give the American workforce an ‘Iron Man suit’... we don’t have the skilled labor to reindustrialize.”

This is the missing piece in most discussions about reshoring: productivity. In an age where labor is scarce, especially for defense-linked manufacturing, you can’t just bring back old jobs. You have to build new ones—faster to train, easier to scale, more rewarding to do.

“These jobs can pay well. They can offer great insurance. At Hadrian, we give everyone equity in the company.”

And if you think the cost gap with China is just about wages, think again. Power makes the case that China’s cost advantage is artificially inflated by massive subsidies—on materials, energy, capital, and exports.

“Because China operates with looser environmental regulations and offers energy subsidies, the cost of raw materials there can be 10 to 20 times cheaper than in the U.S.”

“Chinese exporters get export rebates — they might get 30% to 40% of the sale price back when selling to an American buyer… It’s a policy to de-industrialize the U.S.”

So the story isn’t just about labor or capitalism or even trade. It’s about who’s playing the game, and how. Hadrian isn’t winning because the rules changed—they’re winning by playing differently.

“Everyone has been trained to think that manufacturing isn’t a growth industry. That you can’t put a billion dollars of capex to work and expect it to pay off over ten years. And that’s what makes Hadrian different.”

The bigger picture Power paints is this: America’s industrial base didn’t just erode—it bifurcated. The defense side stayed (regulated by ITAR), while the commercial side fled offshore. That split killed scale, energy, investment—and morale.

“If you’re an engine manufacturer, your only growth comes from defense — and defense doesn’t offer long-term visibility. Very hard to justify a billion-dollar investment in an advanced factory on that basis.”

But Power is making the long-term bet. And he’s doing it by treating manufacturing like tech: high-capex, growth-oriented, driven by systems and automation.

“We punished capex like it was a sin. It’s the opposite of how you’d treat a tech company.”

This is why we need imagination. Not for some distant future… but to see what's already here.

Watch the whole discussion had with Chris Power on YouTube:

If you read this far, thank you! We clearly have a shared interest. Please consider subscribing to a paid tier to support my work. Also consider hitting the like button, it helps others discover us. Comments are only open to premium members, but feel free to DM me with any thoughts.

Subscribe now

Read more of my posts on Industrial Policy/Commons

I learned a ton from this episode. Go listen on YouTube, Overcast, Spotify, or Apple! Below I share a transcript with light-to-medium edits for those that want to dig deeper here.

Read more

Thank you for subscribing. Occasional forwarding is okay. If you are not a member, please sign up here.

1. Disappearing Data

Charts in this section are sourced from WSJ’s How Bad Is China’s Economy? The Data Needed to Answer Is Vanishing. Many economists have theories about why China discontinues economic indicators. My view: the data stops when it no longer fits the story they want to tell.1

National land sales, by area and value

You can go take a look yourself at NBS’s data portal, click “Real Estate” then “Development and Sales of Real Estate”. Monthly data has stopped since Dec’2022.

Unemployment

Five months later, Beijing began releasing a new data series. The real youth jobless rate, it said, was 14.9%.

Officials said the new data series excluded nearly 62 million people who were studying full-time in universities, and so shouldn’t be counted as jobless. But that didn’t make sense to economists. Statistics typically count anyone actively looking for a job as unemployed, including full-time students.

Toll roads’ balance sheet, New stock market investors, Number cremated bodies, Soy sauce production

Demographics: Fertility rate

The true picture is likely worse than it appears. From the Wall Street Journal (emphasis added):

The country’s low fertility rate has become a major economic liability—and some data pointing to it is gone, too. In the mid-2000s, an economist named Yi Fuxian questioned the accuracy of China’s population data and argued that tuberculosis vaccinations were a better measure of population growth because every newborn in China is required to be vaccinated.

In 2020, 5.4 million such vaccines were administered, according to data compiled by the private Chinese think tank Forward Business and Intelligence. Chinese authorities said the country recorded 12.1 million births that year.

A year later, the National Institutes for Food and Drug Control discontinued the weekly data release of tuberculosis vaccines administered, along with other vaccine data.

In a country where statistics must serve the story, silence often speaks loudest.

(Sidenote: We previously cited Yi Fuxian’s work in a piece arguing China needs year-over-year disposable income growth of 9% or more to make real progress toward becoming consumer-driven and avoid the demographic wall.)

