Full Report

The numbers behind The Trade Desk, Inc.: as-reported financial statements and company metrics for FY2021–FY2025, traced to the source filings, opened with the share-price history those statements have to justify. Every linked figure opens the exact page of the filing it was printed on, with the statement row highlighted. Amounts in US$ thousands unless noted.

Reading notes: Cross-check: every FY2021-FY2025 core-statement cell was taken from the filed 10-Ks and agrees with the standardized SEC XBRL feed (revenue, operating income, net income) within rounding - no material discrepancies were found. Display scale is thousands of US dollars, exactly as printed on TTD's 10-K/10-Q statements ('In thousands'). FY2023-FY2024 income-statement and cash-flow figures are the comparative columns of the FY2025 Form 10-K; FY2021-FY2022 come from the FY2023 Form 10-K. Balance-sheet FY2021 is the comparative column of the FY2022 Form 10-K (a 10-K prints only two balance-sheet years).

Share Price — Full Available History — 10 Years

The stock closed at $19.53 on Jul 10, 2026 — up 6,388% over the window shown (+53.1% a year), trading between $0.23 and $139.51. At that close the stock trades at 22× FY2025 diluted EPS as reported below.

Source: market price feed, weekly closes, sampled from 2,463 source observations, Sep 2016–Jul 2026. Price return only, excludes dividends. Prices are split-adjusted (1:10 on Jun 17, 2021).

FY2025 at a Glance

Source: FY2025 consolidated statements [1] [2]. Click any linked figure to open the filing page with the row highlighted.

Revenue by Geography

Revenue by Geography	FY2021	FY2022	FY2023	FY2024	FY2025
United States	—	—	1,696,911	2,133,502	2,476,683
International	—	—	249,209	311,329	419,601
Total revenue	1,196,467	1,577,795	1,946,120	2,444,831	2,896,284
Total revenue growth, derived	—	+31.9%	+23.3%	+25.6%	+18.5%

Source: Note 12 - Segment and Geographic Information (revenue by principal geographic area; first disclosed on a revenue basis in the FY2025 Form 10-K, which recast FY2023-FY2024) [3] [2]. Click any linked figure to open the filing page with the row highlighted.

Income Statement

Source: Consolidated Statements of Operations (Form 10-K) [1] [2]. Click any linked figure to open the filing page with the row highlighted.

Columns marked E are consensus analyst estimates shown alongside reported results for direct comparison; they are not company guidance.

Estimate source: Yahoo Finance analyst consensus, as of 2026-07-11. Estimate figures link to the consensus source, not to filing pages.

Balance Sheet

Source: Consolidated Balance Sheets (Form 10-K) [4] [5] [6]. Click any linked figure to open the filing page with the row highlighted.

Cash Flow

Source: Consolidated Statements of Cash Flows (Form 10-K) [7] [8]. Click any linked figure to open the filing page with the row highlighted.

Long-Term Record

Fiscal year	Total revenue	Income from operations	Net income	Diluted EPS	Net cash provided by operating activities
FY2016	202,926	57,518	20,482	—	75,031
FY2017	308,217	69,356	50,798	—	31,224
FY2018	477,294	107,323	88,140	—	86,603
FY2019	661,058	112,196	108,318	0.23	60,205
FY2020	836,033	144,208	242,317	0.49	405,069
FY2021	1,196,467	124,817	137,762	0.28	378,513
FY2022	1,577,795	113,654	53,385	0.11	548,734
FY2023	1,946,120	200,480	178,940	0.36	598,322
FY2024	2,444,831	427,167	393,076	0.78	739,456
FY2025	2,896,284	589,321	443,304	0.90	992,721

Source: consolidated statements across filings; older years from the standardized feed [7] [1] [8] [2]. Click any linked figure to open the filing page with the row highlighted.

Operating KPIs

KPI	FY2021	FY2022	FY2023	FY2024	FY2025
Gross spend	—	7,741,000	9,611,000	12,040,872	13,394,683
Adjusted EBITDA	—	—	—	1,010,649	1,196,449

Source: company-reported operating metrics [9] [10] [11]. Click any linked figure to open the filing page with the row highlighted.

Analyst Consensus

Estimate source: Yahoo Finance analyst consensus, as of 2026-07-11. Estimate figures link to the consensus source, not to filing pages.

Traceability

249 of 313 figures on this page (80%) link to the filing page where they are printed — click a linked figure to open the source PDF at that page with the row highlighted. Unlinked figures come from standardized data feeds or pre-filing years.

• Cross-check: every FY2021-FY2025 core-statement cell was taken from the filed 10-Ks and agrees with the standardized SEC XBRL feed (revenue, operating income, net income) within rounding - no material discrepancies were found.

• Display scale is thousands of US dollars, exactly as printed on TTD's 10-K/10-Q statements ('In thousands').

• FY2023-FY2024 income-statement and cash-flow figures are the comparative columns of the FY2025 Form 10-K; FY2021-FY2022 come from the FY2023 Form 10-K.

• Balance-sheet FY2021 is the comparative column of the FY2022 Form 10-K (a 10-K prints only two balance-sheet years).

• Revenue by geography (United States / International) was first disclosed on a revenue basis in the FY2025 Form 10-K, which recast FY2023 and FY2024; earlier 10-Ks disclosed only geographic Gross Billings, so FY2021-FY2022 US/International revenue is left blank (Total revenue is still shown and cited).

• TTD operates as a single reportable segment (advertising technology platform); the revenue breakdown therefore uses the company's geographic disaggregation.

• KPI Gross spend: FY2024-FY2025 are as printed in thousands in the FY2025 10-K (p.77); FY2022-FY2023 were printed in millions in the FY2023/FY2024 10-Ks (anchors '7,741' and '9,611') and are shown here scaled to thousands.

