Every company still on Veeva's original CRM must move by the end of 2029, and the old landlord has opened a rival building across the street. Fourteen of the twenty largest drug makers are moving with Veeva. The share price assumes a great deal more than that.
At $273.33, its close on October 2, Veeva Systems is priced for a decade of growth that few companies sustain. Take the cash the business earns for its owners after paying its people in stock, $950 million in the first year on the generous count, grow it 15% a year for nine years and 3% a year after that, and the shares offer about 8.7% a year. Start from the $820 million that this year's guidance implies, and they offer 8.0%. To offer 10%, that cash must grow 18.8% to 21% a year for nine years, until Veeva throws off more cash each year than its entire revenue today. Management's own target, a $6 billion revenue run rate by 2030, implies growth of 11% to 14%. The rest must come from much wider margins or from revenue that artificial intelligence has not yet produced. Two numbers decide this, and neither appears in any filing: how much of this year's cash is a tax benefit in its final year, and what an AI agent earns once its model costs and its human reviewers are paid.
Every working day, somewhere in America, a pharmaceutical representative walks out of a doctor's office and records the visit: who was seen, what was discussed, what samples were left behind. The record serves the sales manager and, when they ask, the regulators. At many of the world's largest drug companies it has been kept for years in Veeva's CRM, software that Veeva built on a platform belonging to Salesforce. That arrangement is ending. The two companies' agreement expired in September 2025, Veeva will support the old CRM only until the end of 2029, and every customer still on it has to move: onto Veeva's own Vault platform, or to someone else's.
On September 30, Barron's made Veeva one of its stock picks under the headline "AI Will Revolutionize Medicine. This Company Provides the Plumbing." Dan Victor's case is the one most investors would make. Veeva's software runs through the regulated work of making and selling drugs, from clinical trials and regulatory filings to the sales call. That position, and the data around it, should make Veeva the natural place for the industry to put artificial intelligence. The move to Vault CRM is going better than feared, with 14 of the 20 largest drug makers committed. And Falcon, Veeva's new software for doing clinical, regulatory and safety work rather than recording it, opens a market Veeva has never sold into.
Plumbing is the right word, and it cuts both ways. Plumbing is valuable because nobody wants to replace it. This decade, every one of Veeva's CRM customers has to replace part of theirs.
It is a good hypothesis because it can be tested, in specific ways, against the company's own numbers. This note does that, and then asks what the price already assumes.
Veeva has built its business on one industry. It had 1,552 customers at the end of January, from the largest drug makers to small biotechs, and its ten largest have paid 28% of its revenue in each of the last three years. Most subscriptions run a year at a time, and some contracts raise the price at each renewal by inflation or 4%, whichever is lower. Delivering a subscription dollar costs about 14 cents: the subscription gross margin was 86.2% last quarter.
Two of Adrian Slywotzky's profit models are candidates, and the obvious one does not fit. Installed base profit, selling the machine thin and earning on everything that follows, is how the artificial intelligence thesis is usually told: sell agents into the systems customers already run. Its arithmetic would score Veeva a winner, since between fiscal 2024 and 2026 customers rose 8.4% and subscription revenue per customer rose 30%, to $1.73 million. But the model begins with a thin first sale, and Veeva's first sale is a subscription at an 86% gross margin. A model that fits only after you ignore its first clause does not fit.
The model that does fit is specialist profit: earning more than anyone in one segment by knowing it better than anyone. Slywotzky's example was EDS, which took on one industry at a time (The Profit Zone, 1997, pp. 44 to 45). Veeva took on one industry and, until this August, stayed there. The test is old and simple. Compare the specialist's margin with what the generalists earn, and watch whether the edge holds at home and travels to new ground:
Here m is operating margin; G is the specialist's gap over the generalist, which should hold or widen; and C(k) is how close ground entered k years ago has come to the home margin, which should rise toward one. For the generalist we use Salesforce: Veeva's primary CRM competitor, in Veeva's own words, and the owner of the platform under Veeva's original CRM. Salesforce does not report what it earns in life sciences, so its company margin stands in.
| Fiscal year (ends January) | Veeva | Salesforce | The gap |
|---|---|---|---|
| 2024 | 18.2% | 14.4% | 3.8 points |
| 2025 | 25.2% | 19.0% | 6.2 points |
| 2026 | 28.7% | 20.1% | 8.6 points |
GAAP operating margin, which charges stock pay, from each company's annual results. On the adjusted figures both companies prefer, the 2026 gap is 10.8 points.