2. US-China Tariffs

The updated chart from Peterson Institute (PIIE) is something to behold.

Read more

This one started with a clip from Palmer Luckey on Rick Rubin’s podcast. Large excerpt from that podcast in today’s Recommended Listen. Luckey, who founded Anduril, doesn’t pull punches when talking about what the U.S. lost in letting go of manufacturing.

Here, I’m exploring the downstream effect of those choices:

1. Why smart students in the U.S. and China make such different decisions

2. How subsidies can create labor gluts, not golden tickets

3. And why, in China, state support often lowers wages instead of raising them

Thank you for subscribing. Occasional forwarding is okay. If you are not a member, please sign up here.

What Shapes a Nation’s Talent?

Imagine a smart 18-year-old in the U.S. and one in China. Both are top students, both at a crossroads. What they choose to study — and eventually do for work — isn’t just a matter of personal interest or ability. It’s shaped by something much larger: the structure of opportunity that surrounds them.

In the U.S., that means scanning a labor market where the best-paying jobs are in tech, finance, or law. In China, it might mean heading straight into engineering — because that's where the jobs are, and where the state is pushing investment.

These decisions don't happen in a vacuum. They’re a feedback loop between labor markets, education systems, and industrial policy. And when subsidies flood a sector, and firms outgrow those subsidies—they don’t just reshape industries. They reshape lives.

The Slow Erosion of America’s Industrial Commons

I previously covered Andy Grove’s warning not to abandon so-called “commodity” manufacturing. Those mundane operations—casting, welding, molding—aren’t low value. They are part of an industrial commons: the shared know-how, infrastructure, and labor pool that support entire industries.

We hollowed out our manufacturing base. Not overnight, but steadily — driven by flawed economic models and free trade orthodoxy. The idea was simple: let comparative advantage do the work. But free trade doesn’t exist. It’s managed trade. And we let other nations manage it better.

As Palmer put it:

The problem is — and I mean, there are a million problems — but what we did was hollow out our country by allowing China into the World Trade Organization and letting American companies outsource manufacturing to China without penalty, without import tariffs, without any disincentive to do it. Why wouldn’t you, if you’re allowed to send production to another country where everything’s cheaper, there are no environmental regulations, and no labor laws? Why wouldn’t you?

We’ve been able to get a bunch of cheap stuff over the last 50 years as a result. That has helped in some ways — everyone in the U.S. can buy cheap TVs, cheap cars, and cheap goods because of China’s rise. But the flip side is there’s no more manufacturing here.

The erosion of our industrial commons was gradual, except the labor part—that went quickly. Here we can see it with respect to our manufacturing employment statistics:

Smart Kid in America

Let’s look at it through the eyes of a high-achieving American student.

Read more

For today’s piece, we dig into the newly released Q3 and Q4 2024 results. Let’s dive in.

Thank you for subscribing. Occasional forwarding is okay. If you are not a member, please sign up here.

Key takeaway: Green shoots are visible, but structural pessimism still dominates.

PBOC Depositor Sentiment: A Mixed and Cautious Mood

Urban depositor surveys from the People's Bank of China (PBOC) reveal a population that remains cautious, risk-averse, and deeply influenced by long-term structural headwinds. Sentiment has weakened across income, employment, and housing—while spending preferences show subtle shifts.

1. Structural Weakness in Income and Employment Confidence

Both Income Sentiment and Confidence have been in a downtrend since 2018—the year US-China trade tensions escalated. The decline accelerated around COVID and deepened with domestic lockdowns. The same deterioration is visible in Employment Sentiment and Expectations, which now sit at series lows.

• Downtrend in income sentiment started around the trade war kickoff.

• Employment expectations have steadily worsened over the last 5 years.

• Together, these reflect diminished confidence in personal economic prospects.

2. Deep Pessimism in Housing Market Outlook

The housing market—a key pillar of China’s economy—continues to show signs of stress:

• Home price sentiment has been bearish for 6 consecutive quarters, unprecedented in this dataset.

• Home sales peaked in mid-2021, and both volumes and YoY growth remain well below historical norms, although there are signs of stabilization in recent months.

• The number of people expecting an uptrend in prices has risen slightly in Q4’24, hinting at a potential inflection point.

The most optimistic read: the worst may be behind us—but this is still a fragile bottom.