• FY2016-FY2018 diluted EPS is omitted from the Long-Term Record: the standardized feed reports those years on a pre-2021-stock-split basis, inconsistent with the split-adjusted figures used for FY2019 onward.

• FY2016-FY2020 long-term figures other than those cited to the FY2021 10-K (p.89) are from the standardized data feed (SEC XBRL) and are shown without page links.

• Quarterly block covers the four most recent 10-Q periods (Q1 FY25-Q1 FY26). TTD files 10-Qs for Q1-Q3 only; Q4 FY25 is not separately filed as a statement (the Q4 earnings press-release exhibit tables were not machine-readable), so it is omitted rather than derived.

---

The Trade Desk: what it is, and why the price fell

The Trade Desk runs the largest independent demand-side platform in digital advertising — software that ad buyers use to place campaigns across the open internet, priced as a fee on the spend that flows through it. The business still works: revenue grew 18% to $2.9 billion in 2025 on $13.4 billion of ad spend, at a take rate that has held near 21% for years, and it converts more than a quarter of revenue to free cash. Yet the shares have fallen roughly 86% from their 2024 high. This chapter sets out the company for a reader meeting it for the first time, and fixes the question the rest of the report answers.

What the company sells

The Trade Desk sits on the buy side of advertising. Its clients are advertising agencies and brands, who sign ongoing master service agreements and then use a self-service, cloud-based platform to create, manage, and optimize campaigns across display, video, audio, native, and — increasingly — connected television. The company takes no position on the other side of the trade: it does not own media inventory. It charges a platform fee, generally a percentage of the client's total spend, plus fees for data and value-added services [1].

That "buy-side only" stance is the company's central identity claim. Founder and CEO Jeff Green frames The Trade Desk as the neutral buyer's agent for the open internet — every ad-supported destination outside the walled gardens of Amazon, Google, and the large social platforms. On the FY2025 calls he argued that the demand-side platforms run by Amazon and Google are structured mainly to sell their own inventory, leaving the independent decision of "what should a brand buy across the open internet" as The Trade Desk's distinct territory, anchored by tools like the UID2 identity framework and retail-data integrations [2]. Whether that neutrality is a durable moat or a shrinking niche is a question later chapters take up; for now it is simply how the company is built.

The market it addresses is large and shifting. Management cites digital advertising at over $700 billion of annual spend — the largest, fastest-growing part of a global advertising market that passed $1 trillion for the first time in 2024 — with the generational move from linear TV to CTV, the rise of AI-driven automation, and audience fragmentation all cited as tailwinds toward programmatic buying [3].

The scale of the business

Two numbers describe the size of The Trade Desk. Gross spend — the client ad dollars transacted on the platform — reached $13.4 billion in 2025. Revenue — the company's own top line, the fee it keeps — was $2.9 billion. The ratio of the two, the effective take rate, has stayed remarkably stable near 21%.

Sources: FY2025 revenue, net income, and free cash flow per reported financials; gross spend and take rate derived from the FY2025 10-K MD&A gross-spend measure [4].

The growth record is the reason the company earned a premium reputation. Revenue compounded at roughly 28% a year from 2020 to 2025, and — unusually for a company still called a growth stock — it has been GAAP-profitable throughout, with cash generation scaling alongside.

Source: FY2021–FY2025 Annual Reports (Form 10-K), Consolidated Statements of Operations, as reported [5].

Two features stand out for a first look. The business is overwhelmingly domestic — the United States supplied about 85% of 2025 revenue — so the international opportunity is still mostly ahead of it rather than behind. And the customer base is sticky: The Trade Desk has reported client retention above 95%, a level it sustained for 32 consecutive quarters through 2021 across roughly 980 active clients [6]. Growth has come less from winning logos than from existing clients routing more of their budgets through the platform each year [7].

The cash economics reinforce the picture. The Trade Desk carries no drawn debt and ended 2025 with about $1.3 billion of cash and short-term investments; operating cash flow was $993 million and free cash flow $796 million — a 27% FCF margin. It returned more than it earned to shareholders, buying back $1.4 billion of stock during the year.

The re-rating

For all of that, the stock has been one of the sharper de-ratings in large-cap software. The Trade Desk closed 2024 at roughly $118 and 2025 at $38; by mid-2026 it traded near $20 — down about 74% over twelve months and roughly 86% from its late-2024 all-time high of $141, near the bottom of a 52-week range of $17 to $91.

Source: daily price history (split-adjusted); last point is the 2026-07-10 close. Company filings, as reported.

The valuation the market now assigns is the mirror image of that history. At roughly $20 a share the equity is worth about $9.6 billion, or near $8.3 billion of enterprise value once net cash is removed.

Source: market capitalization derived from 493.6M shares outstanding at the 2026-07-10 close of $19.53; earnings and free cash flow per FY2025 reported financials.

A company growing revenue near 20% with a 27% free-cash margin and net cash now trades at about 22 times earnings and 12 times free cash flow — multiples ordinarily attached to businesses expected to grow far more slowly. At its 2021 peak the same company traded at roughly twenty times revenue. The market has repriced both the growth rate and the premium it will pay for it.

Two events sit under the fall. In February 2025 The Trade Desk reported the first quarter in which it fell short of its own guidance in eight years as a public company — even as full-year 2024 revenue grew 26% to $2.4 billion on a record $12 billion of platform spend [8]. Management attributed the stumble to execution rather than demand, and disclosed that it had run the largest reorganization in the company's history that December, alongside a shift toward direct brand relationships and the rollout of its Kokai AI platform and Ventura operating system [9]. Growth then decelerated further through 2025, from 26% to 18%, and the multiple compressed the rest of the way.

