Toast: Sticky SaaS - [Business Breakdowns, EP.247]

Summary

• Sean Barrett of Counter Global lays out a 15% position in Toast at what he calls “a generational opportunity”: an 18x 2027 GAAP PE for a “durable twenty-plus percent revenue compounder and thirty-plus percent EPS compounder.” The business does ~$2B of recurring gross profit at ~35% EBITDA margins on a ~$12B enterprise value, with almost $2B of cash on the balance sheet; his DCF puts fair value “something like fifty dollars per share or more” versus a recent $22-23 stock price.
• His answer to the “SaaSpocalypse” is history: today rhymes with 2014-15, when AWS-distributed open source crushed software multiples to 3-4x revenue before the category killers “kept putting up numbers… and the stocks went parabolic” by early 2016. Vertical-market multi-tenant SaaS category killers are, in his view, in a better spot because of AI, not a worse one.
• Toast has gone from one TAM to five — core SMB restaurants plus enterprise (Applebee’s), grocery/liquor/gas, hotels (Marriott), and international (UK, Ireland, Canada, Australia) — while margins swung from -16% in 2022 to ~35%, headed to a “line in the sand” of 40%+. It’s been a “ninety-seventh percentile grower” in public markets for three years, holds ~20% US restaurant share at 160k+ locations, and is winning roughly half of all new US restaurant openings.
• AI is “the best thing to happen to Toast since their founding”: Toast IQ (conversational AI plus a system of action, including agentic inventory management, used weekly by ~50% of customers) and the new Toast Grow marketing engine, which for $500/month has driven ~8% total revenue uplift — “about a 20X ROI right out the gate” on a $1.3M average customer. Internally, R&D dollars have been roughly flat for two years while the business doubled.
• The DoorDash threat gets the boots-on-ground treatment: a colleague walked into 30-40 San Francisco restaurants running both, and “we couldn’t find a single customer that said they would switch if DoorDash offered free point of sale.” The economics are “non-rational” — saving ~50bps on payments to pay a 13-15% delivery take rate — and Toast’s free-delivery module, in partnership with Uber Eats, has an estimated ~80k restaurants signed up, flipping the innovator’s dilemma onto DoorDash.
• The industry’s 15% annual churn is reframed as the challenger’s edge — ~100k-120k restaurants shut down and generally restart or open each year, giving Toast at-bats it would never get in a 99%-retention industry — while a typical Toast customer runs ~15% margins versus a typical restaurant’s 10%, so Toast customers survive more and churn less. The runway: $2B gross profit to $10B by 2035 against a 15M-location global TAM ex-China, implying a 4x MOIC at “a horrible multiple” and ~10x at a NASDAQ multiple.

Deep dive

1. A fifteen-percent position priced like a value stock

• Barrett’s origin story with the name: he first invested during COVID in 2020, opened the model management sent him and said “This can’t be right. These numbers are too good. Retention’s too high for the restaurant industry” — and the reply was “Nope, you’re not missing anything. These are the numbers.” It’s now a 15% position at Counter Global.
• Today’s snapshot, once you parse the interchange accounting noise in the 10-K: ~$12B enterprise value, ~$2B recurring gross profit, ~35% EBITDA margins, minimal capex, 25%+ gross profit growth — versus the 2020-21 hypergrowth phase of substantially negative EBITDA and stock comp at 30%+ of revenue. His headline: “eighteen times next year’s GAAP PE for a durable twenty-plus percent revenue compounder and thirty-plus percent EPS compounder.”
• Matt Russell’s early probe — is that gross profit really recurring? Barrett’s honest distinction: it’s “recurring in nature.” Two-thirds is payments gross profit at a net take rate, one-third software; payments are more volatile than a contract-based business, but customers run seven modules on average, “if the customers turn off Toast, they can’t run their business,” and payments revenue grows with inflation and GDP.

2. The SaaSpocalypse rhymes with 2015

• Barrett doesn’t dismiss AI disruption — “we aren’t blind to this,” and coding/DevOps software is genuinely exposed — but for vertical-market category killers he reaches for the 2014-15 analog: AWS put open source on its platform, software multiples “collapsed to three or four times revenue, basically where they are today,” and everyone asked why anyone would pay for software.
• The resolution took ~18 months: by early 2016 “the category killers kept putting up numbers. They kept innovating… and the stocks went parabolic.” His conclusion: “we have a generational opportunity to invest in something like Toast… at a GAAP net income multiple.”
• The multi-tenant architecture is the reason AI widens rather than closes the gap: Toast can “code once and then deploy all the updates with new AI models… rapidly to the whole customer base,” while an on-premise competitor may need to send a technician to update a server every couple of weeks as OpenAI or Claude ships new models — “good luck. It’s not gonna happen.”

