There is no shortage of billion-dollar plans. There is, however, a shortage of people who can wind transformers, nuclear fuel that actually exists, and tanker captains willing to dodge missiles. This week's riffs follow the industrial boom to the places where throwing more money at the problem stops working.

Meanwhile, Stellantis has too few batteries and too few buyers, True Food discovers that opening restaurants isn't the same as building a restaurant empire, and an abandoned nuclear reactor gets another shot at life. We've also got revealing earnings numbers, freight and commodity signals, some excellent rabbit holes, and a few things worth keeping an eye on next week.

U.S. transformer workforce ~15,000 would need to roughly triple; coil winding is "almost an art form" with the vocational pipeline gone "for a generation." Standard power transformers at 128-week lead times (vs. 50–80 weeks five years ago); distribution transformer prices up 78–95% since 2019. GOES steel from a single domestic mill in Butler, PA, roughly doubled since Jan 2020. GE Vernova disclosed a total backlog above $100B — the largest power-equipment backlog ever reported. The White House invoked DPA Section 303 for grid infrastructure in April, stating "domestic industry cannot meet demand in a timely manner." A DPA declaration can commandeer capacity; it cannot conjure coil winders. You can build a transformer plant in two years; you cannot train a coil winder in two years. The take: transformer lead times won't collapse even when GOES supply eases — the recovery curve is measured in apprenticeship cohorts, not quarters. The companies quietly acquiring transformer service/repair businesses may be the better bet than the manufacturers.

Centrus signed a long-term HALEU (5–20% U-235) supply agreement from its Piketon, Ohio centrifuge plant for Radiant Nuclear's transportable Kaleidos microreactors. Large HALEU volumes won't exist until 2029, after Centrus's Oak Ridge center builds "thousands" of 45-foot AC100M centrifuges, trucked in pieces to Piketon and reassembled. Radiant's prepayments help fund the expansion. The supply agreement is simultaneously a purchase order and a financing vehicle — because project finance won't touch it. The take: everyone models SMR economics from the reactor side; the constraint is the fuel chain. The interesting company in the advanced-reactor story may be the centrifuge maker, not the reactor designer.

Sochaux (Oct 23–30) and Rennes (Oct 22–30) halted because ACC can't supply enough long-range EV batteries; Mulhouse (Oct 15–30) halted for weak demand, cutting to 1.5 shifts from November. ACC scrapped 2 of 3 planned cell plants; only Douvrin, France is ramping — September module output ≈ all of 2025's volume, still not enough. One company, one month: two plants can't build cars because batteries are scarce; one plant can't sell cars because demand is absent. The take: the gigafactory boom's graveyard is a single working plant — the whole continent's EV production plan now routes through Douvrin's module line. Watch whether Chinese cell imports fill the gap or tariffs make the shortage structural. (Sources search-returned, pages not loaded — verify before citing.)

Shell now expects Q3 refining margins of roughly $42/barrel, up from $24 in Q2 and a company record, as the Hormuz disruption squeezes global fuel supply; integrated-gas production is simultaneously expected at 740,000–780,000 boe/day following the $16.4B ARC Resources acquisition, but the refinery number is the better tell. Crude scarcity gets the geopolitical attention, yet barrels are only useful after somebody turns them into diesel, gasoline, jet fuel and petrochemical feedstocks, and refinery outages, constrained trade routes and product-market dislocations can make that conversion step more valuable than simply owning another barrel underground. Shell even expects lower utilization at Rheinland because low Rhine water levels are restricting operations, which is the entire Bottleneck Economy in one earnings update: the commodity is scarce, the conversion margin explodes, and then a river becomes the constraint on capturing it. The take: energy shocks migrate through the chain until they hit the least elastic physical node. Right now the margin is telling you that refining capacity, product logistics and working inventories are capturing more of the scarcity rent than the crude headline suggests.