Who controls the company

One structural fact belongs in any first description. The Trade Desk has a dual-class share structure: Class B shares carry ten votes each, Class A one. As of the end of 2025, Class B holders — the executive officers, directors, and employees, with founder-CEO Jeff Green the dominant holder — controlled about 49.9% of the total voting power. The Class B shares convert automatically to Class A on December 22, 2035 unless converted sooner [10]. Public shareholders own most of the economics and little of the control; the founder's strategic choices are, for now, effectively unchallengeable. That cuts both ways, and a later chapter weighs it.

The question this report answers

Strip the story to its tension. Here is a business that still compounds revenue near 20% at a stable ~21% take rate, generates a 27% free-cash margin, holds net cash, and retains almost every client it wins — and whose shares have nonetheless lost roughly 86% of their value from the 2024 high. The market has already priced the pessimistic case. The business, on its reported numbers, has not confirmed it.

The question this report exists to answer: is The Trade Desk's 2024–25 deceleration — the drop from 26% to 18% growth and its first guidance miss in eight years — a self-inflicted, fixable stumble inside a still-expanding open-internet advertising market, or the first evidence of a structural ceiling as walled gardens and platform maturity cap the independent demand-side platform's runway? Everything that follows — the durability of the take rate, the CTV and identity bets, the competitive reality against Amazon and Google, the quality of the cash, the capital returned, and what the collapsed multiple now implies — is an attempt to answer it on the evidence rather than the price.

---

The Independence Moat

The Trade Desk's defensible advantage is structural, not technological: it buys advertising but owns no media, so — unlike Google or Amazon — it has no incentive to steer a client's budget toward its own inventory. That objectivity underpins the retention and pricing established in The Business. But the same conflicted giants are far larger and, in 2025, grew faster: Amazon's ad revenue alone was roughly five times TTD's total gross spend. The moat is real and narrow — durable across the fragmented open internet, exposed wherever a giant owns the audience.

What the moat is, in the company's own words

TTD's founding argument is that a platform serving both buyers and sellers cannot be trusted by either. "We believe that there are inherent conflicts of interest when market participants serve both buyers and sellers" [1], the company wrote at IPO-era scale, and it built the business to sit only on the buy side, "free from the conflicts of interest inherent in our competitors that also own and operate media" [2]. A decade later the framing is unchanged: the platform "delivers valuable insights and results to clients without the conflict of interest and lack of objectivity that come with also selling owned advertising inventory" [3].

A moat claim only counts if it shows up in numbers. Here it does, in two established metrics: customer retention above 95% and an effective take rate that held near 21% even as revenue growth halved (both detailed in The Business). Neither has cracked under the slowdown — the deceleration has been a volume story, not a pricing one. That is the observable evidence that clients value the neutrality enough to stay and keep paying for it. What that evidence does not settle is whether neutrality is a durable barrier or simply a good position in a market the giants can out-scale.

The scale problem

The uncomfortable fact for the independence thesis is size. Amazon does not report its DSP separately, but its total advertising revenue — the closest disclosed proxy for its ad-tech reach — reached $68.6 billion in 2025, up 22% year on year [4]. That is 24 times TTD's $2.9 billion of revenue and roughly five times the $13.4 billion of gross spend that flows across TTD's entire platform [5]. And in 2025 Amazon's ad business grew faster than TTD's — 22% against 18%.

Sources: Amazon FY2025 10-K, disaggregated revenue [6]; TTD FY2025 10-K, MD&A Executive Summary [7].

The comparison needs a caveat, and it works in TTD's favor. Most of Amazon's $68.6 billion is owned-and-operated inventory — sponsored product ads on its own retail pages — a different market from the third-party open internet TTD buys across. Google's advertising base, larger still, is likewise dominated by its own Search and YouTube. The direct overlap is narrower than the raw figures suggest: it sits at the frontier, in Amazon's DSP for third-party buying and in Prime Video's connected-TV inventory. But the risk framing in TTD's own filing is candid: larger competitors "may have significantly more financial, technical, marketing, and other resources," and walled-garden inventory owners "may exclusively sell their own inventory directly to advertisers, which prevents us from competing with them entirely for such inventory" [8]. Independence is a reason to choose TTD; it is not a barrier to a giant deciding to keep its best inventory for itself.

The regulatory cross-current

The largest competitor is also a supplier — and, increasingly, a defendant. "Google is one of our largest advertising inventory suppliers in addition to being one of our competitors" [9], the filings note. In April 2025 a U.S. federal court found Google had illegally monopolized two ad-tech markets — the ad server publishers use and the exchange that matches bids; a remedies phase concluded in late 2025 with the Department of Justice pressing for a forced sale of Google's ad exchange, and as of mid-2026 the court's remedy ruling was still pending. CEO Jeff Green frames it plainly as a tailwind: "Many have asked what the likely impact of the most recent DOJ victory over Google will be… I do expect most walled gardens to behave" differently [10]. A structurally weakened Google, and one whose stack is scrutinized for self-preferencing, is worth more to a neutral buy-side platform than a free one.

The identity question cuts the other way, and the timeline is instructive. Google spent three years threatening to remove third-party cookies from Chrome — the change that made TTD's Unified ID 2.0 look essential. It delayed the plan in July 2024 [11], then abandoned it: "on April 22, 2025, Google announced that it would maintain its current approach to offering users third-party cookie choice in Chrome" [12]. Keeping cookies removes an imminent disruption but also removes a forcing function for UID2 adoption. The net is mildly positive — the antitrust cloud constrains Google's ability to favor its own tools more than the cookie reprieve helps it — but it is a reminder that the largest platform sets the rules TTD operates under.

