Teardown

Digest · 2026-09-18

Scan #055: four incumbents whose defence-of-margin playbook has now become the wedge — Sotheby's needing a $909M ADQ rescue after FY2024 sales collapsed 23%, UGI's Pennsylvania rate base subsidising an AmeriGas propane franchise losing six-figure customer counts to electrification, Trex's Home-Depot/recycled-poly moat outflanked by an AZEK now inside James Hardie's ~$11B distribution machine, and BigCommerce (Commerce.com) trading at ~2.5% of Shopify's mcap on 3% growth — against four emerging companies each betting a specific mechanism outruns the incumbent bundle: Nowports betting Nowports Capital financing attach reaccelerates net take-rate before TriumphPay-style rails or a Nuvocargo bundle catch it, Physical Intelligence betting a single π-series VLA foundation model compounds cross-embodiment before Skild AI or Nvidia GR00T ship native, Ascend Elements now in Chapter 11 with a May 21 2026 Section 363 sale hearing testing whether Bluegrass Infrastructure Partners can finish Apex 1's pCAM line, and Prewave whose €63M Series B thesis depends on the EU's Feb 2025 CSDDD scope-cut not gutting the regulatory tailwind

Eight companies. Four incumbents whose margin-defence playbooks are now the exposed wedge — Sotheby's post-2019 LBO needing a $909M ADQ rescue after FY2024 sales fell 23%, UGI's Pennsylvania rate base cross-subsidising an AmeriGas propane franchise structurally losing to electrification and nat-gas conversions, Trex's ~$4.4B market cap looking exposed against a James Hardie / AZEK combined ~$11B distribution book, and BigCommerce (renamed Commerce.com) trading at ~2.5% of Shopify's mcap on FY2025 growth collapsed to 3%. Four emerging: Nowports' 2023-2025 retrenchment against Nuvocargo + TriumphPay, Physical Intelligence at $5.6B on the π-series generalist VLA bet, Ascend Elements' April 2026 Chapter 11 filing and May 21 Section 363 sale hearing, and Prewave's €63M Series B thesis exposed to the February 2025 EU Sustainability Omnibus scope cut.

The thread through today’s eight is that four legacy franchises each ran the same playbook — buy your way into scale, use that scale to defend a margin the market has now started to reprice — and none of the four is still compounding. Sotheby’s 281-year-old brand has been carrying a 2019 leveraged buyout by Patrick Drahi’s BidFair, and the FY2024 sales collapse (total consolidated sales $6.0B, -23% YoY per The Art Newspaper January 2025) forced a $909M rescue from Abu Dhabi’s ADQ and a $825M April 2026 bond refinance at ~8.5% — even after H1 2026’s record $4.4B sales, the interest bill and the Christie’s / private-sale share loss remain. UGI’s regulated Pennsylvania gas rate base is fine; the problem is AmeriGas, whose FY2026 EPS guide was cut to $2.75-$2.90 in August 2026 (from $2.90-$3.15) as customer counts, distributor volumes and margin per gallon all keep sliding under the twin pressures of heat-pump conversion and natural-gas hookups. Trex sits at ~$4.4B market cap (September 17, 2026) after a 27% one-day drop on the Q3 2025 print; AZEK, its main tier-1 rival, now sits inside a James Hardie balance sheet ~2.5x its own with mass distribution and a builder-specification motion Trex’s homeowner-DIY channel does not match. And BigCommerce — now trading as Commerce.com after the July 2025 rebrand — has cut two rounds of staff, ended FY2025 with $359M ARR up just 3% YoY per the February 2026 release, and sits at roughly 2.5% of Shopify’s public-market cap on the same total customer count Shopify processes in a quiet week. Against them: four emerging companies each betting a specific mechanism widens the wedge before the incumbent bundle catches it or the balance sheet gives up — the questions are whether Nowports Capital’s financing attach can lift net take-rate above pure-brokerage benchmarks before a Series D becomes required, whether Physical Intelligence’s π-series VLA compounds cross-embodiment before Skild AI, Nvidia GR00T or Google DeepMind ship native, whether Ascend Elements’ court-supervised Section 363 sale (bid deadline May 9, 2026; hearing May 21, 2026) produces a going-concern buyer that finishes the Kentucky pCAM line, and whether Prewave’s €63M Series B thesis survives the February 2025 EU Sustainability Omnibus that pushed CSDDD transposition to July 2027 and cut in-scope companies to those with >5,000 employees and >€1.5B turnover.

