Newsletter · · Ashutosh Agarwal

Trade War, Tariffs & Reshoring - Week of June 22, 2026: Operators treat tariffs as permanent and pile into the picks-and-shovels

Trade war, tariffs and reshoring newsletter for the week of June 22, 2026. Across the podcast tape, operators have stopped waiting for tariffs to be negotiated away and are spending against a permanent regime, putting capex into automation and electrical infrastructure while permitting and power emerge as the binding constraints.

Trade War, Tariffs & Reshoring

Week of June 22, 2026: Operators treat tariffs as permanent and pile into the picks-and-shovels

The week's through-line: the debate has finished moving. Nobody on the operator side is still waiting for tariffs to be negotiated away or for factories to "come back from Asia." They're treating the regime as permanent and spending against it. The investable question has shifted from who gets hit by tariffs to who is putting capex into domestic capacity, and who controls the electrons to run it. This week the operators and the pundits agreed on the destination and split sharply on the timing.

1. The tariff regime is now load-bearing

The clearest signal of permanence came from the legal mechanics. When the courts threatened the IEPA-based tariffs, the administration simply re-based the same rates onto Section 122 of the Trade Act, a fresh 10% across all countries with strong signals it goes to 15%, while the Section 232 stack (50% on imported steel and aluminum, 25% on autos and copper) was never touched. As FinPod put it on "Tariffs, Trade Policy, and Reshoring" (Jun 18), "the level of the tariff persists even when the legal authority vanishes overnight." Note the open balance-sheet item: refunds on IEPA tariffs already paid are "a black hole of billions of dollars" companies can't yet book.

On the enforcement front, the pundits at The Trade Guys (Jun 15) flagged ~60 pending Section 301 forced-labor cases backed by only "2-3 paragraph" assertions per country, legally fragile and a live appeal risk, plus overdue Section 232 investigations on aircraft, pharma, chips and copper. Boeing has already lobbied its way to an aircraft-sector exclusion. Watch the copper and pharma 232 determinations as the next discrete catalysts.

2. Margins: who's eating it, who's escaping

The single most PM-actionable segment of the week, again from FinPod (Jun 18):

  • Apple assembled ~55M iPhones in India in 2025 (+50% YoY, ~a quarter of global output, including the entire iPhone 17 line) and shifted iPad/Watch to Vietnam. India costs an estimated 5–8% more per unit, paid as an "insurance premium," yet Apple still flagged a ~$900M quarterly tariff hit. Reshoring mitigates; it does not eliminate.
  • GM took a $3.1B gross tariff cost in 2025, compressing margin from ~9% to 6.2%, and cited up to $5B of potential future exposure when cutting guidance (still posted ~$3.4B quarterly profit).
  • Ford faced a similar ~$3B gross hit but pulled it to ~$1B net through decontenting trims, dynamic pricing, and swapping tier-2 suppliers out of tariff-heavy regions.

The trap worth underlining for any domestic-manufacturing long: even a US-built vehicle absorbed $1,600–$2,000 of added cost from steel/aluminum 232 tariffs bleeding into domestic material prices; fully imported cars, $5,000–$8,900. Moody's pegged the 2025 hit to global automakers' operating profit at >$30B. There is no insulated product.

3. The reshoring trade is in the picks-and-shovels

The standout operator voice was portfolio manager Chris Semenuk on Other People's Money / Monetary Matters (Jun 18), and his framing is the one to steal: the re-industrialization "infrastructure" phase (roads, shells, production centers) is done, "now those factories will be kitted out." The dollars flow to automation and efficiency equipment, not jobs (7M manufacturing jobs lost over 15 years "are not coming back"). His named beneficiaries: Rockwell, Emerson, Cognex (machine vision), Parker Hannifin, Fastenal, Ingersoll Rand, Caterpillar, Gates, Applied Industrial, Timken. He anchors it in hard data: April industrial-production growth led by communications equipment, electrical equipment, semiconductors, chemicals and utility equipment.

Two operator caveats on execution:

  • Permitting is the binding constraint. Patrick Wolf of IAMT on U.S. Manufacturing Today (Jun 16) called fragmented permitting "the single greatest structural barrier" to turning announcements into output, with a dependency chain running semis to data centers to power to manufacturing.
  • Logistics arbitrage is real money. On FreightCasts (Jun 16), operator Curtis Spencer noted average tariff rates jumped from ~2.5% pre-Trump to 20–25%, a 10x move that makes Foreign Trade Zone duty deferral genuinely valuable; the de minimis repeal has shifted warehouse demand toward Chinese 3PLs serving Temu, Shein and TikTok.

