The Trade Desk is Worth $62 Per Share. Here’s Why.
April 25, 2024
See my previous Discounted Cash Flows last month for Coupang and Rocket Lab.
Digital advertising is an industry that moves at the speed of money.
Ad budgets that used to go to print newspapers, radio, and linear television are transitioning to internet display ads, podcasts, and streaming channels.
Consumers preferences and technology changes have a say as well. Privacy concerns about Big Tech hoarding users’ data have led to regulations and lawsuits, while the deprecation of the third-party cookie has large tech companies searching for a more user-friendly identifier to help them place personalized advertisements.
One innovative who’s quickly made a name for itself in the digital world is The Trade Desk (Nasdaq: TTD). As a demand-side programmatic platform, it helps large advertisers place the right ads in front of the right people and on the right devices. “Programmatic” means that there are AI-tuned algorithms, to win bids and place ads in less than half a second and without requiring any human oversight.
Many investors are already quite familiar with The Trade Desk, whose stock has increased 26-fold since its 2016 initial public offering. Even in this fast-moving industry, it has been a perpetual winner — always finding ways to stay ahead of the competition and outperform even lofty expectations.
Yet there’s also another word that often get thrown into the discussion about The Trade Desk, which is “overvalued”.
Critics and bears are quick to point out that even though the company is fantastic, its stock now sells at a Price/Sales multiple of more than 20x and a Price/Earnings multiple of more than 200x.
Even incredibly companies can turn out to be lousy investments if we overpay for their shares. Other high-flyers like Tesla and Amazon serve as reminders that stocks hitting the stratosphere can fall back down to Earth if they’ve been been priced with too aggressive of expectations.
So is all of the good news already fully priced in to The Trade Desk’s stock? What’s a reasonable price for us to pay for TTD shares?
Introducing the Discounted Cash Flow Valuation
To answer that question, we need to bring in a useful new tool.
Doing so involves a discounted cash flow analysis. A DCF estimates future free cash flows — i.e. the cash that a company generates after paying all of its operational and capital costs — and then discounts them to the present day. The end result is a fair value, representing what that shares are worth for investors to pay to be the owners of those future free cash flows.
DCF models are the primary way that Wall Street firms set price targets for stocks. They’re not simple and are a huge time commitment to do properly. Here’s a quick look at the financial magic and voodoo that’s being run in the models.
This past month, I’ve published DCFs on small-satellite launch provider Rocket Lab (Nasdaq: RKLB) and on South Korean e-commerce leader Coupang (NYSE: CPNG). I came to the conclusion that Coupang is fairly valued, though Rocket Lab appears to be significantly undervalued.
Just as before, I always build my DCF models from scratch. I don’t look at other reports or price target estimates because I want to avoid any bias. My inputs are purposely conservative, to result in a price target that investors should be very comfortable in paying. Using conservative assumptions will ultimately result in a lower price target, though certain drivers could provide significant additional upside.
So all of that said, let’s jump right to the punchline. I believe The Trade Desk’s stock is worth $62.55 per share.
Today’s price of $82 suggests that shares are quite a bit overvalued. But as I’ll discuss in the following sections, The Trade Desk innovates quickly and tends to perpetually find ways to outperform expectations.
But before we get into the trenches of my entire model, I’d like to quickly promote our 7investing service.
This article contains institutional-grade research and is only possible to be published for free because of our generous paying subscribers. If you find value in this article, please consider sharing it on social media to help our brand awareness or joining 7investing for just $1 to get similar research reports on all of our official stock recommendations.
Your 7investing membership also includes complete access to our Community Forum, where our advisors and more than 1,200 other investors are discussing stocks on a 24/7 basis.
Now, let’s get into the details of my The Trade Desk DCF.
7investing’s Discounted Cash Flow model for The Trade Desk
Inputs and Assumptions
Revenues
As a demand-side platform (DSP), The Trade Desk provides a way for advertisers to place their ads within others’ websites, podcasts, and connected TV streaming channels. The company takes a portion of the gross ad spend — knows as the “take rate” — in exchange for placing the ads and providing the analytics.
The Trade Desk doesn’t directly control how much their customers’ spend on advertising. It’s much more a function of the state of the economy and how well consumers are responding.
I’ve modeled The Trade Desk’s revenues as a “top down” approach. I’ve first projected the total advertising spend across a variety of formats, and then drilled-down to figure out what The Trade Desk’s cut of it would be.
Here are my assumptions used to estimate the global ad market’s total spend:
Global Ad Spend can be classified into five categories: Digital, TV, Print, Out of Home (billboards), and Radio.
