Teardown

Daily digest · 2026-08-10

Scan #030: What breaks the model, and what compounds it

Four incumbents whose economics are either compounding on the AI-power build-out or bleeding to entrants who don't need the sales rep — and four challengers each betting a specific mechanism dislodges an entrenched way of doing business.

Two questions run through today’s eight. The first is whether AI actually rewires an operating model or just cuts a line item — Dili on Davis-Bacon paperwork, ZyG on the DTC agency stack, Pace on the insurance back office, RobCo on the SMB factory floor. The second is whether structural advantage compounds or corrodes under it — AutoZone and Duke Energy compounding on density and rate base, Werner and Sysco corroding under autonomous freight and disintermediated ordering. The picks are deliberately asymmetric: the emerging companies name a mechanism, and the incumbents force a verdict.

Dili — Construction / Infrastructure · Emerging. YC S23 outfit that reads certified payroll reports and enforces Davis-Bacon and prevailing-wage rules on the IRA/CHIPS/IIJA build-out; $21.7M in from Khosla on July 30 with 700+ projects and $1.4B of wages already processed. The pitch is that state DOLs still audit CPRs on manual six-week cycles while every federally funded project must now pass wage documentation to keep its tax credit. The mechanism-level question isn’t traction — it’s whether federal and state auditors accept AI-generated CPR review as evidence in a Davis-Bacon audit, or whether Dili is stuck selling time savings rather than legal cover.

ZyG — Ecommerce · Emerging. Ex-ironSource founding team (Bar-Zeev, Kaplan, Ashkenazy) plus three AI leads from Israel’s Unit 81 raised $60M Series A at a $500M valuation from Accel on May 5, less than a year after founding, pitching an “agentic OS” that runs DTC ecommerce end-to-end. The uncomfortable comparison is the $10B+ of aggregator write-downs — Thrasio, Perch, Berlin Brands — that happened the last time a well-funded team tried to industrialise the DTC playbook. Best team ever assembled to attempt the AI-ecommerce-OS thesis, but the question is whether agents close the two gaps (brand equity, predictable CAC) that killed the aggregators, or just make the same mistakes cheaper.

Pace — Insurance · Emerging. $46M Series B co-led by Thrive and Sequoia at ~$375M (May 27), founded 2024 by Jamie Cuffe (ex-Retool, sold Cheer to Retool in 2020). Positions as “agentic process outsourcer” for P&C insurers — submission intake, policy servicing, FNOL — priced per completed transaction rather than BPO hourly. Real customer logos with published outcomes (Palomar 90%+ on three top service flows, Ryze 30% cycle-time reduction across 250,000+ live workflows). The live risk is Duck Creek launching its own insurance-native Agentic AI Platform on April 28 and acquiring Send: does a horizontal AI ops layer survive when the core suites bundle native AI into policy admin they already own?

RobCo — Supply Chain / Manufacturing · Emerging. Munich TUM spinout founded 2020 by Roman Hölzl (ex-KUKA, ex-Tesla); raised $100M Series C co-led by Lightspeed and Lingotto on January 29 at ~$500M+ per Bloomberg. Sells modular robot arms that snap together like Lego with a no-code layer, priced as RaaS to SMB manufacturers Universal Robots never fully cracked. Just absorbed the customer book of failed US peer Rapid Robotics. The mechanism question is whether RaaS pricing survives when SMB manufacturers historically buy robots on five-year depreciation schedules — and when Formic already sets a ~$8-15/hour price floor.

Werner Enterprises — Logistics · Incumbent, at risk. 70-year-old Omaha truckload carrier (NASDAQ: WERN) whose Q2 2026 GAAP operating margin collapsed to 1.8% from 8.8% and EPS to $0.11 from $0.72 — even after the January FirstFleet acquisition padded revenue by 24% YoY. One-way ex-fuel revenue was still down 15.9% in the quarter; organic revenue has declined three straight years ($3.29B FY22 → $2.97B FY25). The at-risk case rests less on numbers and more on Kodiak (100+ commercial driverless loads for Atlas since December 2024), Waabi (driverless target end-2026 on Uber Freight’s Dallas-Houston lane), and Aurora (Werner’s own Fort Worth-El Paso pilot partner) beginning to commercially haul on the exact Texas lanes Werner staffs with 10,500 drivers.

AutoZone — Retail · Incumbent, well positioned. ~$18.9B FY2025 revenue across ~7,850 stores, ~130,000 employees, with the most disciplined buyback machine in retail — ~89% of shares retired since 1998 against ~$42.2B in cumulative buyback authorisations. The moat isn’t brand or price; it’s density — 156 mega-hubs carrying ~110,000 SKUs each feed satellite stores multiple times a day and deliver rare parts to a repair bay in under 30 minutes across most metros. Commercial (DIFM) accounts sit inside PROSPRO, warranty labour reimbursement, and 30-day credit terms — hard to rip and replace. O’Reilly runs a different capital model (higher retained capital, higher multiple, ~49% commercial mix) and both work; Advance is closing ~700 stores and giving the share away.

Duke Energy — Energy · Incumbent, well positioned. ~$33B regulated utility (NYSE: DUK) serving ~8.6M electric customers across the Carolinas, Florida, Indiana, Ohio, Kentucky. Q2 2026 adjusted EPS $1.43 (vs $1.25 YoY); reaffirmed 5-7% EPS CAGR through 2030 on an industry-record $103B five-year capex plan; 7.8 GW of signed data-center ESAs with 15.4 GW more expected by mid-2027. The North Carolina rate case settled at a 9.8% ROE with earnings sharing to 10.3%. Because Duke’s territory covers three of the top-five US data-center corridors, every gigawatt of hyperscaler load converts almost mechanically into rate base at a return investors pay ~20x for. Real risks: behind-the-meter bypass (Amazon/Talen), a $10B equity issuance overhang 2027-2030, and unhealed coal-ash scars from Dan River (2014).

Sysco — Supply Chain / Foodservice Distribution · Incumbent, at risk. The largest US foodservice distributor (NYSE: SYY, ~$84.6B revenue, ~14,000 tractors) whose share of US broadline foodservice has slid from a peak near 75% in the 1990s to roughly 32% today. Gross margin compressed again in Q4 FY2026 (down 17 bps to 18.7%) as mix shifted to lower-margin national accounts. Melius went to Sell in July; Citi cut price target to $72. US Foods is taking metro/chain share; PFGC dominates C-store and specialty via Vistar and Reinhart; Gordon owns Midwest healthcare/education. And underneath them, Choco/Notch/Cheetah/Restaurant365 are letting independent restaurants order around the DSR sales rep the entire tractor-and-warehouse cost model was built to fund.

Full deep dives