Pinterest: $300 million of free cash flow, $292 million of stock pay, and a hold at $19.92

Pinterest generated $300.3 million of free cash flow in the third quarter. It also booked $291.7 million of stock-based compensation, which is pay in shares that the cash flow statement adds back. Subtract one from the other and $8.6 million is left, about 0.7% of the $1,174 million of revenue. That subtraction is a crude alarm and not my valuation, but it is why a company with a 25.6% free cash flow margin and a 28.9% adjusted EBITDA margin needs a second look.

At $19.92 on October 29, 2026, the shares trade 39.3% above my Base value of $14.30, so the Base is 28.2% below the price. The Bear value of $5.06 is 74.6% below the price and the Bull value of $22.13 is 11.1% above it. A deep stress case gives $1.25, which is a stress test and not a floor. I assign no probabilities to the cases. A 30% margin of safety gives a buy range of $6.10 to $12.51 and a conservative limit of $6.10. Pinterest released the results at about 4:06 pm New York time, after that close. The stock traded sharply lower in extended hours, and one report put it at $17.19, down 14.2%, shortly after the release. I do not have a regular session close after the results, so every comparison with the price below uses $19.92.

Valuation route: internet advertising platform, valued as free cash flow to common shareholders. Economic classification: provisional Standard, with an evidence range of 4 to 12 of 12 and Compounder not established. Primary method: ten years of cash flow to common shareholders discounted at the cost of equity, with new stock pay counted as a cash cost and existing unvested awards counted as claims. Cross-check: the same model with five explicit years. Values are present fair-value estimates, not 12-month price targets.

What Pinterest is

Pinterest is an advertising business built on visual search. People use it to plan a kitchen, a wedding or a wardrobe, and advertisers pay to appear next to that intent. It reports one segment and had 661 million monthly users in the third quarter, up 10%. Management says about 57% of them belong to Gen Z. Users are not revenue, though, and the split by region shows why.

The U.S. and Canada produced about $894 million of third quarter revenue, or 76.2% of the total, at $8.38 of revenue per user. Europe produced $193 million (16.4%) at $1.21, and the rest of the world $87 million (7.4%) at $0.23. About 394 million users sit in the rest of the world. If they ever monetize at even a fraction of the North American rate, revenue rises a lot. They have not done that so far, and more users in a low-revenue region do not repair weak advertiser demand.

The product list is long: AI visual search, a shopping assistant, automated performance ad tools, app-install ads and connected-TV inventory from the tvScientific acquisition (about $465 million, per the filing). They are options on higher revenue per user. I do not value any of them as a separate profit stream, because the filings do not show their returns yet.

The latest quarter

Revenue growth slowed from 18% in the second quarter to 12%. Management put roughly 3 points of the slowdown on European cross-border sellers cutting budgets after regulatory changes, 2 points on the timing of Prime Day and World Cup demand, and 1 point on currency. Those are the company's own estimates, and I have no advertiser-level data to test them. What the numbers do show is that Europe is the weak spot and North America is not. Management also said ad impressions grew 17% while the price per ad fell 4%, which is volume without pricing power.

The EBITDA line landed inside guidance, but it landed $5.8 million under the midpoint of $345 million, and revenue missed its range by $16 million at the low end. Management said third quarter costs benefited from a contractual infrastructure item and that fourth quarter costs will include more GPU and cloud spending.

What the earnings are made of

Stock pay is 24.85% of third quarter revenue and 25.2% over nine months. The last row is not a valuation input, because the pay was earned over past and future service, shares are withheld for taxes and vesting timing differs from the accounting charge. It does show that "record free cash flow" and "profit for owners" are two different statements here.

My model treats the two parts of stock pay separately. Shares already granted and unvested (58.2 million) are claims on the equity, with a finite credit for the service still to come: $1,185 million of unrecognized cost over about 2.0 years at September 30. Future grants are replaced by a cash cost, so I do not add a dilution factor on top, which would count the same labor cost twice. In the Base that cost starts at 21% of revenue and falls to 12% by year ten, against about 25% in 2026. That decline is the single largest bet in the valuation.

Taxes get the same caution. Pinterest carries a $1,603 million deferred tax asset on its balance sheet. I let it shield only $500 million of tax, used at no more than 15% of positive taxable income a year, and charge the full 23% after that.