Residential Sales Volumes (NBS data)

Read more

China still exports way more than it imports, which means they cannot be a source of global demand—they can’t be the buyer of last resort. And more importantly Xi doesn’t want to be.

Thank you for subscribing. Occasional forwarding is okay. If you are not a member, please sign up here.

Globalization model quietly stopped working

If someone asked, where will future global demand come from? The instinctive answer might be: globalization. That it will continue spreading, lifting people out of poverty and increasing trade and consumption.

But that model quietly stopped working a decade ago.

In June 2019 Catherine Mann was on Bloomberg Surveillance (overcast).1 Tom Keene asked her about global inter-dependencies—the way one country’s policies are often blunted by those of others. Her response:

"The trade relationship we have right now with China is emblematic of 10 [now 15] years of stagnation in terms of global integration. Earlier in my career we talked about more deeply integrated global supply chains, a greater variety of products crossing borders, lower prices available to both businesses and consumers. And all that stopped about 10 [now 15] years ago."

So if globalization is no longer driving demand, what about growth-driven imports?

, Council on Foreign Relations, November 2018 wrote a blog post entitled "China should import more." He included this chart, showing the declining trend in China's trade in manufactures with the US as a share of GDP.

Brad updated a version of this chart on March 31, 2025 that shows China’s trade in manufactures as a share of GDP. Non-processing imports are manufactured goods that are not going to be processed and re-exported. The updated chart includes non-processing imports from all countries. Quoting Mr. Setser:

Xi's vision of openness seems to be a world that is open to Chinese exports, not a China that is open to imports of manufactures

Read more

I’m not claiming these tariffs are the perfect strategy or even a good strategy. I know some of the ideas here are simplified and might invite pushback. That’s fine. My goal is to offer a clear and honest explanation (take?) of how the world trading system ended up so unbalanced… and why that matters.

Thank you for subscribing. Occasional forwarding is okay. If you are not a member, please sign up here.

Explaining the Trump Tariffs to My Chinese Colleague

The global trading system is unbalanced 不平衡. This is evident in the persistent trade imbalances. In the theoretical world of “free trade” this shouldn’t happen. Ricardo’s theory of comparative advantage assumes everyone benefits. But that theory doesn’t hold up in the real world today. I imagine Ricardo would update his views. Massive-scale manufacturing, built with state subsidies, was never what he had in mind.

Of the persistent trade imbalances, China has the largest surplus and the U.S. has the largest deficit.

China’s trade surplus, using China customs data.1

And here’s World Bank balance of payment data2, U.S. is the light blue line on the bottom. Negative = deficit.

So what’s causing all this?

There are 2 key angles and they have recently converged.

1. The populist view

A lot has changed since Ricardo’s time. Capital can now move freely across borders. So can technology and raw materials, thanks to digitization and global shipping. But labor is still stuck, just as it was in the 19th century.

The populist view:
Free Trade Agreements made it safer for capital to move abroad and invest in less developed countries. This was great for capital and pretty good for the workers in these countries. But not so great for workers in developed countries.

explains it well:

Opening trade between countries that had different resource endowments and different relative prices would trigger redeployment of capital, labor, and goods (trade), which would drive the full set of relative prices closer together. That would certainly increase total production for all countries combined. But it would not result in the free trade argument that each nation’s output goes up as a result.

To put names to the story, opening trade between capital-rich America and capital-poor China has induced capital owners to redeploy it from where it is abundant to where it is scarce. The capital owners–today we call then the 1%–are unambiguously better off as a result. Workers in China are better off because they are not [now] more productive. Workers in the US are now worse off because they have fewer factories and machines for producing product.

The result: industrial commons in the U.S. have eroded and our workers are worse off—opening the doors for populism.

2. China’s barriers

China’s economic structure is set up so that everyday people 老百姓 don’t get much money. Instead, resources are funneled into businesses, especially exporters and factories. For example, labor unions are illegal, and subsidies are aimed at production (producers), not consumption (households). For more on this, see my post Welfare for Whom?

China also follows a strategy of import replacement. In simple terms: “We sell you our stuff, and we do not buy your stuff.” When China must import something, it works to incentivize3 domestic alternatives and eventually replace the foreign version/good.

Some of the other barriers to trade (read imports) that China uses/used are foreign currency exchange management (making the CNY weaker), harsh customs polices4, forced JVs, cheap financing from state banks, market access restrictions, among others.