At the very top-line, I forecast Total Global Ad Spend will grow 6% in 2024 and be mid-single-digits for most years. It will gradually decline to 3% and hit $1.6 trillion by 2038. That’s an overall compound annual growth rate of 4.6% per year during my 15 year forecast.
Out of Home will continue to grow slowly, at just 1% annually throughout my DCF window.
TV, Print, and Radio all see ad budgets decline; primarily because these budgets move to digital.
Digital Ad Spend will be the fastest-growing category, growing at a consolidated 10% in 2024 and moderately slowing to 4% by 2038.
That fast-growing Digital Ad Spend category is the most interesting for the purposes of this discussion. It can be further broken down into five sub-segments:
Walled Gardens (29% of total Global Ad Spend in 2023) – Owned & operated sites like Facebook or YouTube, where the company sets the pricing and collects all of the advertising revenue.
Open Programmatic (10%) – Platforms like The Trade Desk and DSPs that place ads on other publishers’ sites and collect a % fee for doing so.
Search – (23%) – Search platforms like Google or Microsoft’sBing where ads are paid for companies to appear in search results
Reservations (1%) – Direct placements into publishers’ inventory with agreed upon pricing & duration
All Other (1%) – Not accounted for in any of the above.
I have a lot of thoughts about a lot of those segments.
The first is the “walled gardens”, which many believe will crumble down due to their lack of price transparency and regulators hating Big Tech companies hoarding personal information.
I don’t necessarily buy that. I believe our internet was created in a way so that we could visit websites for free, which means they needed their costs to be subsidized by advertising. The walled gardens — which provide our every day web browsing activities like search, social media, music, and video streaming — are able to exist because of advertising. I think the walled gardens — especially Meta, Google, and Amazon — will continue to grow at double-digit rates for at least the next five years. It’s also worth noting that the nine largest walled gardens account for 98% of the total global revenues of this category.
The Trade Desk is part of the Open Programmatic category. This allows large platforms to place advertisements into others’ sites that they do not directly own or operate, then charge a “take rate” of 10%-20% for placing the ads and providing the analytics.
The Trade Desk’s job is to use data to give advertisers and edge and an attractive ROI on their total ad spend. They train their AI with as much data as possible, to optimize the outcome requested by their customers. Customers tend to stick around and spend more over time.
I expect The Trade Desk and Google to be a two-horse race when it comes to the DSP buy-side. Not only will the entire Open Programmatic segment grow at 10% annually (as advertisers transition their budgets from traditional TV, Print, and Radio). But there will also be a market share shift within Open Programmatic, as smaller Buy-Side platforms can’t compete with the sheer amount of data and ROI efficiency of the larger players.
I estimate The Trade Desk will place $97 billion of gross ad spend by 2038 and capture $19 billion in revenue from that (a 20% take rate).
Elsewhere, I see Search maintaining its 20% market share of total ad spend even in the ChatGPT-world. Direct reservations will become inefficient and unnecessary by the end of the decade.
So the result of this top-down ad industry modeling results in $2.4 billion of Trade Desk revenue in 2024, which rises all the way to $19.5 billion in 2038.
Costs of Goods Sold
The Trade Desk’s costs of goods sold are primarily the computing required for the massive amount of data they’re processing. This is something that can moderately scale over time.
Gross margin was 81% last year. I expect that will gradually improve to 85% over the next 15 years.
Operating Expenses
The operating expenses is where things got tricky, because The Trade Desk is a people business. It hires smart software developers to optimize its real-time bidding algorithms. It hires efficient sales people to manage relationship with large advertisers and to close new deals. And it pays its employees — and especially its visionary CEO and Founder Jeff Green — very handsomely with bonuses and stock-based compensation.
So after scouring through the 10K and the proxy, I determined that employee headcount would actually be one of the key inputs of the DCF model. The vast majority of TTD’s operating expenses are people-related costs. Not only stock-based comp, but also sales expenses/commissions, facilities (especially internationally), marketing budgets, and administrative support/overhead.
During the past three years, The Trade Desk’s Revenue Per Employee has expanded from $541k in 2020 to $642k in 2023. Its operating expenses per employee also expanded, from $332k in 2020 to $443k in 2023. During those three yeras, headcount doubled — from 1,545 in 2020 to 3,115 in 2023.