Country-specific read-throughs: Canadian softwood lumber now carries a >45% combined duty (≈35% AD/CVD + a new 10% 232 added in October, up from ~14%), per Strategic Alternatives (Jun 18); containerboard pricing is up $50/ton YTD on 8.5% capacity rationalization. And Cross-border Tax Talks (Jun 19) noted Taiwan's $250B US investment commitment (semis into Arizona), with manufacturers rerouting China to Vietnam, Malaysia, India and Mexico.

4. Automation & humanoids: the labor arb is going live

The humanoid story moved from thesis to deployment. Via the Elon Musk Podcast (Jun 15): BMW Spartanburg running Figure 3 on sheet-metal handling at 99% accuracy; Mercedes deploying Apptronik's Apollo (55-lb payload); BMW Leipzig using Aeon for battery assembly. The unit economics that make operators take the meeting: Robot-as-a-Service at ~$25/operating hour vs. $45–50 fully burdened human labor. Capital is validating it; Rivian spin-out Mine Robotics hit a $3.4B valuation on a $400M round (Kleiner Perkins-led, a16z/Accel/VW/Salesforce in), >$1B total raised.

On the sell-side, William Blair's "Race to Infinite Labor" (Jun 16) flagged that the consensus commercialization timeline has compressed from "decades away" in early 2024 to roughly five years now, driven by aging-population labor headwinds. Treat as a structural tailwind, not a 2026 revenue line.

5. The real bottleneck is electrons

Every reshoring/AI thread converges on power, and the operator data here is alarming:

  • PJM (via The Banker Next Door, Jun 16): wholesale power +76% YoY in early 2025, capacity costs +~400%; 220 GW of interconnection applications against a 154 GW summer peak; 7+ years to energize. Gov. Shapiro filed a federal complaint over "the largest unjust wealth transfer in the history of US energy markets" and threatened a PJM exit; AEP signaled it may leave too.
  • ERCOT (researcher Joshua Rhodes, Renewable Rides, Jun 16): the queue ballooned from 220 GW (70% data centers) to 435 GW (90% data centers), versus an all-time peak delivery of 85.5 GW. Transformers are up 200%, switchgear 180%+, growth running 40–50%/yr vs. a 2.5% historical norm. His read: rate pressure is locked in regardless of whether the bubble bursts.
  • Transmission (ETCC chair Paul Kjellander, Factor This, Jun 18): EEI sees transmission spend at 3x the 10-year average; competitive procurement delivered 38% cost reductions vs. 84% overruns on non-competitive projects, a policy lever to watch.
  • Nuclear (Southern Co's Tom Fanning, SunCast, Jun 19): a 20 GW national target by 2040, Meta's 10 GW Ohio proposal, and a live SMR field (X-energy, TerraPower, GE Hitachi BWRX-300, Holtec, Oklo, Radiant). Behind-the-meter is the near-term workaround, Verse's $54M Series B (Supercool, Jun 18) targets shrinking 7-year grid waits to 1–4 years via on-site storage.

6. The bear whisper

Worth holding in tension with all of the above: macro strategist David Woo on Monetary Matters (Jun 15) argued the five hyperscalers' combined capex actually declined QoQ and YoY in Q1 2026, the nominal "growth" being component inflation, not real capacity. Forward Guidance (Jun 19) cited JPM data showing capex YoY growth decelerating from ~80% to ~45%. And per Morgan Stanley on Thoughts on the Market (Jun 18), the buildout is now debt-funded, $250B AI-related issuance YTD, ~$500B expected for 2026, with construction risk the dominant near-term concern. If real capex is inflation rather than units, the electrical-equipment longs face a pricing-vs-volume question into 2027.

Operator vs. pundit scorecard: the operators (Semenuk, Wolf, Spencer, Fanning, Verse) are unanimous that the spend is happening and bottlenecked by permits, power and supply-chain inflation. The pundits (Woo, MS, the capex skeptics) are doing the useful work of questioning whether the dollars equal capacity. Our bias: the most defensible longs are the equipment and electrical-infrastructure suppliers with pricing power into a supply-constrained buildout (transformers, switchgear, machine vision, automation) rather than the hyperscaler capex line itself, and the cleanest near-term catalysts are the pending copper/pharma Section 232 determinations and the competitive-transmission policy fights.