Pershing Square Challenge 2026 winners on DoorDash $DASH

Summary

• The winning team’s core variant view is that US restaurant delivery is “still in the middle innings” despite years of Street skepticism that the category is penetrated. They model US-restaurant MAUs compounding 8%, with roughly 40–45% of growth attributed to demographics—the transcript cites ~40% of North American growth in one passage and ~45% of MAU growth in another—plus share gains and low-frequency users becoming more active. Frequency rises from ~5 to 6 orders/month (3%/yr), benchmarked against Meituan’s 8–9x monthly in China.
• Valuation: a ~$320 price target in 2029 based on 2030 forecasts, using a 10x terminal EBITDA multiple that implies ~18x free cash flow after deducting stock-based comp, with EBITDA compounding “over 30%… for the next four years.” They’re 16–17% above Street 2028–2030 EBITDA, arguing the Street “terminalized” the November tech-investment guidance that knocked 2027–28 estimates down 11–12% and never rebased.
• Andrew’s sharpest pushback: on GAAP the $70B company is “at best break even and at worst slightly negative” — ~$3B trailing EBITDA is heavily adjusted, with $1B of 2025 SBC (which he said could rise to $1.3B in 2026), ~$1B LTM capex and $1.3B capitalized software. The team’s answer is that terminal-year framing plus SBC-adjusted cash flow, not near-term multiples, is where they “got comfortable.”
• The research was the differentiator: about 90 primary calls (former DoorDash employees, competitors, merchants, investors, sell-side) plus a proprietary Wolt case study using Denmark’s mandatory subsidiary disclosures to show post-2022-acquisition share gains — the piece multiple people flagged to Andrew as genuinely proprietary, and the basis for trusting the Deliveroo integration playbook.
• On new verticals, the team concedes Andrew’s point that grocery is a structurally worse business — commission rates are less than half of restaurants’ 20%+ take — but tracked basket sizes growing from the $30s to north of $50, and models the segment (currently losing ~$1.5B) inflecting to profitability only in the terminal year. “It’s not this crazy thing that it’s gushing cash in 2030.”
• This is explicitly a Tony Xu bet as much as a DoorDash bet: a “culture of operational excellence” visible in calls where DoorDash employees could explain how their work tied to KPIs, while an Uber counterpart described her role and some broader linkage less fully, and alignment where Tony earns $400K all-in and executive comp excluding the CEO is 4% cash, 96% equity.
• Against the Citrini AI-agent bear case — which Andrew thinks is probably wrong but illustrated with a $30 DoorDash pizza that cost $24 direct — the team argues the moat is the three-sided network, not software: physical-world handoffs “can’t just be vibe-coded away,” and history backs them — Google shut its food-delivery effort in 2024 and ByteDance deprioritized its 2021–22 attempt in China.

Deep dive

1. They picked the stock for the game they were playing

• Aaron’s screening criteria, laid out candidly: a business “somebody would already know” so no pitch time is wasted on the model; genuine risk-reward asymmetry (“if everything goes right, how much money do you make? If everything goes wrong, how much can you lose?”); and key issues you can actually research in three months — surveys, channel checks, app scraping. Dozens of names got looked at in two to three weeks; some got “10 minutes before somebody vetoed it.”
• Andrew’s endorsement of the meta-game: candidates constantly pitch him “a complicated biotech long that requires a PhD” to “mouth jewelers like me” — whereas even if the case is 100-going-to-400 rather than 100-going-to-1,000, “this is just a better pitch. It’s a better stock story.”

2. The core edge: US restaurant growth is mid-innings, not tapped out

• The thesis rests on three legs — US restaurant durability, earlier-than-expected profitability in new verticals and international, and underappreciated opex leverage — but Aaron is clear the first is the biggest driver versus consensus “just because US restaurants is the biggest part of the business.”
• The MAU math: ~8% CAGR, with roughly 40–45% attributed to demographics—the transcript gives ~40% of North American growth in one passage and later ~45% of monthly-active growth. About 4M people turn 18 each year; they may be entering their first disposable-income or full-time-work period, and the team believes Gen Z grew up in an on-demand world. The rest comes from share gains versus Uber Eats and from people ordering “three or four or five times a year” — who only show up in monthly actives a third of the time — stepping up frequency and entering the count.
• Frequency goes from ~5 orders/month (their 2025 estimate; DASH doesn’t disclose US-only) to 6 — just 3%/yr — cross-checked against cohort disclosures, third-party data showing tenure drives frequency, and Meituan’s 8–9x monthly in China: “the fact that China’s at eight or nine today and we’re modeling the US getting to six over the next four years… seems reasonable.” AOV is modeled below forecast inflation given two years of stable baskets.
• Why the young skew to DoorDash over Uber Eats/Grubhub: partly speculation, Aaron admits, but “when we look at DoorDash from every angle that you slice the product, it’s better” — fees, speed, selection, satisfaction. ZK: “management has been maniacal about making this product amazing.” Andrew’s revealed-preference confession: a longtime Seamless user who drifted to DoorDash without noticing why — “the DoorDash product, it is just simpler.”

3. Valuation: the GAAP pushback and the $320 answer

• Andrew’s challenge: $70B market cap on ~$3B trailing EBITDA that’s “pretty heavily adjusted” — $1B of 2025 stock comp, which he said could rise to $1.3B in 2026, ~$1B LTM capex and $1.3B capitalized software — so a GAAP purist sees “at best break even,” and 25–30x adjusted EBITDA is already rich before any growth story.
• Elliot’s construction: assume multiple compression, put 10x on terminal EBITDA, and get a ~$320 target in 2029 based on 2030 forecasts. They then sanity-check on cash — “cash flow is king” — with SBC treated as “a real expense,” so free cash flow minus SBC at that valuation implies ~18x, which Andrew concedes is “below market multiples now.”
• Aaron’s framing of the underwriting: with EBITDA compounding over 30% for four years, “if we were to make this investment and then not look at it for three years, we think this is going to be a much bigger and more profitable business” — while acknowledging the modeled downside (no growth, no margin expansion) loses money.