With as much as The Trade Desk talks about international expansion in its IR presentations, I don’t expect the hiring to slow down any time soon. I’m modeling for headcount to expand another 15% in 2024, then taper down slowly over time. I am estimating the company will have 12,661 employees by 2038. This is an extremely scalable platform; i.e. there’s not as much hand-holding required for customers to spend more money on advertising. I expect Revenue per Employee will increase to $1.5 million by 2038. And Operating Expenses per Employee to increase to $689,000. These numbers include all stock-based compensation. And as I’ll show in the next section, that’s not exactly evenly distributed as $689,000 per employee. It’s actually highly-skewed by the massive stock incentive plan of Jeff Green and the company’s top-brass executives.
Regarding how the operating expenses are categorized:
TTD has historically kept R&D constant at 20% of total revenue. I believe that’s a managerial target, so I’ve kept it constant throughout the forecast.
Sales, General, and Administrative includes stock-based comp, facilities costs, and marketing budgets. I took the increase in each of these during 2023 and divided it by the new employees hired, to give an estimate of the total OpEx burden per new employee.
The SG&A expense is a sum-of-the parts of each of those items.
R&D remains constant as 20% of revenue in all years.
SG&A as a % of revenue declines from 50% in 2023 to just 25% in 2038.
GAAP Operating Margin increases from 10% in 2023 to 40% in 2038.
Green’s Lucrative Incentive Plan
10% last year to 40% in 2038?! That’s a huge increase in profitability!! What’s the catch???
A huge portion of TTD’s operating expenses are the Stock-Based Compensation expense that is tied to CEO Jeff Green’s incentive plan.
The punchline is that Green gets a boatload of stock-based comp, but only if TTD’s stock hits certain price thresholds. But due to GAAP accounting requirements, it estimates a certain portion of those in each upcoming year, even if the stock doesn’t hit the threshold and the new shares don’t get issued. When that happens, TTD reduces its SBC expense for the year.
Here’s an example from 2023:
“General and administrative expense decreased by $5 million, or 1%, for the year ended December 31, 2023, as compared to the year ended December 31, 2022. The decrease was primarily due to a $47 million decrease in stock-based compensation, partially offset by increases of $36 million in personnel costs and $5 million in allocated facilities costs. The decrease in stock-based costs was primarily driven by a $64 million decrease in stock-based compensation cost related to the CEO Performance Option driven by the graded-vesting attribution method, under which more expense is recognized earlier in the option’s life.”
The point is that the vesting of Green’s stock options based upon TTD’s share price will have a huge impact on the valuation of the company. It’s difficult to make assumptions around this. Nostradamus isn’t in my lineage and I have no idea what price TTD’s stock will trade hands at here in the near-term. But I do think a reasonable assumption is that the price target thresholds will trigger every third year. Meaning Green’s shares would execute — and also be recognized — in 2025, 2028, 2031, and so on.
Accounting for this into the DCF is more art than science. But the important punchline is that SBC will be lumpy on an annual basis. It spiked in 2021/2022 because TTD’s price hit $90 and that first traunche got executed. But then actually decreased y/y in 2023 because it didn’t hit $115.
Stock Based Compensation and Share Repurchases
Another important consideration is that TTD is offsetting any dilution by immediately repurchasing shares issued as stock-based compensation. The share count has remained at around 500m for the past three years. And they have another $700m buyback authorized and ready to deploy, to offset any shares issued this year. This is another interesting twist. Typically when companies pay stock-based comp, we just treat it as a non-cash expense and increase the share count accordingly. But because TTD is buying back all of the newly-issued shares, that means SBC is really more of a cash expense. It’s essentially the company paying cash to the employees/executives that’s equivalent to the value of the shares. So this is definitely a cash expense that will reduce FCF in all years. I’m holding the outstanding share count constant at 499.7 million. But I’m also not adding back the SBC value when making the free cash flow calculations.
Non-Cash Adjustments and Other Assumptions
A few more odds-and-ends:
TTD has a capital lease agreement that I’m essentially treating as long-term debt. It’s estimated at 3.6% cost of debt and a five-year term; fully repaid by 2029.
Share price hits the triggers every third year. $115 by 2025, $145 by 2028, etc. This is what triggers the SBC payouts in those years.
GAAP Net margin increases from 10% in 2024 to 28% by 2038.
CapEx is difficult to forecast, though it’s tied to international expansion and infrastructure. CapEx was $60m in 2021, $92m in 2022, and then $55m in 2023. I’ve estimated $70m in 2024, rising to $100m annually by 2034, and then settling back to maintenance-only CapEx by 2038.
No acquisitions and no long-term debt borrowing. Expansion is organically funded.