Connected TV is where the moat holds

The strongest ground for independence is the channel driving TTD's growth. Green's structural argument is that the walled-garden model fails in CTV because no single streamer has the share to dictate terms — "nobody has enough share to be draconian," as he put it, so premium content owners are choosing to join the open internet rather than wall themselves off. In one quarter he counted Netflix adding TTD as a main programmatic partner, FOX expanding UID2 and OpenPath across its brands, and Roku adopting UID2 [13]. The 2025 surge in ad supply reinforces the point: with impressions far outpacing demand, "the objectivity of our position (by not owning media) is more valuable in this strengthening buyer's market" [14].

The exception proves the boundary. The one CTV player with the scale to wall itself off is doing exactly that: Green describes Amazon's Prime Video ad rollout as "another rollout of a walled garden" [15]. The moat holds where audiences are fragmented and no owner is a must-buy — Green's own test, that the walled-garden strategy "only works when a publisher is massive, and a must-have on a media plan" [16]. Prime Video, Search, and YouTube pass that test; the long tail of streamers, publishers, and audio does not.

The read: a narrow moat, not a wide one

On the playbook's scale — wide, narrow, none, or not proven — the honest verdict is narrow. The independence advantage is company-specific and hard to copy (a conflicted incumbent cannot credibly promise neutrality), it shows up in retention and pricing, and it has survived a demand downturn and a technology reversal without cracking. What it does not do is cap the giants' reach into the open internet, and their scale advantage is widening, not narrowing.

That distinction is what the through-line turns on. A narrow moat is consistent with both readings of the deceleration: growth cooled to 12% in Q1 2026 — revenue of $688.9 million against $616.0 million a year earlier [17], which is what a still-expanding but maturing niche looks like, not a collapsing one. The read would tilt structural if Amazon's DSP began taking third-party open-internet share rather than monetizing its own pages, or if the deceleration continued into single digits with CTV no longer offsetting it. It would tilt cyclical-and-fixable if a Google remedy pried open inventory to independents and CTV re-accelerated the top line. The competitive map explains why the business is defensible; it does not, on its own, settle how fast it can still grow.

---

Stock Comp and Cash

The cash-generative profile that makes The Trade Desk look cheap rests partly on a non-cash cost. Stock-based compensation was $491 million in 2025 — 111% of net income, 17% of revenue, and half of operating cash flow [1]. Charge that comp as the real economic cost it is, and the "12x free cash flow" headline moves closer to 32x. And $2.3 billion of buybacks since 2023 has shrunk the share count by only about 3%, mostly recycling shares that equity comp put back into the float.

The size of the number

Stock-based compensation is not a rounding item at The Trade Desk. Reported SBC has sat near $490–500 million for four straight years — $491.6M in 2023, $494.7M in 2024, $490.6M in 2025 — even as revenue grew from $1.9B to $2.9B [2]. In every one of those years the company paid its people more in stock than it earned in GAAP profit.

Source: FY2025 Form 10-K, Note 10 Stock-Based Compensation and Consolidated Statements of Cash Flows [3] [4].

Source: SBC per FY2023 and FY2025 Form 10-K Statements of Cash Flows; net income as reported [5] [6].

The largest single ongoing bucket is technology and development — $163 million of the 2025 total — with sales-and-marketing next at $113 million; both are recurring costs of running and building the platform, not one-off grants [7]. Management itself describes SBC as a "recurring expense and a key part of our compensation strategy" that "will continue for the foreseeable future" [8].

What it does to the cash-flow multiple

The Trade Desk's free cash flow is real cash, but it is calculated after adding SBC back to profit. Operating cash flow was $993 million in 2025; nearly half of that add-back — $491 million — is stock compensation [9]. Reported free cash flow of $796 million (operating cash flow less $197 million of property and equipment) is genuine, but it flatters the underlying economics because it treats a $491 million compensation cost as free.

Treat SBC as the recurring expense it is — subtract it from free cash flow — and the picture changes materially.

Source: derived from FY2025 Form 10-K Statements of Cash Flows — free cash flow $796M less SBC $491M [10].

Source: derived from reported FY2025 financials against the ~$9.6B market capitalization discussed in The Business (close of $19.53 on 2026-07-10) [11].

At the roughly $9.6 billion market value established in The Business, reported free cash flow of $796 million is about 12x earnings-power and an 8% yield. Deduct the $491 million of stock comp and the same company trades near 32x free cash flow, a yield closer to 3%. Neither figure is the "true" number — the honest read sits between them, because some SBC is a cost the company would bear in cash under a different pay mix and some is genuine upside-sharing — but the reframing is the point: the multiple that makes the stock look inexpensive depends heavily on how a reader treats stock compensation.

The buyback treadmill

The Trade Desk has spent heavily on buybacks, which in principle should convert that cash generation into a shrinking share count. The record is more modest. Across 2023–2025 the company repurchased $2.26 billion of Class A stock — $646.6M, $234.8M, then $1,380.4M [12]. That bought back 38.8 million shares. But equity compensation issued 24.3 million new shares over the same three years, so the net reduction was just 14.5 million — the share base fell only from 490.5 million at the end of 2022 to 475.9 million at the end of 2025, roughly 3% [13].

Source: FY2025 Form 10-K Statements of Cash Flows (buyback dollars) and Statements of Stockholders' Equity (share activity) [14] [15].

About 63% of the shares repurchased went to offsetting comp issuance rather than shrinking the float. The 2025 buyback was the first large enough to bite — 26.2 million shares retired against 6.0 million issued — helped by a depressed price that let the company retire more shares per dollar [16]. Read fairly, the buyback is doing two jobs at once: neutralizing dilution and, only recently, reducing the count. A reader should not assume the reported per-share figures carry a steady tailwind from repurchases; for most of the period they carried a headwind from issuance instead.

