The Cava Put: What Fast-Casual Investors Are Really Paying For

It’s a little after 6:30 on Friday morning and I’m in the office before the bagels arrive, which is how I like it. My second coffee is going cold next to a legal pad where I’ve been sketching a pairs-trade model since the Cava print hit the tape last night. The screen on my right has four browser tabs open — Cava, Sweetgreen, Wingstop, and a half-finished comps sheet — and the screen on my left has a single sticky note with the only number that matters this morning written in red Sharpie: 10.8.

That’s Cava’s first-quarter same-store sales growth. Traffic up 7.5%. AUV at $2.9 million. The release is right there on the SEC site if you want to read it cold (earningsrelease2025q1.htm), and Restaurant Dive has the operator-flavored write-up (Cava Q1 2025: traffic growth continues). The number is real. The traffic is real. The AUV is real.

And the multiple is real, too. That’s the part the buy-side group chats spent last night arguing about.

Here is the Bottom Line take, and I’m going to put it at the top so nobody has to scroll: Cava is fairly priced as an outright long. The cleaner risk-reward right now is the relative-value trade — long Wingstop, short Sweetgreen, with Cava sitting in the middle as the benchmark you’re paying market price to be exposed to. That’s the trade I’d put on this morning if I were running money instead of writing about the people who do.

Let me show you the work.

The Cava print is the kind of print that makes analysts lazy

A 10.8% comp with 7.5% traffic is not a number you fade. It’s the rarest combination in casual and fast-casual right now: a brand that’s growing the check by less than it’s growing the line. Most “great” restaurant prints over the last eighteen months have been ticket-led — menu-price flow-through doing the heavy lifting while traffic flatlines or bleeds. Cava flipped that. The customer is showing up more often, and Cava isn’t leaning on price to get the comp.

AUV at $2.9 million matters for a different reason. It’s the unit-economics anchor that supports the whole growth story. When new units open into a $2.9M AUV trajectory, the four-wall math gets very forgiving — payback compresses, the IRR on a new build improves, and the company can keep guiding to an aggressive opening cadence without the development pipeline turning into a margin tax. That’s why the stock can carry the multiple it carries. It’s not a vibe. It’s the model.

Here’s the problem. The model is now the consensus model. Every sell-side note I read overnight had some version of the same line: “premium multiple is justified by category-leading comps and AUV.” When the sell-side and the buy-side and the talking-head channel all agree on the same justification at the same multiple, you are no longer being paid to hold the asymmetry. You’re being paid the market rate to hold the asymmetry someone else identified eighteen months ago.

That’s fine. That’s a perfectly respectable place for a stock to be. But it’s not where I want my next dollar of fast-casual exposure.

Wingstop is the trade nobody wants to make twice

I keep coming back to Wingstop, and I keep coming back to it because the print is, frankly, ridiculous. The Q1 release is on EDGAR (a991wingq12025earningsrele.htm) and the two numbers I have circled on my pad are these:

• Digital sales: 72.0% of system-wide sales.
• 126 net new units, delivering “18% unit growth” — that’s CEO Michael Skipworth, verbatim, in the Q1 8-K.

Take a breath and think about what 72.0% digital actually means. It means three out of every four dollars that ran through a Wingstop in the quarter touched a piece of first-party software before it touched a piece of chicken. That’s not a restaurant chain anymore. That’s a software-distributed kitchen network with a wing brand bolted on. The data flywheel — customer files, frequency, recapture, daypart shaping — is the moat, and it gets cheaper to operate every quarter because the marginal digital order costs roughly nothing to take.

Now layer in “18% unit growth.” Skipworth used that exact phrase, and he used it for a reason — it’s a number that, if it holds, compounds into something the comps line alone can’t tell you. A franchise system growing units at 18% with 72.0% digital penetration and operator-level economics that franchisees keep voting for with their own balance sheets is, structurally, a different business than a company-owned fast-casual rolling out 60-70 boxes a year. The unit growth rate isn’t just an operating metric here. It’s the asset-light compounding engine.

My base case: Wingstop is the only name on my fast-casual screen where I think the next two years of multiple expansion is still coming from a place the consensus model doesn’t fully price. Not because the consensus is dumb — it isn’t — but because the digital mix is still treated as a margin story when it should be treated as a category-redefinition story. Three-quarters of the system running on first-party software changes what kind of business you’re underwriting.

That’s the long leg.

Sweetgreen is the short, and I don’t say that with any joy

The Sweetgreen release is also on EDGAR (q125sweetgreenearningsrele.htm) and the line item I want to draw your attention to is the one that ate the tape: SSS -3.1%.

Negative three-point-one. In a quarter where Cava printed +10.8 on a comparable customer base in overlapping urban markets. That gap — almost fourteen points of comp differential between two brands selling bowls to broadly the same lunchtime consumer — is the most informative single data point in the entire sector right now.

I want to be careful here. Sweetgreen is not a failed company. The Infinite Kitchen rollout is a real piece of operational engineering, the brand still resonates in core markets, and the management team has earned a hearing. But when you’re trading at a fast-casual growth multiple and your same-store sales line is going the wrong way while your closest peer is putting up double-digit prints, the multiple has nowhere to hide. Either the comps reaccelerate quickly or the multiple compresses to meet the comp. There isn’t a third door.

