$56.35
$225.64
bear $56.35
base $155.27
bull $225.64
margin of safety $90.00
market price* $240.70
Salesforce sells subscription software that large and mid-sized companies use to run sales (Sales Cloud), customer service (Service Cloud), marketing and commerce, plus Slack, Tableau, MuleSoft, the Agentforce AI-agent product and, since its November 2025 purchase of Informatica, enterprise data management. Revenue was $41.5 billion in the year ended January 2026: about 65% from the Americas, 24% from Europe and 10% from Asia Pacific. Customers sign multi-year contracts and pay a year at a time in advance; the company reports losing only about 8% of contract value a year to cancellations, a rate it says has held steady, and at July 31, 2026 it had $66.3 billion of contracted revenue still to deliver, $33.5 billion of it due within twelve months. How we measure the earnings. We start from operating cash flow, strip out the swings in receivables, customer prepayments and payables, and charge as real costs the stock paid to employees ($3.5 billion last year), the sales commissions the company pays to win and renew contracts, rent on its offices, and the equipment and data-center leases it pays down each year. We credit back the part of commissions and equipment spending that went into growth rather than keeping the business where it is. The large yearly amortization of past acquisitions (Slack, Tableau, MuleSoft, Informatica) is a bookkeeping charge for money already spent, so it is not charged again; instead we do not assume those acquired products keep growing forever. On that basis the business produced cash, before paying its lenders, of about $7.1 billion in fiscal 2024, $8.6 billion in fiscal 2025 and $8.4 billion in fiscal 2026. We use the middle of those three years, about $8.4 billion, as today's earning power. Earlier years are left out because the January 2023 restructuring (layoffs, office closures, slower hiring) changed the cost structure; fiscal 2023 and before describe a different, much less profitable business. Growth in customer prepayments added roughly another $1 billion a year of cash over the last decade; we do not count it, which is a conservative choice. Recent trend. Cash earnings slipped about 2% in fiscal 2026 even though revenue grew nearly 10%, mostly because stock pay and commissions rose faster than sales. In the first half of fiscal 2027 (February to July 2026), cash from operations before working-capital swings and after stock pay was lower than a year earlier despite Informatica's added revenue, and organic revenue growth in the July quarter was about 6%. So our central case starts with another 3% decline next year, below last year's actual decline, before the contracted backlog above lets earnings grow again from year two. Informatica is only partly in the earnings base, but we do not add it on top because the recent half-year shows no net lift yet. What changed since our previous analysis. Two things lower the value sharply. First, we previously treated sales commissions and lease payments as temporary timing items and left them out of costs; they are recurring costs of running this business, and charging them lowers the earnings base by about 30%. Second, we previously valued the cash left for owners after interest, assumed growth continued indefinitely and did not deduct the debt itself. Since then the company borrowed $25 billion in March 2026 (eight bond issues due 2028 to 2066 at 4.5% to 6.7%) and spent it on an accelerated buyback of about 129 million shares, and it has since refinanced the Informatica bank loans with a $6 billion five-year term loan. Total borrowings are now $39.5 billion, against $8.5 billion a year earlier and the $14.5 billion we previously described; yearly interest rises from under $0.3 billion to about $1.8 billion. We now value the business before its lenders and then subtract what it owes. The buyback also cut the share count to roughly 795 million once the final shares are delivered in October 2026, which spreads the remaining value across fewer shares. What it owes. Borrowings are $39.5 billion. The $8.5 billion of older bonds carry fixed rates of 1.5% to 3.7%, several running to 2041, 2051 and 2061, and most of the new bonds and the term loan also cost less than 6%, while the 2046, 2056 and 2066 bonds cost a little more; on balance that locked-in cheap money is worth about $2.6 billion to shareholders compared with borrowing at 6%, so we count the debt at about $36.9 billion. We add $4.85 billion committed to buy Fin (about $3.6 billion, signed June 2026, expected to close by early 2027) and Contentful (reported at about $1.25 billion, closed September 1, 2026), and we give credit for about $9.4 billion of cash and marketable securities beyond working needs. The company also holds $11.3 billion of stakes in other companies, mostly private; we count half of that in the bear case, three-quarters in the base case and all of it in the bull case. Equipment leases are counted through the yearly payments above, not again as debt. Net debt is therefore about $26.7 billion in the bear case, $23.8 billion in the base case and $21.0 billion in the bull case. The first bond maturities are $6.0 billion in 2028 and $4.25 billion in 2029; even in the bear case yearly cash before interest stays above 3.6 times the interest bill for the next ten years, long enough to repay most of the debt before the fade, so refinancing these is not a threat to the equity. Base case ($155 per share). Earnings fall 3% next year, grow 2% the year after, then about 5% a year for three years, easing to 3% by year ten, supported by the contracted backlog, the steady 8% cancellation rate and the company's record of raising list prices on its existing products. We do not credit Agentforce winning new markets, share gains or the company's revenue targets. After year ten the customer data and processes built into Salesforce keep earnings flat for fifteen more years, then they fade to nothing over the following fifteen. The business is worth about $147 billion ($185 per share), it owes a net $23.8 billion ($30 per share), leaving about $155 per share. Bear case ($56 per share). The risk the company itself discloses, customers buying fewer user licenses and shifting to consumption-based pricing as AI agents do work people used to do, shrinks earnings 4% next year and 3% in each of the two years after, then they hold flat to year ten and fade to nothing over the following fifteen years. The company keeps buying businesses, as it has with Informatica, Fin and Contentful, and those purchases return little, costing about $1 billion a year of cash for ten years. The business is worth about $71 billion ($90 per share), it owes a net $26.7 billion ($34 per share), leaving about $56 per share. Bull case ($226 per share). Revenue growth close to the roughly 9-11% of the last three years continues on the backlog and demonstrated pricing, so earnings grow 4% next year and 8% a year for the four years after, then taper to 4% by year ten; earnings then hold flat for twenty years before fading over fifteen. The business is worth about $200 billion ($252 per share), it owes a net $21.0 billion ($26 per share), leaving about $226 per share. What would make this wrong. On the downside, faster seat losses to AI or a costly acquisition failure would push toward the bear case. On the upside, if commissions and stock pay stop growing faster than revenue, earnings would rebuild faster than we assume. The spread between cases is wide because the value rests on how long a subscription franchise lasts, not on a contract or patent with a known end date. Because of the new debt, the lower earnings base and real uncertainty about AI's effect on per-seat software, the estimate calls for a wide margin of safety, and the margin-of-safety price is around $90 per share.