FIELD GUIDE · NINE KINDS OF HARD CHEMISTRY, AND WHAT A BUYER PAYS FOR
Custom Chemistry Makers
A custom chemistry plant is worth more than its reactors when it holds a permit, a skill, and a qualified customer that a buyer can't rebuild quickly: about a year at the fast end, and years or decades for a fab, implant or defense approval. I sort the US makers here by that skill. End markets drive demand, but those three things set the price. Two demand stories run through it: the US rebuilding its defense chemicals, and the AI build-out pulling on chip materials and cooling fluids.
Nine skill segments, read through the three buyers that share the same reactors: pharma, industrial tech, and agro. Built from permits, filings, releases, and the deal tape. Current as of October 11, 2026. This preview is ten sheets of the 114-sheet guide, and every sheet number matches the full guide.
OCTOBER 20261
AGA·DWG / SHEET 3 · FROM THE NOTEBOOK
FROM MY NOTEBOOK · WHAT I'M TRYING TO FIGURE OUT
This is a page from my research notebook. I'm trying to work out what a custom chemistry plant is worth beyond its reactors.
From the outside, fine chemistry, pharma contract manufacturing, and toll plants look like one market. They share reactors, chemists, and vocabulary. I've spent a lot of my working life around plants like these, and I kept noticing that buyers don't pay for them the same way. So I started writing down why.
THE QUESTIONS I'M WORKING THROUGH
What can't a buyer rebuild?
Is it the permit, the skill, or the qualified customer? Does a US address earn anything on its own? What happens to a permitted site when the rules shift, or when the equity runs out first?
WHY I'M WRITING IT DOWN
In the open, so you can check it
Curiously Optimistic is my open notebook on chemicals, materials, growth, teams, and AI in industrials. Every number here comes from a filing, a register, a permit, or a release, dated on the sheet. Nothing from a client engagement shows up.
HOW TO READ IT
Start with the map
Sheet 4 is the plain-English primer. The grid (sheet 11) and the dashboard (sheet 14) are the map, and my ranking (sheet 18) is my ranking. Each of the nine segments runs in the same order and ends with my call.
WHERE I MIGHT BE WRONG
I'm curiously optimistic about US custom chemistry, and I might be wrong on any call in here. Where a number is soft, I say so. If you see something I've missed, tell me which sheet and what the record says. When I change a call, I'll say which one and why.
Source: my notes, Oct 11, 2026. The questions and the reading order are mine, not a document. Disclosure: I also work with the U.S. Critical Chemistry Coalition.
August Grace Advisory · curiouslyoptimistic.com3
AGA·DWG / SHEET 6 · THE GUIDE ON ONE PAGE · AS OF OCT 11, 2026
THE GUIDE ON ONE PAGE · MY GRADE FOR EACH SKILL
The whole guide on one page. My top grades go to permits nobody can rebuild quickly.
Sources: grades and comparables from sheet 18; step contents and money dots from sheet 5, my read of the deal and capex tape on sheets 32, 41, 46, 53, 59, 69, 78, 86 and 94; the middle-step hypothesis from sheet 13. August Grace read, Oct 11, 2026 (judgment, not a document). CASE is coatings, adhesives, sealants and elastomers; row 5 covers the makers of the molecules behind them and the formulators (sheet 57).
August Grace Advisory · curiouslyoptimistic.com6
AGA·DWG / SHEET 7 · TEN YEARS OF DEALS · AS OF OCT 13, 2026
TEN YEARS OF DEALS · ONE ROW PER SKILL, SIZED BY PRICE
60 deals in ten years. A buyer or seller stated a multiple on only ten of them.
Sources: segment tapes on sheets 32, 41, 46, 53, 59, 69, 78, 86 and 94, and the cross-segment tape on sheets 99 and 100, each row dated to its own release or filing. M&A rows only; capex, grants, Army awards and pre-2016 deals left out. Bands use the stated currency without FX. Curium is dated 2026 only and drawn mid-year. The bottom lane holds deals that appear only on sheets 99 and 100. Gelest and Pressure Chemical sit on two lanes and count once.
August Grace Advisory · curiouslyoptimistic.com7
AGA·DWG / SHEET 18 · MY RANKING · AS OF OCT 11, 2026
MY RANKING · WHERE THE EVIDENCE AND THE DEAL RECORD POINT
Here's where the evidence and the deal record point the money. The permit and the qualified customer earn the premium. Reactors don't.