At home the specialist is winning, and by more each year. The travel test is half done. Inside the industry the business has traveled across functions: R&D and Quality, Veeva's software for clinical trials, regulatory filings, drug safety and manufacturing quality, overtook the commercial business in fiscal 2025 and is guided at $1.675 billion of subscriptions this year, growing 17.4%, against Commercial's $1.405 billion, growing 11.7%. Whether the margin traveled with it, C(k), cannot be computed, because Veeva reports a single segment. Outside the industry the test has barely begun. In August Veeva announced Aspen CRM for companies outside life sciences, and its chief executive wrote that success would depend mostly on execution and somewhat on luck. It is an honest sentence, and the specialist's warning label.
Software that runs regulated work is protected by the cost of leaving it: the validation, the data, the retraining, the risk of a failed audit. A customer stays while the incumbent's system, plus the cost of leaving it, is worth more than the rival's. A forced move rewrites the inequality:
Here uV and uS are what Vault CRM and Salesforce's life-sciences CRM are worth to a customer, s is the cost of abandoning a working system, and mV and mS are the costs of moving to each. Once the old system must go, s disappears from the left side. The moat becomes the difference between two moving bills, and for a company whose CRM data already lives on Salesforce's platform, the bill for staying with Veeva need not be the smaller one. None of Slywotzky's profit patterns fits the move exactly. It is older economics: switching costs, and outside options.
For the largest accounts, the scoring is nearly done. Veeva ended January with 10 of the top 20 biopharmas committed to Vault CRM and said it expected about 14. In August it had 12, with two decisions still open. On September 23 it announced 14, with more than 190 customers live. Six of the twenty did not choose Vault CRM, and management now speaks of winning them back. Veeva's own filing says some customers have told it they are moving to Salesforce. Salesforce, for its part, says its life-sciences CRM has more than 140 customers, twice as many as it had less than a year earlier, and names Novartis, AstraZeneca, Pfizer, Takeda and AbbVie among those that have selected it for their commercial and medical teams. The two companies count different things, selections on one side and live customers on the other, and customers split their work: AstraZeneca, on Salesforce's list, is presenting its pilots of Veeva's AI for regulatory work at Veeva's research summit in Boston this month.
Fourteen chose Veeva. We do not know on what terms. Each of them chose at the moment its own alternative was strongest, and in a negotiation the outside option decides the split. Veeva's filings warn of growing purchasing scrutiny when large customers renew. Whatever the fourteen negotiated will show up slowly, in price per seat and in the scope of each contract, and not in any count of logos.
Artificial intelligence is the second reset, and it can run either way. Veeva's annual report names the risk plainly: when customers cut sales representatives, they need fewer user subscriptions. Veeva's Agentic Call Report, which one top-20 company has deployed for its entire U.S. field force, is meant to make each representative more productive, and in time that may mean fewer of them. Veeva's answer is to sell the work as well as the seat. Falcon, which the company calls agentic labor, is meant to do clinical, regulatory and safety work itself; five early adopters are testing it, the first are expected to go live this year, and the first top-20 company in the first half of 2027. Falcon MLR, which came with the acquisition of Copli in June, automates the review of promotional material, and management believes it can eliminate at least 70% of the manual labor in that review within five years. Falcon Safety is built to work with any safety system that meets the industry's E2B standard, rivals' included, which takes it beyond Veeva's own safety customers. Some agents are priced by use, others inside a fixed subscription.
Whether this adds to the spending Veeva captures or merely replaces seats with agents is the question on which most of the price rests, and the filings cannot answer it yet. Veeva does not disclose Falcon's revenue or margins. That is a gap in the disclosure, not evidence that the revenue is small. The agents run on large language models bought from outside providers, so every task carries a cost that Veeva does not set.
We rebuilt the ten-year owner-cash model behind this note line by line, reproduced every one of its figures to the cent, and turned it around, in the manner of expectations investing (Mauboussin, 2006). Instead of asking what a share is worth, we ask what $273.33 offers a year on each set of assumptions. Vista does not publish what a share is worth. What a price assumes is the more useful question anyway.