Two things that soften the read

The bear case on SBC has real limits, and the filings show both.

First, the burden is falling relative to sales. SBC held near $490 million while revenue grew about 85% since 2022, so stock comp has dropped from 32% of revenue to 17% [17]. A company that grows into a fixed comp pool is diluting shareholders more slowly each year, not faster.

Source: derived from FY2023 and FY2025 Form 10-K — SBC divided by revenue [18] [19].

Second, a large slice of historical SBC is a one-off that is rolling off. The 2021 CEO Performance Option — a 17.8 million-share grant to Jeff Green — carried $262 million of expense in 2022, $198 million in 2023, $128 million in 2024, and $67 million in 2025, with only $5 million left to recognize by the first quarter of 2026 [20] [21]. Strip it out and the flat headline hides a rising trend underneath: core employee SBC grew from about $294 million in 2023 to $424 million in 2025, up 44%.

Source: FY2025 Form 10-K Note 10; CEO Performance Option expense per Note 10, core is the residual [22] [23].

The two effects cut in opposite directions. As the CEO option finishes, roughly $67 million of expense disappears, which should let reported SBC drift down or stay flat even as headcount grows — a genuine tailwind to margins and to per-share cash. Against that, the underlying employee grant pool is still compounding at a mid-teens rate, so the fixed-pool comfort only holds while revenue keeps outgrowing it. If growth continues to decelerate — it fell to 12% in the first quarter of 2026 — the "SBC shrinking as a share of sales" argument weakens exactly when it is most needed.

What the depressed price is hiding

One quieter point sits in the earnings-per-share math. Because the stock traded at $19.53 — far below the $47.09 weighted-average strike on outstanding options — 22.7 million equity awards were anti-dilutive in 2025 and excluded from diluted share count, against just 0.3 million a year earlier [24] [25]. The 17.8 million-share CEO option is likewise entirely out of the money, with zero intrinsic value at year-end [26]. The low share price is temporarily suppressing the reported diluted count; a recovery toward prior levels brings a meaningful block of those awards back into the denominator. Dilution deferred by a weak stock is not dilution avoided.

The judgment

The Trade Desk's cash generation is real, but its reported profitability and free cash flow lean on stock compensation to a degree that a professional investor should net out before calling the stock cheap. The strongest fact for the company is that the burden is shrinking against revenue and that a $67 million CEO-option drag is about to roll off; the strongest fact against it is that core employee SBC is still growing 44% over two years and that $2.3 billion of buybacks has bought only a 3% smaller share count. On balance, the "12x free cash flow" framing overstates the value — the SBC-adjusted figure near 32x is the more conservative anchor. What would change the read: sustained shrinkage in absolute SBC as the CEO option ends, and a buyback that visibly reduces the diluted count in 2026 rather than merely offsetting new grants.

---

The Growth Engine

The Trade Desk's revenue growth has slowed in each of the last four reported quarters, from 25% year-over-year in the first quarter of 2025 to 12% in the first quarter of 2026, and the second-quarter 2026 guide of at least $750 million implies a fifth consecutive step down, to roughly 8% [1]. The engines meant to reverse that (connected TV, retail media, international) are real, and international is genuinely re-accelerating. But the slowdown is broad and sits in the large US brand-advertiser core, which keeps the cyclical-versus-structural question open.

The deceleration, quarter by quarter

The clearest fact about the business right now is the shape of its growth. Through 2024, quarterly revenue rose 22%–28% year-over-year. Across 2025 that rate fell in every quarter, and the first quarter of 2026 came in at 12% on revenue of $689 million [2]. The second-quarter guide of at least $750 million implies a further step down to around 8% [3].

Source: derived from reported quarterly revenue, company filings; latest quarter and Q2 2026 guide per Q1 FY2026 earnings call [4]. 2Q26E is the low end of company guidance.

A one- or two-quarter dip reads as noise; four consecutive quarterly step-downs into a single-digit guide read as a trend. Management attributes it to the macro environment — most of the company's revenue comes from Fortune 500 brands, and it points to tariffs, geopolitical uncertainty, and softness in consumer-packaged-goods and automotive spend [5]. That is a coherent cyclical explanation. It is also the pattern a maturing large-cap would show, so the driver mix matters more than the headline.

Geography: the segment still accelerating

The company reports revenue by only two segments that matter for growth: the United States and international [6]. They moved in opposite directions in 2025. US revenue growth halved to 16%, while international growth accelerated to 35% — the reverse of the total.

Source: The Trade Desk FY2025 Form 10-K, Note 12 — Segment and Geographic Information [7].

International is the smaller base — $420 million of $2,896 million in 2025, about 14.5% of revenue [8]. But because it grew faster off that base, it supplied $108 million of the $451 million of total revenue added in 2025 — roughly a quarter of the increment from a seventh of the base. On the first-quarter 2026 call the split had moved further, to about 82% US and 18% international, helped by momentum in Europe and Asia-Pacific [9]. Management has said for years that markets such as the UK, Germany, France, Japan, India, and Australia are where it is investing to grow [10], and it puts roughly 60% of the total advertising opportunity outside the United States [11]. The runway is real; the constraint is that international is not yet large enough to move the total on its own.

Channel mix: CTV gains share as the total slows

The Trade Desk does not report revenue by channel, but it discloses approximate mix each quarter. Video — which includes connected TV — has steadily gained share, from the mid-40s percent of the business in early 2024 to the low-50s in the first quarter of 2026, while mobile has ceded ground and audio, though small at around 6%, grew faster than any other channel last quarter [12] [13] [14].