Deindustrialization & national security

Read more

The Economy Has a Stack

The idea of a “tech stack” comes from software engineering. The first one I learned about was the LAMP stack: Linux, Apache, MySQL, and PHP. Linux was the base operating system, Apache was the web server, MySQL handled databases, and PHP stitched it all together to generate webpages. There’s also the MEAN stack—MongoDB, Express.js, Angular.js, and Node.js. Each stack layers technologies on top of one another, with each layer enabling the next.

The key insight is this: every tech stack has limitations. Some things are easy to build within a stack, others are hard. Constraints matter. And in an economy—especially a production economy—physical constraints matter even more than digital ones.

That’s where this graphic comes in:

Read more

Let the Industrial Commons Erode or Start a Flywheel Effect

My nerdy side takes pleasure in finding original source material. This time I’ve found the paper that coined the term “industrial commons.” The name fits—it instantly made sense to me the first time I heard it. (Perhaps living and working for 15 years in China, where industrial commons are taken seriously, helped.)

Published in 2009, Gary Pisano and Willy Shih’s article “Restoring American Competitiveness,” describes the erosion of the industrial commons—the collective R&D, skilled labor, and supplier networks that sustain innovation. Their argument is straightforward: outsourcing has hollowed out America’s ability to make the very products it invents.

Defining the Industrial Commons

Just as "the commons" once referred to shared grazing land that benefited all farmers, industries also have commons—a shared foundation of capabilities critical to innovation and competitiveness:

“Industries also have commons. A foundation for innovation and competitiveness, a commons can include R&D know-how, advanced process development and engineering skills, and manufacturing competencies related to a specific technology. — Pisano and Shih (2009)

These commons often supports multiple industrial sectors:

"technological know-how, operational capabilities, and specialized skills that are embedded in the workforce, competitors, suppliers, customers, cooperatives R&D ventures, and universities and often support multiple industrial sectors.” — Pisano and Shih (2009)

And importantly, the industrial commons is “geographically rooted”:

The geographic character of industrial commons helps to explain why companies in certain industries tend to cluster in particular regions—a phenomenon noted by Michael Porter and other scholars. Being geographically close to the commons is a source of competitive advantage. — Pisano and Shih (2009)

In 2009, The World is Flat by Thomas Friedman was making waves. But Pisano and Shih push back on the idea that technology alone makes geography irrelevant. For manufacturing and R&D, proximity still matters.

“Detailed empirical work on knowledge flows among inventors by our HBS colleague Lee Fleming shows that proximity is crucial. An engineer in Silicon Valley, for instance, is more likely to exchange ideas with other engineers in Silicon Valley than with engineers in Boston. When you think about it, this is not surprising, given that much technical knowledge, even in hard sciences, is highly tacit and therefore far more effectively transmitted face-to-face.” — Pisano and Shih (2009)

How does knowledge spread throughout an economy or industrial ecosystem?

People switching jobs. A robust industrial commons can help diffuse knowhow and technology too.1

“Other studies show that the main way knowledge spreads from company to company is when people switch jobs. And even in America’s relatively mobile society, it turns out that the vast majority of job hopping is local.” — Pisano and Shih (2009)

Read more

International trade agreements shape America’s economy, industries, and workforce. Yet, their complexities often keep them out of public discussion. This post provides an overview of the General Agreement on Tariffs and Trade (GATT) and its successor, the World Trade Organization (WTO), with key insights from No Trade is Free by Robert Lighthizer.

GATT: The foundation of the modern trade system

Established in April 1947, GATT was signed by 23 nations to create a multilateral framework for global trade. Over time, it evolved into a system that facilitated trade negotiations while allowing nations to safeguard domestic interests.

“[GATT] spawned a bureaucracy-laden organization in Geneva, Switzerland, to facilitate the negotiation of international trade deals.” (No Trade is Free, p. 49)

Critical provisions:

• Most Favored Nation (MFN): Each member had to extend any trade liberalization granted to one country to all members, with limited exceptions (e.g., free trade zones and special treatment for developing nations).

• Negotiated Tariff Schedules: Members agreed to cap tariffs at negotiated levels.

• National Treatment: Imported goods had to be treated no less favorably than domestically produced “like” goods once inside a country.

Importantly, GATT did not create a system of unregulated “free trade.”