UTLX Manufacturing, the Berkshire Hathaway-owned parent of Union Tank Car, filed antidumping and countervailing-duty petitions targeting railway tank cars imported from Mexico, alleging Mexican-built cars are unfairly priced and subsidized; the awkward part is that the companies on the other side are Trinity Industries and Greenbrier, two U.S.-based manufacturers that use Mexican production as part of their North American manufacturing networks and are now fighting the petition. A preliminary International Trade Commission decision is expected by Nov. 16 if Commerce initiates the case. This is a tiny trade fight with a much bigger question inside it: after thirty years of integrating Mexican industrial capacity into U.S. supply chains, reshoring policy now has to decide whether "American manufacturing" means an American company, a North American supply chain or literally a factory north of the Rio Grande. Tank cars are especially useful because this is not T-shirts or televisions; these are regulated pieces of industrial equipment moving chemicals, petroleum and food ingredients through the continental rail system. The take: USMCA-era reindustrialization is eventually going to collide with America-first capacity policy company by company, and UTLX just handed us a clean case study.

Tanker captains transiting the Strait of Hormuz are reportedly being offered as much as $100,000 a month plus $50,000 per transit, while some ordinary crew are receiving multiples of normal wages; at least 93 vessels have been attacked and 24 seafarers killed during the conflict, tanker day rates have reached roughly $1.3M and war-risk insurance can cost as much as $20M. The interesting number is not the freight rate but the wage. A tanker can be worth $100M+, carry another $100M+ of cargo and have military protection, insurance and a charterer willing to pay seven figures per day, yet none of it moves if 20 humans decide the paycheck is not worth getting hit by a missile. The take: labor becomes the final bottleneck when every other constraint has already been financially solved. Capital can bid for the vessel, insurance and fuel; eventually it has to bid for courage.

True Food Kitchen entered Chapter 11 with roughly $42.1M of debt, closed 12 restaurants, kept 34 operating, and secured about $20M of debtor-in-possession financing while pursuing a sale. Court materials point to expansion outside core markets, repeated strategy changes, and unsuccessful investments beyond the company's original model. That makes this more interesting than another “consumer is weak” story. Restaurant chains do not get the same economics from 50 scattered stores that they get from 50 dense stores: distribution routes, field management, local advertising, recruiting, purchasing, and brand awareness all improve when units cluster. The take: a lot of restaurant “growth” over the cheap-money years was really geographic dilution disguised as unit count. The winners in this shakeout may be the chains willing to get smaller on the map so they can become denser on the ground.

Babcock & Wilcox's mPower small modular reactor was supposed to help launch America's next nuclear industry. By the end of 2014, the Department of Energy had provided $111 million in development support. But committed customers never materialized, outside funding dried up, and the program was effectively abandoned by 2017.

Nearly a decade later, the reactor is getting another shot. In June 2026, Applied Atomics licensed the shelved technology for land-based applications. BWXT retained ownership of the intellectual property, exclusive manufacturing rights, and royalty rights on components manufactured by third parties. A separate feasibility study is exploring floating nuclear applications.

Here's a potentially attractive way to recover value from failed industrial R&D: let somebody else fund the commercialization gamble while retaining the rights to manufacture the hardware. No reactor has been commercialized through these new arrangements yet, but BWXT has positioned itself to participate if one does. The next questions are whether licensing advances, customers sign binding orders, and actual manufacturing revenue follows.

(Sources: BWXT 2016 Form 10-K | Applied Atomics, June 2026)

Powell Industries ($POWL) booked a record $934 million in new orders during fiscal Q3 2026, producing a 3.0x book-to-bill ratio. One data center order exceeded $400 million, the largest in company history. The company also secured approximately $60 million in LNG equipment orders and $75 million in petrochemical orders. Backlog reached $2.4 billion, up 69% year over year.

What caught my attention is that a single data center project can now generate an electrical equipment order larger than Powell's entire quarterly revenue. And this is a company historically associated with energy infrastructure and industrial electrical distribution, not just hyperscale computing.

