Teardown

Digest · 2026-09-14

Scan #052: four route-based, headcount-scaled, distributor-locked incumbents whose organic growth stopped compounding — NCR Voyix bleeding Aloha to Toast at a ~$1.3B market cap that is 90% below its pre-spin implied value, Rockwell Automation two years into guidance cuts as Siemens Xcelerator and Emerson-AspenTech vertically integrate around it, BrightView taking three years and a $500M One Rock rescue just to stop shrinking, versus Cheniere Energy — the day's one well-positioned incumbent — printing $19.98B FY2025 on ~95% take-or-pay-contracted LNG capacity; and four emerging companies (three still fighting, one already dead) whose wedge is either an AI-native replacement of an incumbent's core workflow (hyperexponential specialty pricing), a physical-asset arbitrage the incumbent will try to buy rather than out-execute (KoBold Metals AI-first minerals discovery, Airspace time-critical logistics orchestration), or — Diamond Age's post-mortem lesson — a robotic homebuilder that took construction risk and got destroyed by the housing cycle before its own pivot could save it

Eight companies where the incumbent's operating model has become the attack surface. Four legacy incumbents — three at-risk (NCR Voyix, Rockwell Automation, BrightView Holdings) and one well-positioned (Cheniere Energy) — whose route-based, distributor-dependent, or capex-heavy models are colliding with AI-native and vertically integrated attackers. Four emerging companies (KoBold Metals, hyperexponential, Airspace, Diamond Age — the last a Dec-2024 post-mortem on construction-risk-bearing robotic homebuilding), each carrying a specific mechanism-level question about whether an AI-first workflow can compound faster than the incumbent's owned-asset moat.

Today’s eight sit on the same fault line: what happens when an incumbent’s operating model — the distributor channel, the route density, the installed base of an on-premise POS — becomes the thing an AI-native or vertically integrated attacker can price against. Three of the day’s four incumbents (NCR Voyix, Rockwell Automation, BrightView Holdings) are printing improving quarterly numbers even as their underlying competitive position erodes; the fourth (Cheniere Energy) is the counterexample where contracted cash flow has genuinely compounded the moat rather than just deferred a reckoning. The four emerging companies are each a specific bet on whether an AI-native workflow (hyperexponential’s Python-native specialty-insurance pricing platform), an AI-native physical-asset play (KoBold’s ML-guided minerals discovery), an algorithmic arbitrage (Airspace’s time-critical logistics dispatch), or — as Diamond Age’s Dec-2024 bankruptcy shows — a construction-risk-bearing robotic homebuilder can reach durable scale before the incumbent or the cycle catches it.

Cheniere Energy — Energy / LNG · Incumbent, well positioned. Houston-based largest US LNG exporter, running Sabine Pass, Louisiana and Corpus Christi, Texas toward a ~60+ mtpa run-rate by 2028, with ~95% of nameplate production locked under 20-year take-or-pay Sale and Purchase Agreements. FY2025 revenue $19.98B (+27.2% Y/Y), net income $5.3B; Q2 2026 revenue $5.73B versus $4.92B consensus; FY2026 adjusted-EBITDA guidance raised to $7.9-8.4B and distributable-cash-flow guidance to $5.3-5.8B; market cap ~$55B on ~$25-26B of debt. The day’s one well-positioned incumbent because the moat is documented rather than assumed: the contracted-cash-flow structure Cheniere invented in 2012 (SPAs signed before construction, project debt raised against them) is now the template every US LNG project copies, and Corpus Christi Stage 3 Train 1 was delivered six months ahead of guarantee in March 2025 while Venture Global’s identical Plaquemines project is losing arbitrations to BP and Shell. The fair critique is that the 2026-2028 supply wave — Qatar’s 48 mtpa North Field expansion, Golden Pass, Plaquemines, Rio Grande LNG and Port Arthur all racing online — will collapse spot margins for any un-contracted volumes, so incremental growth beyond Trains 8 & 9 depends on Cheniere signing new SPAs into a buyer’s market.