Source: midpoints of the approximate ranges management discloses each quarter, Q1 FY2024–Q1 FY2026 earnings calls [15] [16] [17]. Figures are approximate and do not sum to 100 (display, the remainder, runs low-double-digit).

This is the important nuance. Connected TV — the channel the company ties its future growth to, as linear television shifts to streaming and it expands CTV inventory and spend [18] — is still gaining mix share even as the total decelerates. Video/CTV winning a larger slice of a slowing whole is not the same as CTV re-accelerating the whole. Retail media is the other newer engine: the company's retail data marketplace now reaches over 80% of sales from top US retailers [19], and management counts it, alongside AI-driven search, as a reason its addressable market is expanding faster than it once expected [20]. Neither has yet been large enough to offset the gravity on the mature US display and mobile base.

The runway and the read

Management's cyclical case rests on the size of the opportunity. It frames advertising as a $1-trillion-plus market, of which the open internet — the slice The Trade Desk plays in, distinct from walled-garden social and search — is roughly $280 billion [21].

Source: The Trade Desk Q1 FY2026 Investor Presentation, market-size estimates [22].

Against a $280 billion open-internet opportunity, the company's $13.4 billion of gross spend represents under 5% penetration — genuine room to grow. That is the heart of the bull's structural argument, and it is not empty: an under-penetrated, expanding market with a smaller international engine compounding at 35% is what a fixable, cyclical slowdown can look like.

Asked directly to separate the cyclical from the structural, CEO Jeff Green was unambiguous: "The structural drivers of our business are extremely strong… reacceleration is not really about reinventing ourselves. It is about executing against the larger, expanding opportunity… In the near term this is about execution" [23]. That is management's answer to this report's central question: cyclical and self-inflicted, not a ceiling.

The evidence is genuinely two-sided, and it is worth saying which way each fact cuts. Supporting the cyclical read: international re-accelerating, CTV still gaining mix, a large under-penetrated TAM, and a slowdown that traces to specific brand-advertiser verticals — CPG and automotive — under macro and tariff pressure rather than to a channel visibly losing to a walled garden [24]. Cutting the other way: the deceleration is broad and persistent — four sequential quarterly step-downs into a single-digit fifth-quarter guide — and it is concentrated in the large US core, exactly where a maturing platform would slow, while the growth engines management points to are either too small (international, audio) or gaining share of a decelerating whole rather than lifting it (CTV). The read that fits both sets of facts: the engines are real but not yet decisive, so the near-term trajectory turns on execution against the narrow moat the report has already established. What would change it in either direction is concrete — whether international and CTV begin to re-accelerate the total, not merely their own share, once macro conditions stabilize, or whether the ~8% second-quarter guide marks a new, lower plateau. The next two quarters, not the next annual report, hold the evidence.

---

What the Price Implies

At $19.53 The Trade Desk is worth about $9.6 billion, or $8.3 billion once its $1.3 billion of net cash is stripped out. That price is roughly 12 times reported free cash flow — or roughly 32 times the same cash flow with stock compensation charged as the cost it is. Those two multiples describe two different businesses. A reverse-DCF shows the price embeds either flat-to-declining cash flow forever, or steady ~10% compounding — depending entirely on which one you use.

Two multiples, one price

The number a reader attaches to this stock depends less on the price than on which earnings stream is put beneath it. On reported free cash flow of $796 million — operating cash flow of $993 million less $197 million of capital spending [1] — the company trades at about 12 times cash and an 8% free-cash yield. Charge the $491 million of stock-based compensation the company adds back [2], as Stock Comp and Cash argued a buyer should, and the same price becomes 32 times cash and a 3% yield.

Sources: derived from reported financials — cash flow [3] and income statement [4]; price and consensus per market data as of July 10, 2026.

The same split runs through the earnings multiple. Reported GAAP diluted earnings were $0.90 a share [5], which puts the stock at about 22 times trailing earnings. The Street quotes it on adjusted EPS instead — roughly $1.78 for 2025 and $1.85 expected for 2026 — where it trades near 10.5 times, cheap for a software business. The gap between $0.90 and $1.78 is almost entirely the stock compensation added back to reach the adjusted figure. So the multiple is not one number; it is a choice about a single, quantified cost.

Sources: market capitalization and EV from market data (July 10, 2026) against balance-sheet cash and short-term investments [6]; multiples derived from FY2025 income statement [7] and cash flow statement [8]; consensus estimates as reported.

The context for all of it is a multiple that has already collapsed. At the 2021 peak the market paid about 20 times revenue; today it pays under 3 times enterprise value to sales (The Business). The debate is no longer whether the premium is gone — it is — but whether even the current price still asks too much of a business now growing near 10%.

What the price implies for growth

A reverse-DCF turns the question around: hold the price fixed and solve for the growth rate that justifies it. Discounting the operating business at 10% with a 3% terminal rate, the $8.3 billion enterprise value implies free cash flow compounds at roughly negative 2% a year for a decade — if reported cash flow is the starting point. Widen the discount rate to a 9%–11% band and the implied rate stays between about −3.5% and breakeven. On that basis the market is paying for a business whose cash flow never grows again, which sits oddly against 10% consensus revenue growth and a 27% cash margin.

Start instead from the $305 million of free cash flow left after charging stock compensation, and the same price implies that figure must compound at about 9%–14% a year — right on top of, or slightly above, the ~10% revenue growth the Street expects.

Source: derived from FY2025 reported free cash flow of $796M and stock-comp-adjusted free cash flow of $305M [9]; 3% terminal growth assumed.

This is the crux of the valuation, and it is the same fork Stock Comp and Cash reached from the other direction. The two bars are not a modelling artefact; they are the price viewed through two honest definitions of owner earnings. On reported cash, the stock looks priced for decline and therefore cheap. On cash net of the compensation cost, it looks priced for exactly the growth analysts already forecast — fair, not cheap. Nothing about the discount rate settles it; the stock-comp question does.