Capital, shares and debt

At September 30 there were 499.4 million Class A and 73.3 million Class B shares, or 572.6 million in total. Adding the 58.2 million unvested restricted units gives 630.8 million economic units, down 3.7 million from 634.6 million at June 30. The legal count rose 7.1 million and the unvested pool fell 10.9 million. Third quarter grants were 5.5 million units against 12.4 million released and 4.1 million forfeited, which is real progress. One quarter of fewer grants does not prove a permanently lower cost.

There are also 8.55 million employee options with an average strike of $19.96, priced with a Black-Scholes model at 45% volatility. The $1,000 million of 1.75% convertible notes mature on March 1, 2031 and convert at about $22.72 a share (44.0 million shares), with a capped call up to $30.59. I put the coupons and the $1,000 million repayment into the cash flows once, at their real dates, and price only the net premium above $30.59 as a separate claim. I do not subtract the notes again as debt.

Cash and marketable securities were $1,470 million. I treat $500 million as the working cash the business needs and count the other $970 million as excess. A new $700 million secured revolver was undrawn at filing. It adds flexibility, not cash, and it comes with a net leverage covenant of 3.5 times.

Buybacks changed the picture this year. In nine months the company repurchased 111.4 million shares for $2,039.8 million, an average of $18.31 (my arithmetic), against a Base value today of $14.30. That is a comparison with today's model and not proof that anyone overpaid when the shares were bought. The company bought no shares under its programs in the third quarter, and $1,948 million of authorization remains, which is permission and not cash. I model no future buybacks.

Primary valuation: free cash flow to common shareholders

The method values the cash that can reach common shareholders after the business has paid for its growth, its taxes and its notes. I forecast ten rolling years ending October 30 and discount at a 12.5% cost of equity: the 4.98% ten-year Treasury yield (October 29) plus a 4.75% equity premium times a risk loading of 1.6 gives 12.58%, rounded to 12.5%. The 1.6 is my loading, not a measured beta. The starting point is a 2026 revenue anchor of $4,811 million, which is nine months of reported revenue plus the midpoint of the fourth quarter guidance. Base assumptions, all estimates:

• Revenue growth of 10% in years one and two, fading to 3% by year ten.
• A cash EBITDA margin before stock pay of 28.5% rising to 33%, and capex of 1.5% of revenue.
• Stock pay as a cash cost of 21% of revenue falling to 12%, with a finite credit for service on awards already granted.
• Terminal growth of 3% after year ten, with the same investment and cash-buffer needs as the forecast years.

This gives $14.30 a share. The value after year ten is 47.3% of the total, which is material and is why I test the discount rate, terminal growth and the stock pay path below. The year-five figure is negative only because the $1,000 million notes are repaid from cash that year, and the present value at their actual date is in the bridge.

Secondary valuation: five explicit years

The cross-check runs the same model for five explicit years and then holds year-five economics forever: growth of 3%, a 32% cash margin and stock pay at 17% of revenue. It gives $11.10, which is 22.4% below the primary value and 44.3% below the price.

This is not an independent confirmation. It uses the same revenue path, tax assumptions, financing and claims, and differs only in how much of the later improvement it believes. The $3.20 gap is the price of that belief: the primary value needs stock pay to keep falling from 17% to 12% of revenue and margins to keep rising after year five, and the cross-check does not. The terminal part is 60.7% of the cross-check value, in a five-year model. I found no independent long-horizon forecast of cash to common shareholders to blend in, and no analyst target enters this analysis.

Historical cross-check and growth diagnostic

At $19.92, the 572.6 million shares counted at September 30 are worth $11,407 million. Adding the $1,000 million notes and subtracting all $1,470 million of cash and securities gives an enterprise value of about $10,937 million, or 2.34 times trailing revenue of $4,680 million. This leaves out the unvested awards and the options, so it flatters the multiple and does not replace the bridge above. I did not refresh peer multiples for this update, and a consistent owner-cash history for 2019 to 2026 is not available to me.

A price to earnings growth ratio is not meaningful. Reported earnings are depressed and include the stock pay, and mixing them with a forecast on a different definition would give a number that looks precise and means nothing. Growth is valued through the explicit revenue and cost paths instead.

The test that matters is whether the cash flow keeps growing after the notes are repaid. From year six, once the awards already granted have been expensed and the notes are gone, Base cash to common shareholders grows from $817 million to $1,302 million in year ten, or 12.4% a year. That is a structural path with execution risk, and it depends on the falling stock pay. In the Deep case the same ratio is optical: cash rises from $8.8 million to $16.2 million, from almost nothing.