Rockwell Automation — Industrial automation · Incumbent, at risk. Milwaukee-based US industrial-automation leader with 120 years of Allen-Bradley PLC lineage — ControlLogix and CompactLogix PLCs, FactoryTalk software, Plex (acquired $2.22B 2021), Fiix, Kalypso, Otto Motors, Verve Industrial. Q3 FY2026 was a real bounce, but the underlying story is two years of guidance cuts, an “intelligent devices” segment whose growth has decelerated for six straight quarters, and a software strategy where Plex ARR has never been separately disclosed for a reason. Structurally at risk because Siemens Xcelerator is winning enterprise-standardization deals (Ford’s global body-shop, JLR, Airbus) at the exact scale Rockwell needs to defend, Emerson’s $8.2B AspenTech privatization completing 2025 gives Emerson a vertically integrated industrial-software stack Rockwell no longer has, and the ~500-distributor channel that used to be a moat now looks like a two-hop tax versus Siemens’s direct-to-enterprise model. Blake Moret’s ninth year is compounding the pattern, not fixing it.

NCR Voyix — Retail · Incumbent, at risk. Atlanta-based retail-and-hospitality POS + digital-commerce platform, spun from NCR Corporation on 16 October 2023 (VYX = retail/hospitality; NCR Atleos = ATMs). Sold Digital Banking to Veritas Capital for $2.45B in Sept 2024 to pay down debt. James G. Kelly (ex-EVO Payments CEO) took CEO on 4 February 2025. Aloha POS installed base (~100k+ restaurants historically) is churning to Toast and PAR; enterprise retail (Kroger, Walgreens legacy) is losing to GK Software and cloud-native rivals; the April 2023 Aloha ransomware outage that took Buffalo Wild Wings and other chains offline is a reputational tail carriers still remember; market cap ~$1.3B, roughly 90% below the pre-spin implied value. The Digital Banking divestiture bought balance-sheet relief but not a growth story — Q2 2026 revenue was still down ~7% Y/Y on organic terms. Take-private optionality is high; the equity is trading like an option on a whole-company sale.

BrightView Holdings — Commercial services / Landscaping · Incumbent, at risk. Blue Bell, PA-based largest US commercial landscaper, formed by KKR’s 2013 Brickman + 2014 ValleyCrest combination, rebranded 2015, IPO’d 2018 at $22 raising $470M. Dale Asplund (ex-United Rentals COO) arrived Sept 2023 with a $500M One Rock convertible-preferred alongside the appointment; three years and a full segment repricing later, FY2025 revenue is $2.673B (down from $2.816B FY2023) but adjusted EBITDA at $352.3M is a company record — so the turnaround worked on margin but not on the top line. The structural at-risk case is unchanged: a 40-year-unchanged route-density model with ~250 branch locations and ~20-22k H-2B-dependent field employees, in a $130B fragmented US TAM where the top ten hold <15% share, is a target for autonomous-mowing (Scythe Robotics, Greenzie) and SaaS-enabled route-optimizers (Aspire, now ServiceTitan-owned) — not a compounder. Market cap ~$1.03B on 11 Sept 2026 says the same thing.

KoBold Metals — Energy / Critical Minerals · Emerging. Berkeley-based AI-first critical-minerals prospector founded 2018 by Kurt House (Harvard applied physics PhD, ex-Khosla), Josh Goldman (President; Harvard physics, ex-McKinsey) and Jeff Jurinak (COO; 35 years at Conoco/ConocoPhillips). Jan 2025 Series C of $537M at $2.96B post-money led by T. Rowe Price and Durable Capital Partners; cumulative $1B+ raised across Breakthrough Energy Ventures, a16z, Bill Gates, Bezos, Ray Dalio, BHP, Equinor. The Mingomba copper-cobalt discovery in Zambia (80/20 with ZCCM-IH, targeting 300-500kt Cu/year from early 2030s) is the marquee — but the widely cited 247Mt at 3.64% Cu figure is KoBold’s own internal estimate, with no independently audited JORC/NI 43-101 resource statement yet. The open question is not whether ML helps target drilling (it clearly does at the margin) but whether KoBold can retain enough equity through the $2.3-2.5B mine build to compound the data moat into operating economics — or whether the majors that are already limited partners end up buying the deposits at cost.