Here P is the $273.33 close; S is 165 million shares, the company's own fully diluted count; Ft is owner cash in year t, the cash from operations left after capital spending, stock pay and the after-tax interest on the cash pile; h is 3% growth after Year 10; C is $7.24 billion of cash and short-term investments, counted once; and r, the unknown, is the annual return the price offers.
| The case | What it assumes | What $273.33 offers a year |
|---|---|---|
| Guidance cash, growth fading | $820 million in Year 1; 15% a year to Year 5, then 12, 9, 6, 3 and 3% | 6.8% |
| Original cash, growth fading | $950 million in Year 1; the same fading path | 7.3% |
| Guidance cash, less $100 million | $720 million in Year 1, a stand-in for a tax catch-up of that size; 15% a year | 7.4% |
| Guidance cash, steady growth | $820 million in Year 1; 15% a year for nine years | 8.0% |
| The original case | $950 million in Year 1; 15% a year for nine years | 8.7% |
| Faster | $950 million in Year 1; 17% a year | 9.4% |
| The strong path | $950 million in Year 1; 20% a year for nine years | 10.5% |
Annual return at which each case's owner cash, plus Veeva's cash and investments at face value, equals the October 2 close, on 165 million shares, with 3% growth after Year 10. Holding back $1 billion as an operating reserve lowers each figure by 0.09 to 0.15 point; counting 171 million shares lowers it by 0.15 to 0.25 point. Inputs and method are at the end of this note.
Turned the other way, here is what the price needs in order to offer 10% a year:
| Hold this fixed | Solve for | The price needs |
|---|---|---|
| Year-1 owner cash of $950 million | Growth, Years 2 to 10 | 18.8% a year |
| Year-1 owner cash of $820 million | Growth, Years 2 to 10 | just over 21% a year |
| Year-1 owner cash of $720 million | Growth, Years 2 to 10 | 23.1% a year |
| Growth of 15% a year | Year-1 owner cash | $1.215 billion |
| Growth of 17% a year | Year-1 owner cash | $1.066 billion |
| Growth that fades after Year 5 | Year-1 owner cash | $1.59 billion |
165 million shares and full cash. On 171 million shares the three growth rates become 19.4%, 21.7% and 23.7%.
Four things stand out.
The case that looks like the company's plan offers 8% to 9%. Growing owner cash 15% a year for nine years lifts it to $2.9 to $3.3 billion by Year 10, 78% to 91% of this year's revenue. If revenue grows at the 11% to 14% a year that management's 2030 target implies, owner cash rises from about 22% to 26% of revenue today to 25% to 35% by Year 10: a stretch, not a leap. That world offers 8.0% to 8.7% a year.
To offer 10%, the cash must grow four to six points a year faster, for nine years. From $950 million or from $820 million, Year-10 owner cash comes to about $4.5 billion, 1.21 to 1.24 times Veeva's entire revenue this year. At the revenue pace management targets, owner cash would have to reach 38% to 48% of revenue. Hold growth at 15%, and the price needs $1.215 billion of owner cash in the first year, 28% more than the generous count.
The record and the base rates side with the slower path. Veeva's operating cash flow, on its own adjusted measure, grew about 15% in fiscal 2026 once a tax tailwind of about $145 million is set aside, and the company guides to 15% again this year. Chan, Karceski and Lakonishok (2003) found that sustained high growth is rare and no more common than chance would produce, though their evidence is on earnings, not cash. The faster path also asks margins to widen for a decade, while Fama and French (2000) found that profitability reverts toward the average at roughly 38% a year. Neither study rules Veeva out. Both say the faster path is the exception, and at a 10% return the price assumes it.
Most of the value lies beyond Year 10. On the original case, 64% of the operating value sits in the terminal value (57% when growth fades). Veeva buys its growth with research and selling expense rather than with capital spending, which ran $26.7 million over the last twelve months against $40.5 million of depreciation and amortization, so the terminal shortcut implies an unlimited return on new tangible capital. Our house discipline pins the return on new capital at 20%: growth after Year 10 must be paid for at that rate. Pinned, every case offers about half a point less: the original case 8.1%, the guidance case 7.5%.
This reminds me of Luke 14:28 (WEB), on the builder who should first "count the cost, to see if he has enough to complete it." A reverse valuation counts in the other direction. The price has already paid for the tower, and the question is whether the business has the means to finish it. At the 15% pace of its recent record, Veeva builds 63% to 75% of the Year-10 cash the price needs at 10%. The top floors are the part that artificial intelligence has to build.
Is 8% to 9% a year enough? That is the reader's decision, and the literature does not settle it. Claus and Thomas (2001) estimated that the equity premium implied by prices and forecasts is about three percent, well below the historical average; an investor who accepts that can defend a hurdle below 10% for a company with no debt and $7.2 billion of cash. An investor who requires 10% needs the faster path, and has to believe in the top floors.