Sotheby’s — Retail / Luxury auction · Incumbent, at risk. The 281-year-old auction house, owned by French-Israeli telecoms billionaire Patrick Drahi’s BidFair since a $3.7B LBO closed October 2019 (Robb Report, June 2019), employs ~1,700 globally and cleared $8.0B in FY2023 total sales before FY2024 collapsed 23% to $6.0B (The Art Newspaper, January 2025). The uncomfortable finding is not the FY2025 partial recovery (total sales $7.1B, +18%) or H1 2026’s record $4.4B — those are cyclical — but the ~$909M convertible preferred and warrants ADQ (Abu Dhabi) had to put in August 2024 alongside additional Drahi equity, a 150+ headcount cut, an S&P downgrade to B-, and the April 2026 $825M bond issued at ~8.5% just to refinance the October 2027 maturity. Christie’s (privately held by François Pinault’s Artémis) does not carry the leverage, and its private-sale + online auction growth is taking share Sotheby’s needs the brand for. The attack surface: a Sotheby’s-quality authentication + condition-report layer decoupled from the auction commission, sold to the private-sale channel Christie’s, Phillips and Bonhams already run in-house.

UGI Corporation — Energy · Incumbent, at risk. NYSE: UGI. King of Prussia, PA. Multi-utility with three main books — regulated Pennsylvania natural gas + electric distribution (UGI Utilities), the largest US propane distributor (AmeriGas), and European LPG (UGI International, including Antargaz France and Flaga) — plus midstream. FY2024 revenue $7.64B (UGI 10-K, November 2024); market cap ~$8.1-8.3B (September 2026 per StockAnalysis) with total debt across the three financing entities in the $6-7B range against ~$530M FY2025 free cash flow. The regulated utility and European LPG are the good parts of the book; the problem is AmeriGas, whose customer counts, volumes and margins per gallon are all structurally shrinking as US households electrify heating or convert to natural gas. The August 2026 guidance cut on FY2026 adjusted EPS to $2.75-$2.90 (from $2.90-$3.15) is the tell. Standard General’s 2023 activist campaign forced a strategic review that ultimately opted to retain AmeriGas — but every year of retained propane exposure is another year the good rate base cross-subsidises a fuel losing to heat pumps and gas conversions on the same neighbourhood, and BBB / Trustpilot complaints on cylinder-swap pricing and cancellation friction now compound the customer-loss curve.

Trex Company — Construction · Incumbent, at risk. NYSE: TREX. Winchester, VA. The ~30-year composite-decking pioneer, ~380M lbs of grocery-bag film pulled out of the waste stream annually to make WPC boards, FY2025 revenue $1.17B (flat vs $1.16B FY2024) at 42.2% gross margin, market cap ~$4.38B (September 17, 2026) after a 27% one-day drop on the Q3 2025 print and a ~32% YTD decline. The at-risk verdict is not about the product — Trex Signature, Transcend and Enhance still hold ~40%+ composite-decking share by trade-press estimate — but about distribution: AZEK, now inside James Hardie’s ~$11B distribution and channel machine after the mid-2025 close, can specify decking into the builder-facing motion Hardie already runs, and Trex’s homeowner-DIY channel through Home Depot and Lowe’s does not attach as tightly to the new-build cycle. The attack surface is not another integrated compositor; it is a China + Vietnam capstock-decking direct-import brand plus a Shopify-native TrexPro-style installer network that skips both the big-box shelf and the AZEK / James Hardie builder-spec dependency.

BigCommerce (Commerce.com) — Ecommerce · Incumbent, at risk. NASDAQ: BIGC (renamed to Commerce.com in July 2025 with the same ticker). Austin, TX. Public ecommerce SaaS platform, IPO’d August 2020 at $24/share (peak $162.50), now trading around $5-6 for a market cap approximately $380-450M — roughly 2.5% of Shopify’s public-market cap on FY2025 revenue $340.6-345.6M (guided; Q3 2025 print $86M) growing just 3% YoY, and total ARR $359.1M at Dec 31, 2025 (+3% YoY) with Enterprise ARR $287.2M (+10%, 80% of total). The uncomfortable finding: the Shopify Plus / Adobe Commerce squeeze on both ends is now visible in the numbers — Shopify Plus continues to take mid-market SMBs, Salesforce Commerce Cloud + Adobe Commerce hold the enterprise top-end, and the July 2025 rebrand plus AI-agentic / composable narrative has not restarted top-line growth. Two years of layoffs (workforce reduced by ~15% in 2023 and further in 2025), CEO Travis Hess into his second year defending guidance, and Lynrock Lake’s Schedule 13G disclosure is the only external pressure signal so far. The attack: sell headless commerce as a runtime rather than a subscription, and win the composability narrative Commerce.com is now defending from a smaller balance sheet than the buyer expects for a mission-critical stack.

Nowports — Logistics · Emerging. Monterrey, Mexico. Digital freight forwarder for Latin America, founded 2018 by Alfonso de los Rios (CEO, ex-MIT / Y Combinator W19) and Maximiliano Casal (co-founder, COO through late-2023). ~$242.7M cumulative equity through the May 2022 SoftBank-led Series C at $1.1B post-money (TechCrunch / Bloomberg Línea) — the first LatAm logtech unicorn — but no announced round since, and 2023-2025 retrenchment (Peru, Colombia, Brazil office closures per LinkedIn / Glassdoor threads) plus Casal’s COO departure suggest the mark is under pressure. Reported 2024 revenue of $397.7M in Getlatka is almost certainly gross freight billings, not net brokerage margin — the company’s 2022 net revenue was ~$50M. The falsifiable open question: does Nowports Capital’s factoring / trade-finance attach reaccelerate net take-rate materially above pure-brokerage benchmarks before a Series D or debt facility becomes necessary, especially with TriumphPay’s April 2024 peso rails switched on and Cargado / Nuvocargo bundling the same USMCA payables layer at the shipper side? Nearshoring tailwinds help; Cargado + Nuvocargo taking the AP integration back does not.