hyperexponential — Insurance · Emerging. London-based pricing and (as of July 2026) agentic-underwriting SaaS for specialty insurers, founded 2017 by Amrit Santhirasenan and Michael Johnson out of Tokio Marine Kiln. Series B of $73M led by Battery Ventures with a16z closed January 2024, ~$91M cumulative disclosed funding, Marcus Ryu (Guidewire co-founder) joined the board. Product hx Renew is Python + Jupyter + Git-native, targeting the Lloyd’s/specialty pricing seat that WTW Radar has owned for 30 years. July 2026’s launch of hyperoperator repositions the company from pricing platform into full submission-to-bind agentic underwriting — a much bigger swing that puts them into competition not just with Akur8 and Earnix but with Cytora, Sixfold, Concirrus and every startup selling underwriting-workbench automation. Falsifiable test for winning: 5+ top-20 Lloyd’s syndicates displace Radar as primary pricing engine, ARR crosses $100M, US commercial P&C customers clear 20+. Falsifiable test for losing: Akur8 lands 3+ Lloyd’s syndicates first, Guidewire or Duck Creek ships a native competitive engine, or YoY growth decelerates below 40%.

Airspace — Logistics · Emerging. Carlsbad, CA-based time-critical shipment orchestration platform for organ transplants, aircraft-on-ground parts, clinical-trial samples and semiconductor equipment logistics. Founded 2016 by Nick Bulcao and Ryan Rusnak; $138M cumulative funding across four rounds through the Dec 2022 Series D of $70M (Telstra Ventures, DBL Partners) — no round since. Real revenue is undisclosed, but headcount is ~500 as of 2024-2025 and reported ARR is $200M+. The unique wedge: an algorithmic dispatch engine that arbitrages across commercial airline belly-space, next-flight-out cargo, on-board couriers and ground courier networks, in <60-second quotes. The open question: capital-light routing without owned capacity is only as good as the belly-space commercial airlines make available — capacity Airspace does not own — while vertically integrated incumbents (Kuehne+Nagel’s Quick Group, UPS Healthcare’s Marken/MNX, Cencora’s World Courier) own their own aircraft and cold-chain assets. Either Airspace out-executes them to IPO scale, or one of them acquires it before it needs to answer that question standalone.

Diamond Age — Construction / Homebuilding · Emerging (post-mortem). Phoenix-based 3D-printed-and-robotic production homebuilder founded 2018 by Jack Oslan (ex-Kodak, ex-Boeing 3D printing) and Russell Varone. Trucked a gantry-based concrete printer to Arizona lots and printed ~30 production houses for Century Communities’ Century Complete brand in Casa Grande, AZ between 2022 and early 2024 on ~$58M cumulative funding ($8M seed Aug 2021, $50M Series A Mar 2022 led by Prime Movers Lab). Pivoted to light-gauge-steel panels in mid-2024 when the printer economics didn’t scale — then ran out of capital eight weeks after the pivot proved itself, filing for bankruptcy on 12 December 2024 and auctioning its Fanuc robots via Silicon Valley Disposition in January 2025. Included here as a post-mortem: the open question Diamond Age answered in the negative is whether ANY construction-risk-bearing, vertically integrated robotic homebuilder can survive a housing-cycle trough. The category winners (ICON, at a claimed $2B and now shifting from building homes to selling printers) and the equipment-only OEMs (COBOD, Peri, CyBe) that never took the construction-risk are the ones still standing.

Full pages linked above. Yesterday’s scan: Scan #051.

Full deep dives