3. How the model works: 49bps, $10k a year, and hardware nobody can shortcut

• The unit math on a $100 dinner: ~3% interchange, ~$2.50 to banks and networks, ~49bps net to Toast — under-monetized versus competitors at 75-125bps (Square monetizes at ~1% gross profit take rate with no comparable operating system on top). With ~$200B of volume, Toast is “in aggregate, one of the biggest merchants in the US,” giving cost-optimization room to lift the net rate. Software runs $300-500/month; all-in, the average customer pays ~$10k a year on $1.3M of revenue. The whole business is consumption-based; hardware is a loss leader, while lending to restaurants is a small, profitable business.
• Every customer must run Toast hardware, and the shortcut-takers proved the moat: competitors built iPad apps, but “turns out iPads don’t work very well in restaurants” — water, heat, breakage — and “what was a shortcut at the beginning for some of those competitors actually led to their demise.”
• The proof of customer love: NPS of roughly 50 (Barrett notes 25 is “really good”), 95% would recommend, ~20% US share, and — the number that “really gets us excited” — roughly half of all new US restaurant openings, in an industry where the top one or two players historically end up with 40-50% share.

4. AI in production, not experimentation

• Toast IQ is conversational AI plus “a system of action”: multi-location menu and price changes that “used to take an owner all weekend” now happen in real time, custom analytics on demand, and agentic inventory management that automatically orders fresh produce from Instacart Business or other local vendors when needed. Roughly 50% of customers use it weekly.
• Toast Grow is the sharper monetization story: an automated marketing engine that spots a quiet “winter Tuesday in Boston,” fires SMS promos, website deals, and Instagram pushes automatically for $500/month — versus $1-2k/month restaurants pay agencies with little uplift. Early users see ~8% total revenue uplift: “about a 20X ROI right out the gate,” and potentially a 100% uplift to SaaS ARPU.
• The internal leverage is equally striking: “they’ve shipped more product in the last two years than I’ve ever seen them ship” while R&D dollars barely grew as the business doubled — a new operating-leverage lever alongside sales-and-marketing and G&A scale.

5. Competition: legacy is the donor, DoorDash is the debate

• Barrett splits the field: ~half the market is still legacy (NCR Aloha, Oracle MICROS — “you don’t want a hot server sitting in the back of your kitchen”); Square holds 5-10% at the low end and wins mostly from legacy and Clover, not Toast; Clover (~15%, Fiserv) “is not innovating, and customers have noticed” amid Fiserv’s leverage issues. Why Toast wins underneath it all: a typical Toast customer, from Barrett’s work, runs ~15% margins versus the typical 10% — “taking home 50% more dollars for the restaurant owner.”
• On DoorDash’s POS pilots — possibly a response to Toast’s free-delivery module, in partnership with Uber Eats, which Barrett thinks ~80k restaurants have signed up for — Barrett’s checks were categorical: switching means saving ~50bps on payments to pay a 13-15% delivery take rate, while as much as 30% of the order can go to fees from the customer-and-restaurant perspective. Among 30-40 dual-enabled San Francisco restaurants his colleague visited, “we couldn’t find a single customer that said they would switch.” His hedge: both can coexist, with DoorDash better suited to delivery-first quick-service restaurants.
• The churn question flipped: 15% annual industry churn (100-120k of ~800k US restaurants closing and generally restarting yearly) is “a huge positive for the challenger” — those are at-bats a 99%-retention industry never offers. The standing bear debate since 2019-20 — won’t Toast’s churn converge to the market’s as it scales? — hasn’t shown up in the data, which he attributes to survivorship bias among healthier Toast customers.
• On why an AI-native startup can’t just clone it: one high-profile entrant may have thought AI coding could copy Toast quickly; “fast-forward to today, we aren’t seeing them at all in the market,” and the company said it was at least another two years before having something that looks like Toast. It would then still need to build the supply chain, hardware, and “feet on the street in every major city in America.”

6. The 5x gross profit path, the team, and what could break it

• The TAM build runs on flywheel markets: city-level share goes 2%, 4%, 6% — and at 10% “the market decides it will standardize on Toast,” with 25-30%-share cities now adding share faster than smaller ones. Layer on enterprise (Applebee’s), hospitality (Marriott), grocery/liquor/gas, and international markets management says are growing faster than early-days US, against a 15M-location global TAM ex-China: $2B gross profit to $10B by 2035, which “we actually just think that’s a stopping point.”
• The moat package is broader than software: Toast is a mission-critical operating system with network effects, industry standardization, data advantages, and domain expertise. Barrett also emphasizes its purpose-built hardware and supply chain, plus feet-on-the-street distribution in major cities — advantages he says are difficult to replicate.
• On the team of three co-founders who met while getting computer science degrees at MIT (Aman Narang and Elena Gomez among the leaders), Barrett’s Munger-inflected test — integrity, ambition, innovation — recalls 2020, when they had 20-30k restaurants and “were talking about their path to two hundred thousand restaurants… and everyone thought they were crazy.” They’ll likely hit 200k this year or early next. “This is the team you want in an AI world.”
• Valuation three ways: 18x 2027 GAAP earnings (lower still adjusting for ~$2B cash); a 10-year MOIC of ~4x at “a horrible multiple” and ~10x at a NASDAQ multiple, with over $3B of GAAP net income by 2035 and the whole market cap generated in cash; DCF fair value “something like fifty dollars per share or more” versus $22-23.
• Risks, as hedged: macro (though restaurants were down only low single digits in 2008-09 — “people need to eat”); churn convergence he’ll “stay on top of”; ARPU price sensitivity, less concerning because growth is coming from products designed to deliver ROI for customers rather than simply from pricing; and competition, “the risk that we stay on top of the closest.” His closing lesson from the company: resilience — and “stay hungry.”