“The GATT did not introduce an era of unregulated trade. Instead, US policymakers regularly used their constitutional authority to ensure that efforts to lower tariffs and encourage trade with US allies would not disrupt the US economy and hurt American workers.” (No Trade is Free, p. 50)

GATT’s trade liberalization efforts were gradual and negotiated over decades, primarily benefiting U.S. allies. When trade deficits threatened U.S. economic interests, leaders took action:

• President Nixon intervened to address trade imbalances.

• Congress enacted tools to shield U.S. industries, such as Section 232 (national security tariffs), Section 301 (retaliatory tariffs), and stronger anti-dumping/subsidy laws (No Trade is Free, p. 51).

Eventually, GATT transformed into the WTO in 1994—with consequences that Lighthizer argues were disastrous for U.S. manufacturing workers.

The WTO: A shift in global trade rules

…with awful results for manufacturing workers.

United States Trade Balance: Total, Goods, Services (Millions of 2020 Dollars)

The WTO expanded GATT’s framework, introducing new disciplines that extended beyond goods into services, intellectual property, agriculture, and textiles—areas previously outside multilateral trade rules.

Key Changes Introduced by the WTO:

• “Single Undertaking” Rule: Unlike GATT, where some agreements were optional, WTO membership required nations to adopt all trade agreements, creating a system with over 60 binding agreements from the Uruguay Round.

• Development Focus: The WTO placed more emphasis on development, granting special treatment to developing countries, including longer implementation periods and lesser obligations.

However, the most significant changes were institutional:

1. WTO as a Permanent Institution

Unlike GATT, which was signed as a temporary arrangement1, the WTO was established as a full-fledged international organization with:

• A legal personality, allowing it to enter agreements and function as an entity.

• A Secretariat in Geneva to oversee trade negotiations and compliance.

• Defined decision-making procedures under a Ministerial Conference and General Council.

Diagram of the WTO’s organizational structure

Subscribe now

2. The WTO Dispute Settlement Mechanism (DSM): A Powerful Enforcement Tool

The WTO introduced a binding dispute settlement mechanism, a major departure from GATT’s weaker enforcement system.

• Under GATT: Any ruling required consensus—including from the losing party—allowing countries to block unfavorable rulings.

• Under the WTO: The “reverse consensus” rule meant that dispute rulings were automatically adopted unless every member (including the winner) agreed to reject them.

The system also introduced a two-tier structure:

1. Dispute Panels: A panel of experts reviewed trade disputes and issued rulings.

2. Appellate Body (AB): Countries could appeal legal interpretations, and the AB’s decision became binding.

Initially, this mechanism worked as intended, with most countries complying. However, over time, the Appellate Body became a major source of controversy.

The WTO’s Dispute System: A Breakdown in Functionality

Lighthizer and other critics argue that the Appellate Body (AB) evolved beyond its original purpose, engaging in judicial overreach by:

• Creating new trade rules instead of strictly interpreting agreements. (“Became an unelected lawmaking body, creating rules that have vast implications for international trade.” – No Trade is Free, p. 69)

• Issuing “findings of fact” despite WTO rules precluding them from doing so.

• Ignoring procedural deadlines, often exceeding the 90-day limit for rulings.

• Extending judge terms beyond limits, allowing expired members to rule on cases.

• Financial incentives to extend cases, as AB judges were paid daily.

• Perceived bias against the U.S., ruling against U.S. policies on taxation, subsidies, and regulatory standards.

Consequences and the U.S. Response:

By the 2010s, tensions escalated, with the U.S. arguing that the AB had become a self-perpetuating judiciary that weakened U.S. sovereignty. In response, the U.S. blocked new AB appointments, and by 2019, the AB ceased functioning, paralyzing the WTO’s dispute resolution system.

Lighthizer’s Critique of the WTO: A System that Restrains U.S. Sovereignty

Lighthizer argues that the WTO locks the U.S. into outdated commitments, preventing adjustments for changing economic conditions:

• Bound Tariff Rates: Members cannot unilaterally raise bound tariffs, even when global conditions shift.

• Different Legal Traditions: The Anglosphere views trade agreements as rigid contracts with precisely written obligations, while Europe sees them as evolving, allowing reinterpretation to enable favorable outcomes. This difference enabled looser interpretations of WTO rules over time.

The Appellate Body’s overreach was particularly problematic, as WTO rulings began interfering in domestic U.S. policy2, including:

• Taxation: Ruling against U.S. corporate tax structures.