Why it matters: the bottleneck is moving beyond electricity generation into the equipment that safely distributes and controls power. Engineered switchgear and electrical distribution systems are becoming critical infrastructure for data centers, LNG facilities, and industrial expansion. The companies capable of engineering and delivering these systems on schedule may have substantial pricing power. (Source: Powell FY2026 Q3 earnings)

Fastenal ($FAST) reported 14.7% daily sales growth in Q2 2026. Sales to contract customers increased 17.6%, compared with just 7.3% growth among non-contract customers. Heavy manufacturing sales increased 18.1%, while the company disclosed that 75.8% of revenue now comes from contract accounts, up from 73.2% a year earlier.

What caught my attention is the divergence between customers. Large manufacturers with integrated procurement agreements are spending significantly faster than smaller, transactional buyers. But part of this difference is Fastenal capturing additional wallet share, not necessarily broad industrial recovery.

Why it matters: industrial distributors can serve as useful manufacturing indicators, but revenue growth needs to be separated into price, production activity, and market-share gains. Fastenal's contract versus non-contract customer performance is a particularly interesting way to monitor whether industrial strength is broadening or remaining concentrated among larger operators. (Source: Fastenal Q2 earnings)

Republic Services ($RSG) reported that large-container volumes declined 2.2% in Q2 2026, primarily because of continued softness in construction-related activity. At the same time, landfill special-waste volumes increased 10.7% after adjusting for unusually high wildfire-related volumes in the prior year. Recycling commodity prices declined to $136 per ton from $149.

What caught my attention is how differently waste streams are behaving. Construction-related collection remains weak, while certain special-waste categories are expanding. Waste collection provides an alternative perspective on physical economic activity that does not depend on management surveys or construction announcements.

Why it matters: dumpster volumes can be a useful real-economy indicator. Construction containers, landfill tonnage, manufacturing waste, and special-waste movements can help identify which portions of the physical economy are actually operating, expanding, or slowing. Tracking the mix may reveal economic changes before they appear clearly in reported construction spending. (Source: Republic Services Q2 earnings call)

Texas Roadhouse ($TXRH) reported 6.2% comparable restaurant sales growth in Q2 2026, consisting of 3.0% higher guest traffic and a 3.2% increase in average checks. But traffic growth slowed from 4.0% in the prior-year quarter. At the same time, commodity inflation reached 7.0%, driven largely by beef costs, contributing to restaurant margins declining from 17.1% to 16.4%.

What caught my attention is the interaction between consumer behavior and upstream agricultural economics. Customers continue visiting restaurants, but traffic growth has moderated while meat costs are rising faster than average checks.

Why it matters: Texas Roadhouse is a useful downstream indicator for ARC's discretionary-spending coverage. The business captures consumer spending while being exposed to cattle supply, processing economics, transportation, and labor costs. Following traffic, checks, and restaurant margins together helps distinguish resilient consumer demand from growing pressure in the physical supply chain. (Source: Texas Roadhouse Q2 2026 10-Q)

BWX Technologies ($BWXT) ended Q2 2026 with approximately $8.4 billion in backlog, up 40% year over year. After acquiring Precision Components Group, management discussed establishing additional U.S. manufacturing capacity for large nuclear components. CEO Rex Geveden explained that any new large-component manufacturing location would require deepwater port access to transport steam generators and reactor pressure vessels.

What caught my attention is that marine transportation access is an explicit site-selection requirement for nuclear manufacturing. BWXT is evaluating how to combine specialized manufacturing capabilities, an established workforce, and transportation infrastructure to serve domestic and international nuclear projects.