My base case: the multiple compresses first. It almost always does. And that makes Sweetgreen the natural short leg against either a long-Cava or long-Wingstop book.

So why pair them instead of just owning Cava outright?

This is the part of the column where I have to actually defend the contrarian piece of the thesis, because I can already hear the email I’m going to get from a portfolio manager I respect: Marcus, if Cava is fairly priced and Wingstop is the structural compounder, why pair anything? Just own Wingstop and go to lunch.

Fair. Here’s the answer.

When you own Wingstop outright, you’re long the structural fast-casual growth trade and you’re also long the broader restaurant beta — consumer discretionary spend, labor inflation, commodity inputs, the whole macro stack. The single-name long doesn’t isolate what you actually believe. What I actually believe is something more specific: digital-led, asset-light, unit-growth-compounding fast-casual is going to outperform traffic-stalled, capex-heavy, brand-led fast-casual over the next several quarters. That’s a relative-value belief, not a directional one.

To express it cleanly, you pair. Long Wingstop, short Sweetgreen, and let Cava sit in the middle as your benchmark — the name that prices in what consensus thinks a “winning” fast-casual model looks like. If consensus is right about Cava, the pair still works because Wingstop’s structural drivers are different from Cava’s and the digital-mix re-rating happens on its own clock. If consensus is wrong about Cava — too generous on the multiple, or too forgiving on the unit-economics ramp at scale — the pair works even better, because Sweetgreen is the higher-beta version of whatever de-rating eventually hits the category, and the short leg outruns the long-leg drawdown.

You are, in plain English, getting paid to be right about the shape of the category without having to be right about the direction of the category. That’s the trade.

The other reason I prefer the pair: the calendar. The NRA Show runs May 17 through May 20 here in Chicago, and the show floor is going to be wall-to-wall AI-kitchen, voice-ordering, and back-of-house automation pitches. Every operator I’ve talked to this week is heading in with a shopping list. That cycle of “operators see the demo, operators imagine the comp lift, operators write a check, sell-side writes the upgrade note” tends to compress into the two weeks after the show, and it tends to be kindest to the names that already have a digital story the analyst community is willing to extrapolate. Wingstop at 72.0% digital is exactly that name. Sweetgreen, with negative comps and Infinite Kitchen still scaling, is the name that has to defend its capex story under the same hot lights. The pair has a near-term catalyst on the long leg and a near-term vulnerability on the short leg, on the same calendar.

As our later coverage of the structural divergence argues (/blog/posts/what-the-doordash-sevenrooms-deal-actually-buys), the gap between digital-mature operators and the rest of fast-casual isn’t a quarter-to-quarter mood swing — it’s a regime, and the regime is widening. This pair is just one way to express it.

What would make me wrong

I owe you the other side of this. The honest version of “my base case” is that it comes with a list of things that would make me reverse the trade, and I keep that list on the same legal pad as the thesis.

The thing that breaks the long leg: any sign that the digital mix at Wingstop is plateauing, or that the franchisee-level economics are softening enough that the 126-units-in-a-quarter pace doesn’t extend. Skipworth’s “18% unit growth” line is the load-bearing wall here. If unit growth decelerates meaningfully and the franchisee P&L story narrows, the structural-compounding thesis weakens and the multiple has to come in.

The thing that breaks the short leg: a real Sweetgreen reaccel — and by “real” I mean a positive comp print, not a “less-negative” comp print — combined with an Infinite Kitchen rollout cadence that visibly drops labor as a percent of sales at the store level. If both of those land in the same quarter, the short is over and I cover.

The thing that breaks the middle: Cava posting a quarter where traffic decelerates without a corresponding step-up in check. The 7.5% traffic number this quarter is the part of the print that’s hardest to fake, and it’s the part that justifies the multiple. If that line softens, Cava stops being “fairly priced” and starts being expensive, and the read-across hits everything in the category — including, uncomfortably, my long leg.

None of those things happened this quarter. All of them could happen next quarter. That’s why this is a trade with sizing, not a religion.

The piece I’m not writing today

There’s a parallel column forming in my head about the AI premium that’s about to get attached to whichever fast-casual name shows the cleanest automation story coming out of the McCormick Place show floor next week. I’ll write it after I’ve walked the floor. In our subsequent piece on the AI premium (/blog/posts/dont-pay-the-ai-premium-a-buy-side-thesis-on-restaurant-m-a-in-2026), I’ll get into which of these three names is most likely to be the beneficiary, and which is most likely to be the one that has to explain itself on the next call. For now, the pair-trade thesis stands on the numbers we already have on the tape.

My base case in three lines

• Long Wingstop. 72.0% digital, “18% unit growth,” asset-light compounding the consensus still treats as a margin story.
• Short Sweetgreen. Negative 3.1% comp into a category putting up double digits. Multiple has nowhere to hide.
• Cava is the benchmark, not the position. Fairly priced. Beautiful print. Not where the next dollar goes.

Pour another coffee. The bagels just arrived.

— Marcus writes The Bottom Line. Tips: [email protected].