WHERE
THE COMPARABLE BEHIND IT
CONFIDENCE
REVENUE BAND A SPONSOR WOULD HUNT (MY READ)
KEY RISK
1
A US hazardous-chemistry site with a live permit (6): nitration, phosgene, cyanide, energetics
Adjacent. Product makers printed: Orica 7.5x for Cyanco, NewMarket ~$700m for AMPAC. Jayhawk (~9x, buyer-stated) is the one multipurpose print
High on the permit; medium on price
$30m to $150m; Jayhawk's ~$78m sits inside it
The risk case is VanDeMark: phosgene at Lockport since 1951, a WARN for 69 jobs in Mar 2026 citing unfunded retooling and debt, then a sale at an undisclosed price. The permit can outlive the equity
2
A Western fluorination or fluorine-reagent site (1), or a partner to an HF owner
None. No site sale has printed; the evidence is capex (Inhance's F2 plant, a Texas fluorination house tripling its site, Arkema's $150m HF plant)
Medium
$10m to $75m
Indian HF (Navin, SRF) undercuts the US reagent step before a first comp prints
3
DEA-registered or high-containment capacity in the US (8)
Direct. Siegfried paid "below 10x" for Noramco's US sites, its words
Medium
$50m to $250m; Siegfried guides ~$100m of 2026 sales from those sites
Quota: opioid classes roughly halved since 2016; stimulants could be next
4
An organometallics or high-pressure independent with Western customers (2, 4)
Adjacent. Grace 9.5x for Albemarle's fine chemistry sites; JM 13.3x and Indorama 6.3x for adjacent businesses; the independents sold undisclosed
Medium
$20m to $100m
A sponsor exit that prints at a reagent distributor's price
WATCH
A US silicon maker with a qualified medical or specialty position (3)
Adjacent. Momentive's ~$3.1bn take-private and Silbond's stated range; Gelest, NuSil, ICM and SiVance sold undisclosed
Medium
$10m to $75m
Almost nothing independent is left, and the rest prices against Chinese siloxane
WATCH
Small monomer and photoinitiator makers (5); fab-adjacent high purity (7)
Adjacent. H.B. Fuller under 6x for formulators; Fujifilm $700m for ~$360m of fab-acid sales
Low
$10m to $50m
CMMC ends up as paperwork and the bench never thins
AVOID FOR MID-MARKET ENTRY
Peptide or oligo capacity without a contract book (9). This grades API capacity; building blocks and intermediates can grade differently
Adjacent. Samsung's CHF 44.31 offer for PolyPeptide, no multiple; CordenPharma bought AmbioPharm undisclosed
Medium
Not a mid-market ticket: CordenPharma's build is €900m
Lilly and Novo insource; Bachem's commercial API fell 19.3% in local currency in H1 2026 while development grew
PASS
Capacity sold mainly on reactor time, with no permit or qualified customer behind it
Adjacent. EUROAPI sold Brindisi at €5m EV
High
Not a hunt
Price is set by Indian and Chinese capacity. Plenty of multipurpose plants do hold a permit and audited customers; those belong in rows 1 to 4
To be straight with you: no US independent hazardous toll site of the row-1 kind has printed a price. The permitted US sites that did print were strategic carve-outs (Albemarle's to Grace) and one Indian purchase (Jayhawk). Why modality isn't a flat avoid: with a contract book the economics work (Bachem's FY2025 EBITDA margin was 30.9%). A mid-market buyer can't get that book.
Sources: Orica, Feb 21, 2024; NewMarket, Dec 4, 2023; Anupam (NSE), Dec 9, 2025; Grace on Albemarle FCS, Feb 25, 2021; Siegfried, Jan 27 and Apr 24, 2026; JM on Catalyst Technologies, May 22, 2025; Indorama on Oxiteno, Apr 4, 2022; H.B. Fuller 2024 Annual Report; Entegris, May 10, 2023; Samsung Biologics offer for PolyPeptide, Jul 20, 2026; Evonik on the Silbond purchase, Feb 28, 2014; MPM Holdings completion 8-K, May 15, 2019; Milliken on SiVance, Oct 2011; CordenPharma, Jul 16, 2024; Bachem Annual Report 2025 and H1 2026 release, Jul 30, 2026; C&EN, Mar 13, 2026; Pulse 2.0, Jun 23, 2026; EUROAPI, Jul 29, 2026. Ranking, confidence and revenue bands are my judgment, not a document. Read it as where the evidence points, not as advice.