A range, not a target

A forward view gives the same answer with the drivers made explicit. Projecting revenue three years to 2028 under three paths, applying an exit multiple to each, adding back net cash and discounting at 10% produces a spread from roughly $12 to $28 a share. The base case — revenue holding near 10% growth and the EV/sales multiple staying at today's ~3 times — lands within cents of the current price. In other words, at $19.53 the market is already paying for a decade-plus of steady low-double-digit growth at an unchanged multiple; the upside case needs both faster growth and multiple expansion, and the downside case needs only that the multiple keep compressing.

Source: derived from FY2025 revenue of $2,896M [10] and net cash of $1,303M [11]; 10% discount rate; current price $19.53 as of July 10, 2026.

The range is most sensitive to the exit multiple, not the growth rate. Moving the 2028 multiple by one turn of EV/sales shifts value by roughly $6 a share — more than the gap between the 6% and 14% growth paths. A reader who believes the open-internet story (The Independence Moat) re-rates the multiple is making a larger bet than one who believes growth reaccelerates.

The Street's mark

The published targets sit above the price but carry little conviction. The mean analyst target is $24.42 and the median $24.50 — about 25% above $19.53 — inside a wide $11-to-$38 range. Yet the rating distribution is neutral: 19 holds against 13 buys and four sells. And the estimate trend runs the wrong way. In the week before this snapshot, 18 or 19 analysts cut their 2026 EPS estimate while one or two raised it, and the first-quarter 2026 result missed the consensus adjusted EPS by 12%. A target 25% above spot alongside falling estimates and a hold-weighted book describes a market that thinks the stock is worth more than today's price but is unwilling to underwrite it while the numbers are still being marked down.

Source: consensus analyst estimates and price targets, as reported (July 10, 2026); Q1 FY2026 revenue of $688.9M vs $616.0M a year earlier [12].

What would change the read

On balance the evidence points one way: on reported cash flow the stock is genuinely cheap, but that cheapness rests on adding back a compensation cost that recurs every year and that Stock Comp and Cash sized at 111% of net income. Charge it, and The Trade Desk is priced roughly fairly for the ~10% grower The Growth Engine described — not the bargain the headline 12x free-cash multiple suggests. The strongest fact against that reading is the balance sheet: $1.3 billion of net cash, no funded debt [13], and buybacks running at $1.4 billion a year into a depressed price [14] — real optionality the DCF does not capture if growth stabilizes above the ~10% base case.

Two lines decide which way it resolves. The first is revenue growth: a return to a sustained low-teens rate pushes the scenario toward the bull path, while a drift toward mid-single digits validates the bear multiple. The second is stock compensation in absolute dollars — flat near $490 million for four years, with the CEO option that inflated it now rolling off (Stock Comp and Cash); if it falls in 2026, the reported and adjusted cash-flow figures converge and the two multiples in this chapter begin to close on their own. Both are checkable in the next two filings.

---

Control and Execution

The Trade Desk is a founder-controlled company whose long-run credibility rested on a decade of hitting its own forecasts — a record that broke in late 2024. Co-founder and CEO Jeff Green holds 49.7% of the vote, so the reacceleration case rests on trusting management to fix a stumble it says was self-inflicted. The evidence is two-sided: a genuine long-term vision and disciplined buybacks on one side; a first-ever guidance miss, a self-imposed reorganization, and an unusually unsettled senior bench on the other.

Sources: 2026 Proxy Statement, beneficial-ownership table [1]; Q4 2024 earnings call [2]; FY2025 Annual Report, Consolidated Statements of Stockholders' Equity [3].

One holder, near half the vote

The dual-class mechanics were established in The Business; what matters here is how concentrated the resulting control is. As of March 2026, Green beneficially owned 42.1 million Class B shares — 97.6% of the class — and 2.6% of the Class A, giving him 49.7% of total voting power in his own hands; all directors and officers as a group hold 49.8% [1]. The four largest institutional holders — Vanguard, BlackRock, State Street and Baillie Gifford — together carry about 16% of the vote against Green's near-half [1].

The practical consequence is that outside shareholders have little formal recourse if execution disappoints. Board seats, the compensation plan, and strategic direction all effectively answer to one person. That concentration cuts both ways: it lets Green make the long-horizon bets ad-tech peers rarely can, but it also means the reacceleration thesis is, to an unusual degree, a bet on a single manager's judgment rather than on an accountable board.

A forecasting record that broke

For most of its public life, the company's defining feature was reliability. On the Q4 2024 call in February 2025, Green opened by acknowledging that "for the first time in 8 years, we missed the expectations we set, and it was our fault" — the first shortfall in 33 quarters, excluding one COVID-era revision [2]. He framed it as execution, not market: "This wasn't due to a smaller opportunity than anticipated, nor did competition play a role… we made too many turnovers" [2]. The quarter itself was not a collapse — revenue of $741 million still grew 22% year-over-year [4] — but the break in the streak, from a company that treated guidance "like a commitment," was the event that reset the stock's credibility.

The deceleration did not stop with the miss. Growth continued to step down through 2025 and into 2026: Q1 2026 revenue was $689 million, up 12%, and the company guided Q2 2026 to "at least $750 million" — implying single-digit growth [5]. The quarterly trajectory is decomposed in The Growth Engine; the management point is narrower. Whether the slowdown is cyclical or structural, management chose to attribute the initial miss to its own missteps — a claim that is only vindicated if the numbers stabilize, and so far they have kept sliding.