Why it matters: nuclear power expansion faces bottlenecks that cannot be solved simply by permitting more reactors. Large forgings, pressure vessels, specialized welding, qualified manufacturing facilities, and heavy-component transportation are all constraints. Deepwater ports and proximity to suitable industrial infrastructure could determine which regions capture the next generation of nuclear manufacturing investment. (Source: BWXT Q2 earnings call)

Pop-Tarts Backs Away From BuzzBallz Collaboration — Mars says it regrets the 15%-ABV Pop-Tarts-flavored alcohol partnership inherited through Kellanova, while Sazerac appears considerably less apologetic. Just an objectively bizarre CPG story. (Finger)

Mo Money, Mo Problems for Amateur Golf — Fried Egg looks at the modern amateur game through the U.S. Mid-Am and the increasingly blurry distinction between elite amateur and professional golf. The top end of amateur golf is starting to look a lot less amateur. (Fried Egg)

Rome Menswear Shopping Guide — Five years of collected recommendations spanning vintage watches, Fila tennis gear, bespoke shirts, shoes and classic menswear shops around Rome. Basically a ready-made hit list for anyone who cares about clothes and ends up in the city. (Substack)

Intern Pierre Does Luxury Channel Checks in Turin — Hermès was busy with what appeared to be repeat local customers while Gucci looked materially weaker. A fun bit of physical-world channel checking that offers a counterpoint to what the equity charts alone are saying. (Intern Pierre)

A $30,000 Working Sheepdog — Bloomberg profiles elite British border collies doing agricultural work owners describe as otherwise nearly impossible. A very good labor-capital substitution story hiding in the lifestyle section. (Bloomberg)

Ari Emanuel’s Autobiography — Senra works through Roll the Calls and Ari’s progression from the CAA mailroom to building Endeavor, acquiring UFC/WWE and eventually taking the company private. The real lesson is operating velocity: hundreds of calls, constant motion, unusual willingness to ask, and an almost pathological refusal to stop pushing. (YouTube) 

David Senra Sits Down With Ari Emanuel — Two days after the book episode, Senra actually sits across from Ari for 84 minutes on buying UFC, getting the ESPN deal done, taking over William Morris, learning from David Geffen and Egon Durban, dyslexia, rage, psychedelics, and why Ari still spends his life rolling calls. It is basically the perfect companion to #436: first Senra studies the operating system from the book, then he interrogates the guy running it. (YouTube)

10 Things I Learned From Interviewing 40 Dads — Brendan Leonard distills recurring themes from conversations with 40 fathers, including time, patience, uncertainty and the relentlessness of parenting. Simple observations, but a surprisingly good collection of them. (Semi-Rad)

If You Haven’t Worn It in Two Years, Give It Away — O.W. Root on cleaning out the closet, storing what actually deserves keeping and letting unused clothing go. Basically a case for treating a wardrobe more like a curated collection than a storage unit. (The Fitting Room)

October 13 — Domino’s Q3 earnings — A clean read on downstream discretionary spending: watch U.S. same-store sales, order frequency, delivery versus carryout, franchisee margins, and whether higher food and fuel costs are changing customer behavior.

October 14 — September CPI — Ignore the headline first and pull gasoline, food away from home, airfares, vehicle costs, and other categories where the current energy shock should be reaching consumers. The question is how quickly higher physical-economy costs are leaking into discretionary purchasing power.

October 15 — Prologis Q3 earnings — One of the better reads on the warehouse layer of the economy. Watch occupancy, rent growth, customer move-outs, development starts, and whether logistics tenants are expanding space or sweating existing buildings harder.

October 15 — Alcoa Q3 earnings — Watch alumina and aluminum realizations, energy costs, smelter utilization, bauxite economics, and any update on the recently announced South32 asset acquisition. This is a useful test of whether higher metals prices are actually becoming better producer economics.

October 15 — September retail sales — Break out restaurants, building materials, autos, gasoline stations, sporting goods, and other discretionary categories rather than staring at the aggregate. This should give us a cleaner look at where household dollars are going as fuel and food costs absorb more of the wallet.

Know someone who should be reading this? Forward it along, or send them the signup link. The conversation gets better when more of the right people are in the room.

Share with a friend

Independent research answers to no one, except the espresso machine. ARC runs on coffee and 10-Ks. If the research is worth something to you, please buy me a cup.

Thank you and may the force be with you,