• Subsidies: Striking down U.S. measures to assist industries harmed by unfair trade.

• Social Policy: Overturning U.S. gambling restrictions.

It became clear to Lighthizer and others the Appellate Body was biased against the United States3.

Lighthizer’s Proposals to Fix the WTO

Lighthizer argues that the WTO, in its current form, is outdated and unfit for modern trade realities. His solutions:

1. Reset the global tariff system to allow flexibility.

2. End the “FTA end run” around MFN treatment, preventing abuse of free trade agreements.

3. Limit “developing country” status to only the poorest nations (e.g., preventing China from claiming special treatment).

4. Counter Chinese economic aggression by permitting compensatory tariffs and unilateral action against predatory policies.

5. Introduce a “sunset clause” for WTO agreements to ensure periodic renegotiation where the agreement expires when met with negotiation deadlock.

6. Mechanism to assure long-term balanced trade to prevent chronic trade imbalances.

7. Scrap the current dispute settlement system, model a new one after commercial arbitration.

Conclusion: A WTO in Crisis

The WTO was created to enforce trade rules and provide stability, but over time, it has drifted from its original intent. Lighthizer’s critique highlights how the system now limits U.S. sovereignty, prevents tariff flexibility, and has an activist dispute mechanism that undermines national policies.

With the Appellate Body now defunct and trade tensions rising, the future of the WTO remains uncertain. Whether it will be reformed, replaced, or abandoned is an open question—but what is clear is that trade policy is no longer a niche topic. It directly impacts jobs, industries, and national sovereignty, and it deserves a greater public debate.

Whether consuming or engaging in this public debate, please remember that free trade is a nice theory that “works” in abstract vacuum, but it doesn’t exist in practice. Reject the unregulated free trade ideologues. More to come on this in a future post.

If you read this far, thank you—we clearly have a shared interest. Please consider hitting the like button, it helps others discover us, or consider subscribing to a paid tier to support my work. Comments are only open to premium members, but feel free to DM me with any thoughts.

Subscribe now

Diffusion Deficit

Bill Bishop recently introduced me to the concept of a diffusion deficit in a Substack note reply. (Side note: The quality of engagement and writing on Substack continues to impress me. It’s a pleasure to be here.)

The term comes from Jeffrey Ding’s paper, “The Diffusion Deficit in Scientific and Technological Power: Re-assessing China’s Rise.” His research challenges the conventional wisdom that China is poised to become a dominant science and technology superpower.

What is a Diffusion Deficit?

Ding’s work highlights that standard assessments of China’s technological progress—measured by patents, scientific papers, and R&D spending—fail to account for diffusion capacity, or how well a country scales and integrates innovations into its economy.

Ding argues that, much like the Soviet Union in the 1960s, China has a strong innovation capacity but a weaker diffusion capacity. While China excels at producing cutting-edge research and prototypes, its ability to scale and integrate these innovations across industries lags behind.

A diffusion-centered approach shifts the focus from mere invention to the adoption of innovation at scale throughout an economy, society or industry.

While Ding’s efforts to measure diffusion capacity are commendable—it’s an incredibly difficult task! Many of the metrics used are broad and survey-based or just plain estimates, raising questions about their accuracy.

One example is ITU’s ITC Development Index (see interactive data here)1, which measures 3G and 4G mobile network coverage but not 5G, which China leads. It also lists mobile phone owners at 100% in the U.S. but only 83% in China, despite:

1. China ranks number 1 in scale of mobile phone subscribers at 1.7 billion

2. First-hand observations of widespread mobile phone usage in China, by anyone who’s visited.

This discrepancy illustrates the challenge of relying on high-level diffusion indicators without considering industry-specific dynamics.

In my humble opinion, a true measure of diffusion capacity should assess the quality of the industrial commons. Other key factors would include the intensity of industrial competition, which drives firms into a trial-and-error process of turning innovations into efficient, mass-produced realities. This is where industrial commons and manufacturing clusters come into play—something China has in abundance.

Subscribe now

In the following sections, I examine three key factors that influence diffusion:

1. Proximity in supply chains

2. Industrial robots as a flawed diffusion metric

3. Industry-academia collaboration

Proximity and Speed of Innovation

Diffusion often works best when suppliers and production lines are closely integrated. This creates a rapid feedback loop where manufacturing problems are identified and solved in real time. Consider this scenario:

• Production line workers notice a flaw and tweak a part.