August Grace Advisory · curiouslyoptimistic.com18
AGA·DWG / SHEET 19 · MY RANKING, CHARTED · AS OF OCT 11, 2026
MY RANKING ON ONE CHART · HOW HARD TO COPY vs HOW GOOD THE PRICE EVIDENCE IS
My picks are the skills where the permit and the qualified customer are hardest to rebuild. Only one has a price paid for the same kind of site.
MY READ
This is the same claim as sheet 18: hard to copy means the permit and the qualified customer, not the chemistry on its own. Licensed handling is the one pick with a price a buyer stated for this kind of site (Siegfried, "below 10x"). Hazardous and the two pick-four rows lean on adjacent prints, and fluorine has none. Silicon is a watch: Momentive and Silbond sold near it, not on it. * Modality works with a contract book; a mid-market buyer can't get one (sheet 18).
Sources: rank and comparable from sheet 18; replication position read off sheet 16 (August Grace read, Oct 11, 2026; judgment, not a document); bubble counts are M&A rows on sheets 32, 41, 46, 53, 59, 69, 78, 86 and 94, 2016 to 2026. A deal on two tapes counts on both.
August Grace Advisory · curiouslyoptimistic.com19
AGA·DWG / SHEET 20 · WHAT A BUYER CHECKS · AS OF OCT 11, 2026
FOR AN OWNER · WHAT A BUYER CHECKS IN A $30M TO $150M PLANT
A buyer checks eight things. The permit and the customer list come first. The reactor list comes last.
#
WHAT A BUYER CHECKS
WHAT THEY ASK FOR
READY · PARTLY · NOT YET
ONE PROOF POINT FROM THE PUBLIC RECORD
1
Permits, and whether they move
Title V, RMP and PSM files; DEA registrations by drug code; any NRC license
A Title V permit can change owners by administrative amendment (40 CFR 70.7); a DEA registration can't move without DEA's written consent (21 CFR 1301.52).
2
Customer qualifications and concentration
Which customers audited the site, for which molecules, and each one's share of sales
Park Aerospace: one customer family is 39% of sales and a second is 12%.
3
Quality and inspection history
FDA inspections and outcomes, DMFs, ISO certificates, customer audit findings
About a third of US custom sites run to GMP (SOCMA).
4
EHS liabilities, and how they're priced
Phase I and II reports, consent orders, incident history, who keeps pre-close liabilities
Huntsman's ~$2.0bn sale to Indorama included ~$76m of liabilities.
5
Utilization
Reactor hours booked by train, the campaign calendar, idle cost
EUROAPI booked €11.1m of idle costs in H1 2026.
6
Asset age and capex backlog
Vessel inspection records, deferred maintenance, the capex a buyer inherits
VanDeMark's WARN notice cited unfunded retooling.
7
People and key operators
Who runs each hazardous step, their tenure, and the retention plan
Radford's nitroglycerin rebuild drew one bid.
8
Contract terms
Term, volume commitments, price adjustment, change of control
CordenPharma had ~€3bn of contracts before a €900m build.
HOW I'D USE IT
Tick one box per row, from the files you'd hand a buyer today. Start with rows 1 and 2, because that's where a buyer starts. There are no weights and no total score, because nobody prints what each row is worth at this size.
THE ONE PRICE ANCHOR AT THIS SIZE
Jayhawk, a US multipurpose site in Galena, Kansas, sold at ~$134m EV on ~$78m of 2024 revenue and ~$15m of EBITDA. That's about 1.7x revenue (my math on Anupam's figures) and ~9x EBITDA (Anupam's figure). It's one deal, so I wouldn't build a range on it.
Sources: 40 CFR 70.7(d)(1)(iv) and 21 CFR 1301.52(b) (eCFR, Sep 2026); Park Aerospace 10-K, FY ended Mar 1, 2026; SOCMA, Oct 23, 2025; Huntsman 8-K, Jan 5, 2020; EUROAPI H1 2026 press release, Jul 29, 2026; C&EN, Mar 13, 2026; DoD contract announcement, Oct 2023; CordenPharma, Jul 16, 2024; Anupam (NSE), Dec 9, 2025. The boxes are a self-check, not a score.
August Grace Advisory · curiouslyoptimistic.com20
AGA·DWG / SHEET 111 · THE RESEARCH AGENDA · AS OF OCT 19, 2026
THE RESEARCH AGENDA · WHAT I'VE PUBLISHED, WHAT'S NEXT, WHAT I'M THINKING ABOUT
This is my second Field Guide. Here's what's out, what's on the bench, and what I'm still chewing on. Tell me what to build first.