Valuation triangulation

I keep the ten-year value of $14.30 as the central estimate and show the five-year $11.10 beside it. Three ranges answer three different questions and none is a probability interval:

• Operating cases: $5.06 to $22.13 (deep stress $1.25).
• Method range at Base inputs: $11.10 to $14.30.
• Selected input and method set: $8.72 to $17.87, from nine combinations of discount rate and stock pay for each method.

The price of $19.92 is above the method range and above the selected set, and it falls inside the operating range only because that range runs up to the Bull case. A lower quote after the release does not make the stock cheap: the revenue miss and the lower fourth quarter outlook weigh on my growth and margin path as well.

What is the market already pricing in?

Solving the Base model backwards for $19.92, one driver at a time:

These are alternatives, not a market forecast. The price needs either a required return 2.6 points lower than mine or a margin that no quarter reported so far points to. A $300 million swing in cash moves the value by only about $0.46 a share, so the balance sheet cannot close the gap. A lasting cut in stock pay could, and that is what I would watch.

For a holder who buys now, keeps the company's cash inside the business and receives no dividends, the five-year result is:

The example assumes no dividends, no buybacks and no new borrowing, with retained cash earning 3% after tax. Base returns 3.87% a year, far below the 12.5% cost of equity, because the price today is well above my value. Only the Bull case clears the cost of equity. These are conditional outcomes of each case and not a forecast.

Bear, base and bull scenarios

Bear: growth of 5% fading to 2%, a 28% mature margin and stock pay that stays at 19% of revenue. Base: the path above, which assumes slower growth after this quarter's international weakness and the fourth quarter outlook. Bull: growth of 16% fading to 3.5% and stock pay falling to 8.5%, which is a strong execution assumption. Deep stress: revenue falls 12% and then 8% with stock pay stuck at 20%. Real losses can exceed the model.

Sensitivity: cost of equity and terminal growth, Base operations

Sensitivity: one assumption at a time

A deliberately favorable combination, the Bull operations at an 11.5% cost of equity, gives $25.07, or 25.9% above the price. I show it as a limit of the argument and do not substitute it for the Base.

How the buy price is set

Confidence B starts at a 30% margin of safety. I add nothing for the method gap, because both methods are in the range below, or for the stock pay path, because it is already in the operating cases and the grid. The buy range comes from 18 combinations: two methods, three costs of equity and three stock pay paths, each value multiplied by 70%.

The buy range is $6.10 to $12.51, and the conservative limit, rounded down, is $6.10. Reaching it takes a price decline of 69.4%, and even the top of the range needs a fall of 37.2%. That is a policy outcome of stacked cautious assumptions, not a forecast that the stock will fall, and not an order: it is a conditional re-entry trigger. If the drop came from a broken thesis, the limit is void. A single price inside a range this wide would pretend to a precision the inputs do not have, so I publish the range and the conservative end.

Dividend and buybacks

Pinterest pays no common dividend and has no stated policy for one, so this stock does not suit anyone who needs income. The $1,000 million of convertible notes pays 1.75% to its holders and not to shareholders. Buybacks are discretionary: none were made under the programs in the third quarter, and the remaining $1,948 million is authorization, not cash. The valuation counts cash that could reach shareholders, not cash that will be paid out.

Reasons to own PINS

• Reach and intent. 661 million monthly users, up 10%, using the service to plan purchases. That is a useful position for advertisers if monetization catches up.
• A strong North American engine. Revenue in the U.S. and Canada grew 14% and revenue per user 10% to $8.38, in the largest and most profitable region.
• Large international headroom. Europe earns $1.21 per user and the rest of the world $0.23, against $8.38 in North America. Even partial convergence adds revenue, if local advertisers show up.
• Specific product levers. Automated performance ads, measurement tools, app-install ads and connected-TV inventory are testable paths to higher revenue per user, though none has shown separate returns yet.
• Real balance sheet progress. $1,470 million of cash and securities, an undrawn $700 million revolver, and unvested awards down 10.9 million in one quarter.

What could go wrong?

About 47% of my Base value comes from the cash flows after year ten, so discount rate and terminal margin carry more weight than any single quarter. The Deep case assumes some franchise value remains, and in a real failure capital can be lost in full.

Management execution

Revenue of $1,173.9 million fell below the $1,190 to $1,210 million range management itself set, by $26.1 million against the midpoint. Adjusted EBITDA met its range. On the call the chief executive acknowledged the miss and named the international sellers, and the company described changes to its go-to-market team. A new CFO, James Dibbo, joined shortly before the call, according to the transcript, so I do not credit a new capital allocation policy that has not been tested.