I used a discount rate of 10.2%. The Trade Desk has no long-term debt and is almost fully-funded by equity. I calculated the WACC using an average market return of 10.3% and a Beta of 1.0 (TTD’s 3-yr Beta is 0.77 and 5-yr Beta is 1.5).
Capital Expenditures, Depreciation, and Stock Based Compensation
I’ve also made quite a few capital expenditure forecasts and non-cash adjustments:
Fulfillment centers and equipment generally have a 15-year useful life. I’ve tied depreciation to Net PP&E and CapEx.
Depreciation will converge to CapEx in the later years, as Coupang begins spending primarily on maintenance rather than on growth.
CapEx has been explicitly stated by management to be $2.2 billion through fiscal 2027 for investment in Korea + $400 million/year in Taiwan.
I’m modeling $950m of CapEx through 2027, increasing to $1b in 2028, and then gradually walking it down to $500 million/year by 2036.
As a sanity check, MELI spent $509m on CapEx last year. The numbers check-out.
I looked at the proxy and it looks like Coupang’s executives have a reasonable equity compensation plan that won’t result in massive dilution. I’m therefore tying SBC growth to the growth in overall headcount, which grows 12% in 2024 but slows down a bit in later years. SBC never accounts for more than 1.3% of total revenue or 25% of Adjusted EBITDA.
Working capital is tricky due to the stockpiling of inventory. Some years it loads up and other years it works through it. I’m having the two offset one another and am assuming a zero change in working capital needs throughout the DCF.
Acquisitions are also a trump card. We know Coupang paid $500 million for Farfetch this year, but we’re still unclear on the impact it will have on financials.
For the time being, I’m baking Farfetch’s revenues and growth entirely into my expected growth rate of the core platform. This is a conservative assumption that I’ll likely boost in future versions.
Coupang has a defined-benefit pension that is overfunded and a reductions in its estimated discount rate have provided a net cash benefit in recent years. It’s also made a few gains on its short-term investment securities. I’m zeroing out the pension benefit, but I think it’s reasonable to assume that the company can continue to get a 2-3% return on its $5 billion of cash in the bank. So I’ll continue to recognize a $150 million operating gain in each year of the forecast.
I’m also capitalizing the operating lease, which gives it right-of-use for several of its facilities without owning them outright. So I’m recognizing the current value of the liability as a debt, but also adding back the $450 million per year in expenses.
Other Key Assumptions
In addition to everything I’ve mentioned above, here are two final key assumptions:
I assume a terminal growth rate of 3% in all years past 2038.
I assume The Trade Desk will continue to use cash to offset all future stock-based compensation — i.e. we don’t add back SBC as a non-cash expense.
I used a 15 year growth window and assumed a terminal growth rate of 3% for the cash flows beyond 2038. I then discounted all of those future cash flow back to the present.
So What’s The Trade Desk Worth?
So now that we’ve gone through all of the inputs, let’s circle back to that punchline. I estimate The Trade Desk’s present equity value is worth $62.55 per share.This represents the fair value of what The Trade Desk’s shares are worth today.
The Trade Desk’s current stock price of around $82.29 suggests that shares are a big overvalued. That’s pretty much been par for the course though, as this company tends to always find ways to outperform expectations.
One Ace it has up its sleeve is its OpenPath initiative, which allows large advertisers to directly place ads into open inventory offered by large publishers like Disney. This would bypass the traditional real-time bidding auctions entirely, saving advertisers from inefficient/wasted spending and rewarding The Trade Desk with a larger piece of the overall pie.
But this is just the beginning. DCFs evolve over time, and I’ll be updating my model accordingly to keep up with the digital ad industry’s ongoing changes.
I’ll be sharing all of those updates on our 7investing Community Forum. And I would love to invite you to join in on the conversation.
Want to discuss Simon’s assumptions on The Trade Desk directly with him and with other investors?
“Thanks for doing this. So much work goes into these models, that this kind of financial calculus on individual stocks is worth the price of admission by itself.”
“Thanks a lot for working through this publicly, really an excellent exercise.”
“Thanks a lot, Simon. I feel like some of your assumptions are quite conservative, and it is astounding that you get to such a crazy upside anyway. What really makes me believe that the upside is possible is that they keep performing and delivering on their promises.”
“Thanks for the modeling!”
“Thanks for the hard-work, 7innovator! Would be interesting to see if this impacts the conviction rating..”
These are posts in our Community Forum posts from actual 7investing members. Click here to join 7investing’s Community Forum today!