• They walk it over to a nearby supplier—literally next door.

• The supplier immediately adjusts their manufacturing process.

• The new version is seamlessly implemented in future production.

This process accelerates problem-solving and ensures rapid iteration—one of the keys to successful adoption of new technology, or diffusion capacity.

High proximity is a core feature of a robust industrial commons: clustering of resources, suppliers and talent, a.k.a. manufacturing clusters.

China’s industrial commons are robust with strong regional clusters. State media (which I take with a grain of salt) likes to highlight this:

“As a key driver of high-quality manufacturing development, China has established 178 national high-tech zones and 45 national advanced manufacturing clusters, scattered throughout the country.” - Global Times2

Industrial Robots: A Misleading Diffusion Indicator?

Some claim China’s slow tech diffusion is evident in its industrial robot adoption rates. But let’s break this down:

• China has installed 4.12 times more industrial robots than the U.S. since 2011 (chart above).

• But China also has 10 times more factories (6.03 million3 vs. 605,5484 in the U.S.).

• If you compare these figures, China has fewer robots per factory than the U.S.—which could be interpreted as a diffusion gap.

This reasoning misses the bigger picture. Why should we care how evenly distributed industrial robots are across factories? In manufacturing, scale and efficiency matter more than per-factory diffusion.

If China has more industrial robots and a strong industrial commons to produce them at scale, that advantage outweighs concerns about their relative spread.

Read more

Questions

1. Does the US services export surplus offset the goods …

Read more

Industrial Commons

I came across this term after I published Industrial Policy, A Foundation. It’s the perfect term for what I was calling “Scaling Value” in that post.

The industrial commons is what enables scaling. It consists of two key components:

1. The ecosystem – suppliers, logistics, infrastructure, capital access, and more.

2. Technical knowledge – process know-how, workforce expertise, and accumulated experience within the industry.

Despite the word industrial, this concept isn’t limited to manufacturing. For example, Silicon Valley was once a hardware industrial commons focused on silicon and semiconductors; today, it’s a software/digital industrial commons.

Fragility and Importance

Industrial commons are like good reputations—built over time, yet easily destroyed. If one critical link in the chain disappears (a supplier, skilled labor force, or piece of infrastructure), the entire ecosystem weakens significantly.

Why does this matter? In today’s world, many goods and services require economies of scale to compete globally. Without scale, companies lose market share. That decline can start a downward spiral:

China’s Playbook: Building an Industrial Commons

China has executed a textbook example of industrial commons development. Some of its key policy tools included:

• Special Economic Zones (SEZs) – to attract foreign investment and build clusters of expertise.

• State infrastructure investment – to ensure logistics, ports, and supply chains were world-class.

• Taking low-margin industries – electronics, chemicals, and materials were seen as stepping stones, not dead ends.

• Subsidies, market access, and trade barriers – to protect and nurture domestic industries.

See China’s epic progress in the following chart from Robert Atkinson (ITIF).1

China’s industrial commons strategy was simple (to explain):

1. Recognize their starting position – huge population, high poverty, low resource wealth.

2. Invest aggressively in industry and capacity – knowing that scale leads to dominance.

3. Achieve market share dominance – then move up the value chain.

Why Build Industrial Commons?

Robust industrial commons offer major advantages, including:

1. Process Knowledge: The Unwritten Moat

Scaling production requires deep process knowledge—an asset that isn’t found in textbooks or research papers. Unlike theoretical knowledge, process knowledge accumulates within companies and industries over time as workers refine production techniques, solve problems, and develop efficiencies.

A prime example is Taiwan Semiconductor Manufacturing Company (TSMC). Their ability to build high-yielding fabs for advanced semiconductors is process knowledge honed over decades. It’s an enormous barrier to entry. More yield = lower costs = more market share = more profits = more R&D = stronger industrial commons.

2. The Andy Grove Warning: Don’t Abandon “Commodity” Industries

Andy Grove famously warned that losing “commodity” industries can cripple future technological development:

“I disagree [that it’s a success to lose the ‘commodity’ television industry]. Not only did we lose an untold number of jobs, we broke the chain of experience that is so important in technological evolution. As happened with batteries, abandoning today's ‘commodity’ manufacturing can lock you out of tomorrow's emerging industry.” — Andy Grove2

Losing an industrial commons today means losing the ability to scale future innovations tomorrow.