The response: the largest reorganization in its history

Management's answer was structural, and self-imposed in the middle of a slowdown. Green described "four significant changes," the first of which was "the largest reorganization in company history" in December 2024 — a re-drawing of roles and reporting lines for most employees, a split of client-facing teams into brand-focused and agency-focused units, a return to "smaller, agile teams" with the engineering group divided into "nearly 100 scrum teams," and a heavier internal-operations focus paired with senior hiring [6]. Running alongside it was the migration of all clients from the Solimar platform to Kokai, which Green conceded "did proceed more slowly than we expected" [7].

This is the double edge of founder control in a single episode. A reorganization of that scale, executed while growth was already decelerating and a platform migration was underway, is either decisive recalibration or added disruption at the worst possible time — and the two readings are hard to separate from outside. Management's own framing ("something we can control") is the optimistic one; the risk is that a self-inflicted stumble was met with a self-inflicted upheaval.

An unsettled senior bench

The reorganization was accompanied by turnover in the executive suite that is heavy for a company this size. The chief financial officer's chair alone has changed hands three times since 2023, and the most recent transitions came in quick succession.

Sources: Q1 2023 earnings release, CFO transition [8]; 2026 Proxy Statement, executive equity [9]; Q1 2026 CFO-appointment 8-K [10]; company leadership page [11].

Alex Kayyal was appointed CFO in August 2025, having joined the board only in February 2025; by mid-2026 he had separated and the seat passed — via interim CFO and chief accounting officer Tahnil Davis — to Nate Olmstead [9] [10]. A new chief operating officer, Vivek Kundra, also commenced in 2025 [11]. And in May 2026 — reported by the trade press hours before earnings — Chief Strategy Officer Samantha Jacobson left to join OpenAI, though she remains on the board [12].

Green's response was to reframe the churn as upgrade: the company has "quietly been assembling a very strong team of meaningfully senior leaders," and Olmstead (ex-Logitech CFO) and Kundra (the first U.S. federal CIO) are seasoned outside hires [12] [11]. That is a fair defense — bringing in operators to professionalize a founder-run company is a recognizable maturation step. But a finance chair that has held three names in under a year, arriving as growth decelerates, is a difficult signal to read as pure strength.

A pay package built for a share price it no longer has

The compensation structure aligns Green tightly to the stock — which now works against him. His cash pay is modest: a $1.35 million base with a revenue-linked target bonus of about $2.7 million [13]. The wealth is the 2021 CEO Performance Option: the right to buy 16 million Class A shares at a $68.29 strike, vesting in eight tranches only if the stock reaches price targets ranging from $90 to $340 [9]. At today's $19.53, the option is entirely out of the money — the stock would need to more than quadruple just to clear the lowest vesting target, and to exceed the strike at all.

The award's reported cost has whipsawed with the share price. In 2021 Green's Summary Compensation Table total was $835 million — one of the largest single-year figures ever recorded — while "compensation actually paid," a mark-to-market measure, reached $1.2 billion; in 2022, as the stock fell, that same measure swung to negative $620 million [14].

Source: 2025 Proxy Statement, Pay Versus Performance table (the table stops at 2024, before the 2025 collapse) [14].

Two things follow. First, the alignment is real: with a now-worthless mega-option, Green's fortune moves with the same stock his shareholders hold. Second — and this connects to Stock Comp and Cash — the company still booked the accounting expense of that option along the way, so shareholders bore a GAAP cost for an incentive that has so far delivered nothing exercisable. The roll-off of that charge as the option finishes expensing is part of why reported and stock-comp-charged cash flow may converge over time.

Capital allocation: the disciplined counterweight

Against the execution wobble, the record on capital is the strongest management evidence for the bulls. The balance sheet is unlevered — roughly $1.4 billion of cash and no funded debt at Q1 2026 — and the board has repeatedly topped up the buyback: a $700 million program authorized in February 2023, increased to a $1 billion available balance in January 2025, with a further $500 million in October 2025 and $350 million in February 2026 [15].

Crucially, the company leaned in as the price fell. In FY2025 it repurchased 26.2 million shares for $1.4 billion [15], and for the first time the buying outran dilution: period-end shares fell from 496.1 million at the end of 2024 to 475.9 million at the end of 2025, and to 471.0 million by March 2026 [3].

Source: FY2025 Annual Report and Q1 2026 Form 10-Q, Consolidated Statements of Stockholders' Equity [3].

The caution, developed in Stock Comp and Cash, is that across 2020–2024 the buyback mostly offset roughly $500 million a year of stock-based dilution rather than shrinking the count. What changed in 2025 was the price: the same dollars bought far more shares once the stock had collapsed. Read charitably, that is exactly the discipline you want — heavier repurchases into a depressed price, funded from cash rather than debt. Read skeptically, it is a management team buying its own stock on the conviction that its execution problems are temporary.

The read

The reacceleration case that underpins the valuation (What the Price Implies) depends on management executing through a stumble it has diagnosed as its own. The evidence for trusting that is genuine: a decade of met commitments before 2024, a coherent long-run open-internet vision, an unlevered balance sheet, and buybacks that turned the price collapse into a real reduction in share count. The evidence against is equally concrete: the forecasting record broke in late 2024 and growth has since stepped down toward single digits, the fix was the largest reorganization in the company's history layered onto a slow platform migration, and the senior bench — three people through the CFO seat inside a year, a new COO, a strategy chief gone to OpenAI — is more unsettled than at any point as a public company. On balance, control is a double-edged asset here and the near-term record is mixed; the burden of proof now sits with management. What would move the read toward the bulls is straightforward and observable: guidance met again for several consecutive quarters and the executive suite stabilizing. What would move it the other way is another "our fault" quarter — because a second self-inflicted miss would make the structural explanation harder to keep at bay.