It's a plan, not a promise, and the order is yours to influence. Reply with the one you want most, or a topic that isn't here. Paid subscribers and founding members get the deciding vote.
STATUS
PIECE
WHAT IT'S ABOUT
WHEN
PUBLISHED
Electronic Chemicals
Field Guide 1. The ~$73B of materials that make every chip, and where a financial buyer can build. This guide links to it wherever a skill sells into a fab.
Aug 13, 2026
PUBLISHED
M&A Still Finds a Way
Essay. Capital didn't pause this year, it got pickier: spec-driven, domestic, hard-to-copy assets still trade.
Sep 11, 2026
PUBLISHED
What It Takes to Make a Raptor Move Like That
Essay, F-22 part 1. A thousand little shops, and the coatings that keep the F-22 flying.
Oct 7, 2026
PUBLISHED
We Reshored the Buildings
Essay. Ten years of US factory construction, and about a third less chemistry.
Oct 14, 2026
PUBLISHED
Custom Chemistry Makers
Field Guide 2, this guide. What a buyer pays for in a US custom chemistry plant.
Oct 24, 2026
NEXT
F-22, part 2
Essay. The planes before the Raptor, the Raptor itself, and how it got made.
Tentative
NEXT
F-22, part 3
Essay. Where the materials are likely going next, labeled as speculation.
Tentative
NEXT
The F-22 materials map
Deep dive. Mills to formulators to fabricators to primes, with public supplier counts.
Tentative
NEXT
Where's the Chemistry in $100 of AI Capex?
Essay. A widely shared chart of $100 of AI spending has no line for chemistry. The ~$73B of chip materials sits one step back.
Tentative
PARKED
Rates Don't Print Oil
Essay. Rates can cool demand and buy a breather on prices. They don't put missing barrels back.
No date
THINKING ABOUT
The Hormuz quarter
Idea. What multiple do you pay on earnings a shipping chokepoint created? From public company results only.
Not scheduled
Correct me: a filing, a permit, or a transcript line that moves a number changes the guide. Corrections are credited.
Source: my Content Calendar, Oct 19, 2026. Published dates are the Substack dates. Anything marked next, parked or thinking about is tentative, and the order follows reader demand.
August Grace Advisory · curiouslyoptimistic.com111
AGA·DWG / SHEET 113 · SUBSCRIBE
HELP FUND THE RESEARCH · SUBSCRIBE TO CURIOUSLY OPTIMISTIC
This guide took real work, and readers pay for it. Subscribe, and the full guide and the next one are yours.
Curiously Optimistic is reader-funded. No ads, no sponsor, and no paywall on the thinking that matters. Paid subscribers get the full Field Guides, the subscriber-only posts and archive, a say in what I research next, and readers who argue back.
WHO READS THIS · FINANCIAL BUYERS · CORPORATE STRATEGY AND M&A · OPERATORS AND BOARDS
Every number in this guide comes from a filing, a permit, a register or a release, dated on its sheet. A subscription pays for that reading, and keeps the next guide coming.
Curiously Optimistic, by Bob Girton · curiouslyoptimistic.com · My open research notebook on chemicals, materials, growth, teams, and AI in industrials.
August Grace Advisory · curiouslyoptimistic.com113
AGA·DWG / SHEET 114 · CLOSE
CLOSE
Thanks for reading. If the public numbers are missing something, tell me.
If you're inside one of these nine skills and the public record is missing something, reply and tell me which sheet. If this was useful, subscribe. I always welcome a dialogue, so please challenge me on my perspective.
REPLY
Correct a sheet
Every sheet has a number. Tell me which one, and what the public record says.
SUBSCRIBE
Follow the notebook
I write up what I'm working through in Curiously Optimistic. Paid subscribers get every full Field Guide; the options are on sheet 113.
TALK
A live conversation
If you're working a question in one of these segments, I'm glad to talk it through.
Go deeper: want this worked on a question you name? Private work is scoped and priced to the question. Bob Girton, bg@augustgraceadvisory.com.
Curiously Optimistic, by Bob Girton · August Grace Advisory. Field Guide, October 2026.
August Grace Advisory · curiouslyoptimistic.com114
The full guide · 114 sheets, nine skills deep
Ten sheets in, you've got my ranking and my reasoning. The other 104 are where I show my work, and they come with a paid subscription.
Curiously Optimistic is my open research notebook on chemicals, materials, growth, teams, and AI in industrials. Subscribe to read the full Custom Chemistry Makers guide, or if you're working on something more specific, I'd be glad to talk live.