Physical Intelligence — Robotics / Foundation models · Emerging. San Francisco, CA. Founded March 2024 by Karol Hausman (ex-Google Robotics), Sergey Levine (UC Berkeley), Chelsea Finn (Stanford), Brian Ichter, Suraj Nair, Lachy Groom and a partner cohort largely from Google DeepMind Robotics + Berkeley robot learning. ~$1.07B disclosed across three rounds: $70M seed (March 2024), $400M Series A at $2.4B post (November 2024, Thrive + Lux + Bezos + Sequoia), $600M Series B at $5.6B post (November 2025). Bloomberg reported (March 2026) a further ~$1B round at ~$11B in discussion — not confirmed closed. Product: a single vision-language-action “π-series” foundation model (π₀, π0.5, Pi-Zero-Fast, transformer + flow-matching action head), trained on 8+ heterogeneous robot embodiments plus a proprietary corpus “orders of magnitude” larger than the Open X-Embodiment academic set. Pre-revenue; released model weights (subset) in 2025. The falsifiable open question: does a single generalist π-series policy compound cross-embodiment before Skild AI ($4.5B post-money 2025), Nvidia GR00T, Google DeepMind Gemini Robotics, or Figure AI ship native VLAs at similar scale — and does the proprietary data corpus actually stay proprietary, or does an open-source Open-π equivalent (an academic-led effort, or Meta / DeepMind releasing weights) commoditise the moat before the first commercial ARR is booked?

Ascend Elements — Energy · Emerging. Westborough, MA. Lithium-ion battery recycler and cathode active material (CAM) producer using “Hydro-to-Cathode” direct-precursor synthesis. Founded 2015 as Battery Resourcers by Eric Gratz and Yan Wang (WPI); renamed 2022. ~$1.1B+ cumulative equity through Series D + $480M in DOE grant awards ($164M cancelled Feb 2025, another $110M rescinded October 2025 after ~$206M disbursed). Peak valuation ~$1.5B per debtors’ first-day declaration (Southern District of Texas, April 2026). The uncomfortable finding: Ascend Elements filed Chapter 11 on April 9, 2026 with Bluegrass Infrastructure Partners Holdings as stalking-horse bidder; Section 363 sale bid deadline May 9, 2026, hearing May 21, 2026. The IRA 45X sensitivity always loomed — Trump-era changes to Section 45X credits and the $480M grant rescissions closed the runway before Apex 1’s pCAM line in Hopkinsville, KY reached commercial scale. Base 1 (Covington GA) has processed feedstock commercially since Aug 2022 and AE Elemental (Poland) since Sept 2024, so the recycling backbone runs; the higher-margin CAM/pCAM revenue was pre-commercial at filing. The falsifiable open question is now the outcome of that 363 sale: does Bluegrass or a topper bidder actually finish Apex 1 and run Hydro-to-Cathode at commercial rate, or does the estate get sold for the recycling assets and the intellectual property while Redwood Materials and Chinese CAM incumbents own the market Ascend was built to attack?

Prewave — Supply chain · Emerging. Vienna, Austria. AI supply-chain risk platform, founded 2017 as a TU Wien spin-off by Harald Nitschinger (CEO) and Dr. Lisa Smith (CTO). ~€100M ($110M) cumulative through Series B: €63M / ~$67.5M Series B in June 2024 led by Hedosophia with Creandum, Ventech, KOMPAS, Speedinvest and Working Capital Fund following (the “PT1 + DN Capital” I had earlier is not verified — official press names only the above). Enterprise references include Hilti, BMW, Kärcher, DPDHL and Continental; getLatka pegs 2025 ARR at ~$43.9M as an unverified third-party estimate. Product monitors 150M+ global sources in local languages for ESG, disruption, geopolitical, financial and cyber risks, with tier-N supplier mapping and CSDDD / LkSG-aligned reporting. The uncomfortable finding: the February 2025 EU Sustainability Omnibus pushed CSDDD transposition to July 2027 and application to July 2029, cut scope to companies with >5,000 employees and >€1.5B turnover (from the original 1,000-employee / €450M threshold), and softened tier-N due-diligence expectations. Prewave’s Series B was priced on a regulatory tailwind that just got weakened. The falsifiable open question: does the enterprise wedge — Hilti-scale F500 references + local-language NLP coverage no US-first competitor matches — compound into cross-functional ERP-adjacent contracts before SAP Ariba, Coupa Compass and Ivalua Risk Center ship native tier-N risk modules that come free with the source-to-pay contract the customer already owns?

Full deep dives