4. The competition bear case Andrew wouldn’t let go

• Andrew’s scenario-stack: his own membership came bundled via Chase Sapphire/Lyft; he believed his Seamless/Grubhub membership came through Amazon; so what if delivery becomes a bundled product — Amazon buys Grubhub outright, uses its logistics network to deliver from other grocery stores, or Uber outbids for the Chase partnership and “all those Chase Sapphire members like me” default to Uber Eats?
• The team’s evidence: an expert call with “a relatively high up Amazon exec” suggested DoorDash is too small for Amazon to worry about — they’re focused on “the big sort of Walmart type competitor” — and senior Instacart contacts expect all the players in the space to do well over the next three years.
• ZK’s honest hedge: “we will definitely keep monitoring that competition landscape… based on the current evidence, we think the competition is not kind of the headline concern.”

5. Ninety calls, and what grocery economics actually look like

• The ~90 calls (Andrew: “you got to round it up to 100, man”) spanned former DoorDash employees, current/former staff at all of its competitors, merchants — “invaluable” on commissions and sales interactions — plus DASH-owning investors and covering sell-siders.
• The best specimen is grocery unit economics: the dasher walks the aisles and must be paid for time, while grocery commissions run less than half the 20%+ restaurant take — “you need a much higher basket size to make the math work.” Calls with current and former Instacart people plus third-party data showed DoorDash baskets going from the $30s three years ago to “a little bit north of 50,” which directly times when new verticals turn profitable in their model. Elliot says the calls informed forecasts for orders, AOV and MAUs.
• On Andrew’s Tegus-sourced pushback that every new vertical adds TAM but is “a worse business”: Elliot agrees — “I think that’s true” — but points to management discipline: KPI-gated capital budgets, leadership changes when targets are missed, and public exits from full international markets earlier this year, which the team reads as proof “they’re not growing hand over fist ignoring the path to profitability.” New verticals are losing ~$1.5B today and only inflect in the terminal year.

6. How much of this is a Tony bet? Basically all of it

• Elliot doesn’t dodge the Berkshire/Buffett analogy: “Tony created this… it’s a huge part of what we’re quote-unquote betting on,” dating to starting the company at Stanford GSB in 2013. Aaron’s most vivid evidence: back-to-back calls with a three-year DoorDash alum and a similar-level Uber employee — “everyone that we spoke with at DoorDash really gets what’s important,” while the Uber person could explain her own role and some connection to the broader picture. The team would be “definitely disappointed” if Tony left, but thinks the data-obsessed culture is “maybe ingrained… at this point.”
• The alignment kicker, from Andrew’s proxy read and ZK: Tony, age 41, makes $400K all-in while the CFO makes $14M; excluding the CEO, average target executive compensation is 4% cash and 96% equity. Andrew speculated that if Tony left, a replacement CEO might cost $30–40M in stock compensation.

7. Wolt, Deliveroo, and the tech-stack rebuild the Street is mispricing

• ZK’s proprietary case study on the 2022 Wolt acquisition mixed third-party data, employee calls, and a clever wrinkle: Denmark requires international conglomerates to disclose Danish subsidiaries, giving the team subsidiary information with which to track Wolt’s performance. Every market researched showed significant share gains — the basis for the team’s confidence in applying the acquisition playbook to Deliveroo, where one Deliveroo employee described Uber Eats “subsidizing” exclusive restaurant deals that Deliveroo could not fight back against, a problem DoorDash’s capital may mitigate.
• The reinvestment cycle is the setup: DoorDash, Wolt and Deliveroo run on three tech stacks being consolidated into one, at an estimated $300–400M of R&D and other expense — probably three engineering teams (San Francisco, London, Finland) shipping every feature becomes one. Aaron: this is “a huge part of why the stock is down so much from the peak” since the November earnings disclosure, with lower incremental margins scaring investors.
• Elliot’s variance math — the crux of “why are we so ahead of the Street”: since November, Street 2027–28 EBITDA is down 11–12% “and has not rebased,” while the team sits 16–17% above for 2028–2030. “The Street’s terminalizing this as a big expense and dragging their numbers through on their spreadsheets.”

8. The AI-agent bear case, and why physical networks resist it

• Andrew says he probably thinks the Citrini agents-eat-everything thesis (which DoorDash management spent much of the Q1 call rebutting) is wrong, but illustrates the risk with his New Orleans pizza: $30 on DoorDash, $24 direct from the restaurant — an AI agent told to “find me the cheapest way to order that pizza… would have cut out DoorDash completely.”
• Elliot’s rebuttal: the barrier “isn’t actually the software, it’s the actual network effects” of a three-sided local network, and Citrini “mistakenly assumes that after driver costs the delivery fees effectively all flow to profit” — ignoring driver insurance, payment processing and infrastructure that no entrant can escape. On the physical layer, the handoff from merchant to courier to consumer involves so many messy things that it “can’t just be vibe-coded away.”
• ZK’s history lesson on top-of-funnel shifts: Google did food delivery from 2013, leaned in during COVID, and shut it in 2024 “citing that they found it was really hard to change the customer behavior”; ByteDance tried in China in 2021–22 and deprioritized it — talent didn’t want the operationally heavy work and there was “insufficient customer mind share.” His hedge stands: “we will just keep monitoring that progress going forward.”