Owner cash is not a number Veeva reports. It is built from four that it does, and three of the four have a story.
| Twelve months to July 31, 2026 | $ million |
|---|---|
| Cash from operations | 1,665.5 |
| Less purchases of long-term assets | −26.7 |
| Less stock-based compensation | −494.6 |
| Less after-tax income on the cash balance (estimate) | −225.0 |
| Owner cash, trailing | 919.2 |
| The original case, rounded up | 950 |
| This year's guidance: $1,600 of operating cash, less 30, 525 and 225 | 820 |
Fiscal 2026 plus the first half of fiscal 2027, less the first half of fiscal 2026, so that the first quarter, when annual subscriptions are billed and collected, is counted once. The $225 million is trailing other income of $292.5 million after the first half's 23.3% tax rate, rounded; it is an estimate, not a reported figure.
The tax catch-up. The 2025 tax law known as the One Big Beautiful Bill Act restored the immediate deduction of domestic research spending and let companies accelerate the deduction of research costs they had capitalized in fiscal 2023 to 2025. Veeva spread that catch-up over two years, and fiscal 2027, which ends in January, is the second. The company says the law will keep reducing its cash taxes for the rest of the year and that it cannot estimate by how much. Its presentation puts the law's effect on fiscal 2026 operating cash at about $145 million compared with fiscal 2025. That is an aggregate for the law as a whole, which the filings do not split between the permanent change and the catch-up, and it is not a disclosed cliff for fiscal 2028. The deferred tax asset labeled capitalized expenditures fell by $80.5 million last year, to $246.0 million. That is a clue to the size of what remains, not the size: the line does not separate the research costs being caught up from anything else capitalized for tax. A finite benefit belongs in the years it occurs, not capitalized as if it lasted forever. If $100 million of this year's cash is the catch-up, the guidance case starts at $720 million, and the price offers 7.4%.
Stock pay. Veeva paid $494.6 million in stock over the last twelve months, 14.3% of revenue, down from 16.7% in fiscal 2024. The adjusted earnings the company guides to, about $9.21 a share this year, leave it out; on trailing figures the shares trade at 31 times adjusted earnings and 45 times reported earnings, and most of the difference is stock pay. Our model charges it as a cost, as an owner would. It counts the awards already outstanding in the 165 million shares and does not also add future grants to the share count, which would charge the same cost twice. In April Veeva moved its employees other than the chief executive from annual option grants to restricted units that vest after four years and are not expected to recur every year. At the end of July, $893 million of award cost was still to be expensed. Whether stock pay keeps falling as a share of revenue is one of the tests below.
This reminds me of Confucius, asked what he would do first if he were given a government. He would rectify the names, he said, for "if names be not correct, language is not in accordance with the truth of things" (Analects 13.3, in James Legge's translation). Stock pay is pay. The accounting rules have called it an expense for two decades. The adjusted figures still leave it out, and still call what remains earnings.
The literature agrees. Aboody, Barth and Kasznik (2004) found that investors treat stock pay as a real expense whether or not it is recognized, and Mohanram, White and Zhao (2020) found that companies paying more in stock trade at higher multiples and earn lower returns afterward. Core, Guay and Kothari (2002) showed that the usual way of counting options understates their dilution, which is why we also test 171 million shares, roughly the count if every outstanding option and unvested unit were added at $273.33. On that count the original case offers 8.5% instead of 8.7%.
The cash pile. Veeva holds $7.24 billion of cash and short-term investments and no funded debt. The model counts all of it, once, at face value, and subtracts the after-tax income it earns from owner cash, so that it is not counted twice. In the first half Veeva also spent $472.7 million buying back 2.66 million shares, at an average of $175.28; $1.4 billion of its $2 billion program remains. Buybacks spend cash. They do not make stock pay free. Acquisitions are left out of owner cash: $81.8 million in the half, most of it for Ostro, a conversational-AI service for drug brands. If growth needs more of them, owner cash is lower.
Likely, we can get some of it from primary research, which is what Vista is all about. The filings will report the results a quarter at a time. People who made these decisions, or carried them out, can tell us now how such decisions have usually gone.
We would askWhen the old CRM had to be replaced anyway, what decided the choice: the software, the cost and risk of moving, the data around it, or the price? Did the decision change what the company bought from the same vendor in research and quality?
The answer that would change the view"Once CRM was reopened, everything we bought from that vendor was reopened."
We would askAs a matter of past practice, what does each move cost a large company, how long do companies run the old and new systems side by side, and where do migrations stall?
The answer that would change the view"For a company whose data already lives on Salesforce, moving to Salesforce is the cheaper move."
We would askHow much of the manual review could software do without changing who signs off, and who would keep the saving: the company, its agencies or the vendor? Would you rather pay per review or per seat?
The answer that would change the view"The saving shows up as fewer reviewer seats, and nobody pays per review."
We would askWhen case intake was last automated, what share of the labor went away, what did validation cost each year, and who captured the saving?