Thought Experiment: Industrial Commons in the Age of AI

Imagine a world where Artificial Super Intelligence (ASI) is attained. ASI rapidly discovers new scientific breakthroughs, running billions of simulations and directing automated labs to conduct real-world experiments.

Who benefits?

Only nations and companies with robust industrial commons will be able to translate these discoveries into physical reality at scale.

For instance, suppose ASI makes radical advances in robotics. The countries with strong robotics industrial commons will be first to implement them in manufacturing, defense, logistics, and healthcare.

It won’t matter where the discoveries are made—what will matter is who has the industrial commons to scale them.

The AI Economy: From Bits to Atoms

As I was finishing this post, I watched J.D. Vance give a speech on AI where he said:

“It’s my view that tech innovation of the last 20 years has often conjured images of smart people staring at computer screens engineering in the world of bits, but the AI economy will primarily depend on, and transform, a world of atoms.

Now at this moment, we face the extraordinary prospect of a new industrial revolution, one on par with the invention of the steam engine or Bessemer steel, but it will never come to pass if overregulation deters innovators from taking the risks necessary to advance the ball, nor will it occur if we allow AI to become dominated by massive players looking to use the tech to censor or control users’ thoughts.”3 [emphasis added]

Hard agree. The past two decades of tech innovation have been dominated by software, where regulation has been relatively light, and capital has flowed freely due to high ROI and lack of NIMBYism. But AI’s biggest impact won’t stay confined to digital spaces—it will reshape the physical world. Scaling AI-driven breakthroughs in robotics, materials, healthcare, energy, and manufacturing will depend on strong industrial commons, and that means rethinking how we regulate and incentivize the industries of the future.

Building industrial commons isn’t just about investment—it requires a hard look at onerous regulations that may be hindering innovations from crossing the valley of death—the gap between breakthrough research and large-scale commercialization. Right-sizing these regulations for the AI-driven industrial era will be critical if we want to lead in the next wave of technological transformation.

High-Stakes Choice: Nurture, Build or Fall Behind

Neglecting the industrial commons isn’t just an economic oversight—it’s a strategic vulnerability. Once lost, rebuilding it is exponentially harder, requiring not just investment but also time, expertise, and a coordinated ecosystem.

However, there is another option for rapidly constructing a micro industrial commons: extreme vertical integration. This approach, though incredibly difficult, involves a company or coalition amassing capital, talent, and resources to aggressively build an ecosystem from the ground up. An historical example is the U.S. industrial mobilization before entering World War II, where firms like Ford, GM, and Boeing rapidly scaled production with government support—a story well-documented in Freedom’s Forge. But extreme vertical integration still requires alignment with policymakers and, in many cases, broader societal backing.

Whether through long-term industrial commons development or rapid vertical integration, the lesson is the same: if we fail to actively nurture our ability to scale, we won’t just lose market share—we’ll lose the capability to lead in the next technological revolution.

In the 1930s, the Empire State Building—the tallest in the world at the time—took 410 days to build. A decade later, the Pentagon took 16 months.49 In the span of eight years during the Great Depression, President Franklin D. Roosevelt’s Works Progress Administration built some 4,000 new schools, 130 new hospitals, 29,000 new bridges, and 150 new airfields; laid 9,000 miles of storm drains and sewer lines; paved or repaired 280,000 miles of roads; and planted 24 million trees.

Compare those feats to more recent ones. In 2022, an opinion piece in The Washington Post observed that it had taken Georgia almost $1 billion and twenty-one years—fourteen of which were spent overcoming “regulatory hurdles”—to deepen a channel in the Savannah River for container ships. No great engineering challenge was involved; the five-foot deepening project “essentially . . . required moving muck.” Meanwhile, raising the roadway on a New Jersey bridge took five years, 20,000 pages of paperwork, and 47 permits from 19 agencies—even though the project used existing foundations. [emphasis added] — Over Ruled: the human toll of too much law, by Neil Gorsuch

Thank you for reading. If you enjoyed this post, please consider subscribing to a paid tier to support our work. Today’s graphics were generated by Napkin AI.

Share

Read more

Read more

Read more

This is a new post category, “Reflections”, where I’ll be sharing some stories from living and working in China for over 15 years.

Reflections

With China’s stock markets raging, I will share some stories from the Xi bull market 1.0. W…

Read more