On the numbers, cash conversion is strong and accounting is consistent: the 10-Q disclosed the notes, the award flows and the credit facility clearly. The weaker points are the missed guidance, a stock pay bill near a quarter of revenue and a year of buybacks partly funded with debt at an average price above my current value. I have no verified return on the tvScientific acquisition yet. Giving credit for communication is fair. It does not cancel a forecast miss.

Stock Analyza scorecard

I do not rescale the known part: the missing return score is missing evidence, not a poor result. The management score is 1 on a four-point scale after the revenue miss, which is a fact about the business and not a reaction to the share price. Management credibility on its own scores 10 to 12 of 20.

The economic classification is provisional Standard, with an evidence range of 4 to 12 of 12. Compounder is not established: it would need several quarters of better regional revenue per user, positive international growth, lower stock pay and a financing path that leaves owner cash after 2031.

Confidence score

This measures how well the inputs and the model are supported. It is not the chance that the stock reaches the Base. The weakest part is the forward path: the fall in stock pay and the 33% terminal margin are my estimates, and the company has not guided to them.

What I would watch from here

The three most important thesis breakers

1. International weakness spreading to the U.S. and Canada, where three quarters of revenue are earned.
2. Stock pay staying near 25% of revenue, so the 12% cost in my Base never arrives.
3. AI and cloud costs absorbing the margin gains before the $1,000 million notes fall due in 2031.

Stress test and independent check

I rebuilt the ten-year model separately from the inputs and matched all four scenario values to the cent ($5.06, $14.30, $22.13 and $1.25), the five-year value of $11.10, the 18 buy limits, every sensitivity above and the reverse valuation. I also checked the reported figures against the SEC earnings release, the third quarter 10-Q and the second quarter release, including the third quarter guidance range. That checks the arithmetic and the reported facts, not the forecasts. The main challenges and how I handled them:

The inputs that matter most are the stock pay path, the terminal margin and the cost of equity. The strongest bull argument is that grants really have shrunk (5.5 million units in the third quarter against 59.2 million in the first half), North American revenue per user is still rising, and my 12.5% hurdle is too high, so a stock needing 9.91% is not absurd. The strongest bear argument is that Europe does not recover, pricing keeps falling, GPU costs eat the margin, and the year of debt-funded buybacks bought shares above their value. My $5.06 and $1.25 are not floors. Weak points in the evidence: I did not read the earnings slides, and I have no regular session price after the results.

Final verdict: hold at $19.92, do not add

Pinterest is a large, growing, cash-generating advertiser whose owners pay for that cash in shares. At the October 29 close the price was above my Base and above every entry case, and the results released after the close add to the doubts: a revenue miss, a weaker fourth quarter outlook and flat revenue in Europe. The lower extended-hours quote is closer to my value, but I will not buy a stock because it fell, and the first regular session after the results has not happened.

Verdict: HOLD at $19.92, do not add. Base value $14.30, 28.2% below the price, five-year cross-check $11.10, Confidence B (18.75/25). Required margin of safety 30%, buy range $6.10 to $12.51, conservative buy limit $6.10.

I assume no position. A new buyer has no reason to act: the Base returns 3.87% a year over five years against a 12.5% cost of equity, and the buy limit is 69% below the price. A current holder has a real argument for reviewing the position and none for adding. I would rerun the numbers after the fourth quarter results, or sooner if the first full session changes the price a lot or the filings change the stock pay story.

The revenue miss will fill the headlines. I would keep the $8.6 million: it is crude, but it is what a 25.6% free cash flow margin shrinks to once the staff is paid in cash.

Sources

Primary sources are Pinterest's third quarter 2026 earnings release, its third quarter 2026 Form 10-Q, the second quarter Form 10-Q and the second quarter earnings release, which set the third quarter guidance. Management's explanations of the slowdown come from the third quarter call transcript, and are the company's estimates.

Prices and volumes are from StockAnalysis (S&P Global Market Intelligence data). The after-hours move is from StockStory. The ten-year Treasury yield is from daily par yield data. The cost of equity, the economic share count, the forecasts, the stock pay path and the tax limit are my estimates.

The reference market price is $19.92, the October 29, 2026 regular session close, before the results were released. Balance sheet figures are as of September 30, 2026.