The answer that would change the view"The outsourcers cut their prices to keep the work, and the software had to beat them."
Before any call we would write down the number the answer has to move, and how far, to matter: Year-1 owner cash against $1.215 billion at 15% growth, or growth against 18.8% a year from $950 million. If neither end of the plausible answers would change the reader's decision, the call has little decision value unless its purpose is to rule out a downside; that is the value of information. A credible favorable answer can matter as much as an unfavorable one. Interviews cannot tell us Veeva's undisclosed revenue, its customers' current plans or its contract terms, and we would not ask.
Every conversation would draw on past, general experience only: no current employees of Veeva or of its customers or competitors, nothing confidential, and every Advisor screened for conflicts before a word is said. For a client engagement, Vista would combine this note with those interviews: two or three Advisors, structured conversations, and one written brief that says where they agree, where they split, and what would change the answer.
We will score this note after Veeva reports its fourth quarter and fiscal year 2027 (the year ends January 31, 2027; last year's report came on March 4), and publish the result on the scorecard whether it flatters us or not. The tests are fixed today, so they cannot drift:
Four or five passes, and the cash the price needs starts to look earnable. Two or fewer, and the starting cash, the growth behind it, or both are weaker than the cases here assume. Three is too early to say. We will also report what the price then offers on each case, and read the first top-20 Falcon go-live, which management expects in the first half of 2027, when it comes.
A move is the one day a household learns what it really owns. Veeva's customers are packing now, and six of the largest are packing for a different address. The price has already assumed that the rest will like the new house, and pay more to live in it.
Owner cash each year = cash from operations − purchases of long-term assets − stock-based compensation − after-tax income on the cash balance. Year 1 runs from October 3, 2026 to October 2, 2027 and is paid at its end, with no growth before it; growth applies nine times, in Years 2 to 10; Year-10 cash then grows 3% a year forever. Cash and short-term investments of $7.242947 billion (July 31, 2026) are counted once, at face value; there is no funded debt, and lease costs stay inside operating cash and are not deducted again. Shares are 165 million, the company's fully diluted count; 171 million is a stress, roughly what counting every outstanding option (by the treasury-stock method, on the $181.57 average exercise price) and every unvested unit at $273.33 would give. Stock pay is charged as a cost, and future grants are not also added to the share count. The return each case offers is the discount rate at which its owner cash per share, plus cash per share, equals $273.33. The pinned terminal value deducts the investment that 3% growth requires at a 20% return on new capital. We read a revenue run rate as a quarter's revenue times four, which reproduces the company's statement that it reached a $3 billion run rate in early 2025 ($759.0 million in the quarter to April 2025); a $1.5 billion quarter in 2030 then requires 11.3% to 13.7% a year from the $928.0 million of the quarter to July 2026. Every figure in this note was recomputed in code from these inputs, and the model reproduces each of the original draft's figures to the cent before turning them around.
| Input | Value |
|---|---|
| Year-1 owner cash, guidance case | $820 million: $1,600 million of operating cash guidance less 30, 525 and 225 |
| Year-1 owner cash, original case | $950 million: the trailing $919.2 million, rounded up |
| Growth, Years 2 to 10 | 15% a year; or the fade: 15, 15, 15, 15, 12, 9, 6, 3 and 3% |
| Growth after Year 10 | 3% a year |
| Cash and short-term investments | $7.242947 billion, counted once |
| Shares | 165 million; stress 171 million |
| Reserve stress | $1 billion of cash held back |
| Return for the reverse solves | 10% a year, an illustration, not a recommended hurdle |
The trailing bridge uses fiscal 2026 cash from operations of $1,415.2 million, long-term asset purchases of $29.1 million and stock pay of $472.7 million, plus the first half of fiscal 2027 ($1,365.8, $9.8 and $256.1 million), less the first half of fiscal 2026 ($1,115.6, $12.2 and $234.2 million). The specialist comparison uses GAAP operating income over revenue: Veeva $429.3, $691.4 and $916.4 million on $2,363.7, $2,746.6 and $3,195.3 million; Salesforce $5,011, $7,205 and $8,331 million on $34,857, $37,895 and $41,525 million.
This note is research, not investment advice. It states what a market price assumes under labeled assumptions; it is not a recommendation to buy, sell or hold any security, and the decision belongs to the reader. As of October 3, 2026, Russ Rosenzweig, Vista's founder, owns no Veeva Systems shares. The note began as a research question Russ Rosenzweig put to the Vista Research Desk, which challenged his model over several rounds; the model and every figure were then rebuilt and checked against the company's filings.