At $264, Adobe is a 12% premium and a 30% discount at once

A superb earner whose value spans two to one

Adobe sells design and document software almost entirely by subscription. In fiscal 2025, 96 percent of its $23.8 billion in revenue was recurring, earned at an 89 percent gross margin and a 36.6 percent operating margin.¹ After tax, its operating profit was about $7.1 billion on roughly $11 billion of invested capital, a 63 percent return.¹

The market has not priced it like a sure thing. Adobe closed at $264.02 on August 14, 2026, and its market value had fallen about 29.5 percent over the prior year.² The doubt is not about the current numbers, which are excellent. It is about the future: the same wave of generative artificial intelligence that Adobe sells into could just as easily cheapen creative software and shorten the life of its advantage. That doubt lives in one place, how fast the cash grows, and it is enough to move the value a long way.

Earlier lessons built the machine that turns a company into a value: free cash flow, the cost of capital, the discounted cash flow, and the bridge from the whole firm down to a single share. The machine returns a number. But the number is only an estimate, and an estimate of an uncertain future is almost guaranteed to be somewhat wrong, and more wrong the more uncertain the business.³ The right response is not to stop valuing but to value more carefully and to carry the answer as a range rather than a point.⁴ Which raises the question this lesson answers. If the value is a range, what protects an investor who has to buy at one price? Graham gave the answer in three words, a margin of safety: the gap between the price paid and a conservative estimate of value, made wide enough to absorb being wrong. And the margin, he insisted, is a property of the price, not of the business.⁵

Building both ends of the value from one 10-K

To make "a range" concrete, build two honest values for Adobe from the same filing, changing only the estimates the filing does not own. Both start from the same cash base, the same net cash, and the same share count. They differ in five assumptions about the future.

The base for both is Adobe's after-tax operating income, not its reported free cash flow. Operating income was $8.7 billion; taxed at Adobe's 18.4 percent effective rate, \$8.7\text{B} \times (1 - 0.184), that is about $7.1 billion.¹ Reported free cash flow looks larger, $10.0 billion of operating cash flow less $0.2 billion of capital spending, about $9.9 billion.¹ But that figure adds back $1.9 billion of stock-based compensation,¹ a real cost that pays employees in shares and dilutes the owners. Subtract it and the honest figure is $7.9 billion, close to the after-tax operating profit. The conservative base uses the lower number because it already treats that pay as the expense it is.

From that base the value follows the machine the earlier lessons built. The cash grows for ten years at an assumed rate, each year's growth paid for by reinvesting part of the cash at an assumed return on that new capital. What is left is discounted at an assumed cost of capital. A terminal value caps the stream at year ten, growing forever after at an assumed rate, its own growth again paid for by reinvesting at an assumed return on capital. Five estimates, and only those five, separate a cautious reading of Adobe's future from a hopeful one.

Summing the discounted ten-year stream, adding the year-ten terminal value discounted back, adding Adobe's small net cash of about $0.4 billion,¹ and dividing by its 397.5 million shares⁶ gives the value of one share.

On the conservative assumptions the whole enterprise is worth about $93 billion and each share about $236. On the optimistic ones, about $206 billion and $518.¹ Same company, same filing, same night, and a value that runs better than two to one. As the discounted-cash-flow lesson found, most of that worth sits past the forecast. The terminal value is 54 percent of the conservative total and 72 percent of the optimistic one. Growth matters so much here because it is the number that carries the cash into that terminal year. Neither $236 nor $518 is a figure Adobe reported. The filing owns the inputs; the growth, discount, and terminal choices are estimates, and the endpoints are worth exactly as much as those labels.

The same price is a premium and a discount at once

A margin of safety is a single subtraction, 1 - \tfrac{\text{price}}{\text{value}}, the fraction of the value the buyer is not paying.⁵ At Adobe's $264.02 the answer depends entirely on which value goes in the denominator.

Against the conservative $236, the price is 12 percent too high: no margin at all, a premium. Against the $377 midpoint it is a 30 percent discount, and against the optimistic $518, a 49 percent one.² One stock, one price, and it is both dear and cheap by a third at the same time. This is exactly why Graham tied the margin to a conservative estimate of value. An investor who anchors on the midpoint or the high end will almost always find a margin, because the value was chosen to produce one. The discount is real only when it is measured against a value the future would have to disappoint to break.⁵

Read the other direction, the table says what price would buy real protection. For a 30 percent cushion against the conservative estimate, Adobe would have to trade near $165; for 20 percent, near $189.¹ At $264 the only margin on offer is the one measured against the midpoint, a value that already assumes the growth question resolves halfway to the optimistic case. Anchoring there does not find a bargain. It assumes one.

Growth is the input the value hangs on

How wide the cushion has to be is not a matter of mood. It is set by how far the value moves when a single assumption moves. Take the conservative case and change one input at a time, holding the other three fixed.

Lifting the ten-year growth rate from 4 to 11 percent adds $93 a share on its own. The next-strongest input, the cost of capital, adds $39; the two terminal knobs together move the value less than $20.¹ Of the five estimates, one moves the value more than the other four together, how fast Adobe's cash compounds over the next decade. That is what fixes the required margin. A business whose value turns on an input that could plausibly land far apart needs a wide discount to be safe; a steady earner whose inputs barely move needs little. Facing more uncertainty, the answer is to value more carefully and demand a bigger cushion, not to give up, because the aim is to be less wrong than the price, not to be right.⁴ Where the plausible range grows so wide that no honest price leaves any margin, the business is neither cheap nor dear. It is unpriced, and buying it is speculation wearing the clothes of investment.

A good business is not the same as a good investment. Adobe's 63 percent return on capital¹ is what makes the optimistic case even thinkable, but quality is not a margin. A superb business bought at the top of its own value range has none left. The protection lives in the price paid, exactly as Graham said, and nowhere in the quality bought.⁵