The Salesforce Situation Is INSANE

Added: 2026-05-07

[00:00:00]Take a moment to look at this chart. On December 4th, 2024, this company closed at $368 a share. That was $354 billion in market cap. It was the stock market poster child for the last decade of enterprise software and the SAS playbook that built it. Now, have a look at where it is today. $183 a share as of Friday's close, down 50% from that December peak, down 27% just this year and down 32% over the past 12 months. That's roughly $180 billion in market value gone. And all the time, the stock is dropping 50%.

[00:00:39]The company has continued to show strong fundamentals. Since then, it has posted latest Q4 revenue up 12%, fullear revenue of 41.5 billion, up 10%, free cash flow of 14.4 billion. They've made some strategic acquisitions, including the closing of Informatica for 8 billion. They raised FY30 revenue target to 63 billion. They've got 35 billion of contracted future revenue locked in and growing at 16%. And in March, they executed the largest accelerated share buyback in corporate history. $25 billion in one transaction.

[00:01:15]$103 million shares retired in a single day. So what gives? Well, in this episode of the Liam Highland portfolio, we're going to figure out what Salesforce is actually worth right now.

[00:01:27]We're going to ask whether the market has correctly identified Salesforce as collateral damage from this AI transition or whether the fear has run ahead of the math. By the end, we'll have a bare base and bull case with specific share prices. We'll also have the four signals that settle this debate and we'll have the exact KPIs and dates we'll be tracking quarter by quarter to know which case we're in. Let's begin.

[00:01:52]So before we get to the valuation, let's look at what Salesforce actually does because the bare case turns on misreading this. If you ask the average person what Salesforce is, they'll say it's software that your sales team logs calls into. That's true on the surface, but there's a layer underneath that matters a lot more for the long-term value question. Salesforce is the system of record for customer data at roughly 90% of the Fortune 500. Their installed base touches over 150,000 organizations.

[00:02:21]Six product lines do the work.

[00:02:24]Salescloud is the historical core pipeline and sales automation. Service cloud is the customer service and support layer. Marketing cloud handles campaigns. Slack, which was bought for $27 billion in 2021, is the communication layer. They have Mulesoft, which is the integration plumbing, and Tableau handles the analytics. As of November 2025, Informatica now sits underneath all of that as the data management layer. So, I've made a simple diagram to illustrate and visualize where Salesforce operates in the wider enterprise data stack. This is my own simplified map. It's got six layers. It is not a universal architecture diagram.

[00:03:03]At the bottom, we've got data origination where business events first get captured. Things like sales activity, support cases, orders, invoices, product usage. Above that we have storage and query compute. Then we have context and semantics where technical data structure gets mapped into the actual business concepts things like our customers account the opportunity a case churn lifetime value.

[00:03:25]Above that we have the reasoning layer.

[00:03:27]This is where we've got open AI anthropic and Google and other frontier model providers. Then we have the action and orchestration layer where workflows execute. And then cutting across all of it is governance. We've got things like identity, permissions, lineage, and order trails. So, Salesforce doesn't sit horizontally in any one of these layers.

[00:03:46]In fact, it sits across a vertical slice that covers the customer relationship domain. At origination in businessto business, enterprise, salescloud, and service cloud are usually where the operational customer relationship is created and maintained. They've got things like the leads, accounts, contacts, opportunities, cases, service interactions, renewal history. Note how I said usually. Billing, ERP, e-commerce, product systems, and data warehouses all hold pieces of that truth, too. But Salesforce is where the customerf facing workflow lives. At the context layer, data 360 is their attempt to harmonize customer data and map it into Salesforce metadata framework and make it available to workflows, analytics and agents. At the action layer, service cloud, flow, slack, mulesoft and increasingly agent force are where customerf facing work actually gets executed. Now the bare argument is right about one thing. Salesforce, service now and workday are similar species of company. their do their main operating systems for the enterprise, right? They've got the same broad buyers, similar switching costs, and similar implementation pain, similar AI failure modes as well. Workday owns the people and finance domain. Service Now owns the enterprise service operations and Salesforce owns the customer relationship. But now, same species does not mean the same animal. AI will attack all three for sure. Where it attacks Salesforce hardest and where it gets blocked is the entire bear versus bull debate, right? Does AI turn CRM into a thinner seatbased interface or does it make Salesforce the governed execution layer for customerf facing agents?

[00:05:34]That's the moat question. Hold it because we will cover it when discussing signal 4 later. Okay. Okay, so now that we know where Salesforce sits in the wider enterprise data architecture and the domain it covers and how this compares to other software companies, we need to ask why is Salesforce down 50% from its all-time highs. Well, based on what we just covered, the bears see these companies as the same species, same category, and they conclude they are all the same animal. It's pricing Salesforce as a category, specifically as a SAS application company in a world where AI agents may do the work that used to require human seats clicking through the Salesforce customer relationship management screen. And founder and CEO Mark Beni off actually addressed this on the Q4 earnings call in February this year. His exact phrase was, you know, where are we going?

[00:06:25]Because you've heard about the SAS apocalypse >> and you know that this isn't our first SAS apocalypse. We've had a few of them.

[00:06:31]>> Look, for most people, AI is confusing.

[00:06:33]The media is confusing. What the hell is going on? Let me simplify this. I was just looking at our numbers. We've closed $2.7 million and we have 3.5 million more in the pipeline. Those are agents and it works. And so that is exciting. That is exciting that these agents can go out and sell for you. The idea that now our Salesforce instance can run autonomously versus doing manual data entry. I mean, this was always a dream. and I'm quoting, you know that this isn't our first SAS apocalypse.

[00:07:01]He's seen it before. In December of 2022, the stock dropped to $126 a share.

[00:07:07]While there were some company specific shocks, the narrative was near identical. SAS is dead. It can't grow.

[00:07:14]Activists started piling in. Yet, 18 months later, it was up 150% from the lows. Now, I'm not predicting the same setup repeats here, but the framing, this idea that the entire category is broken, that AI replaces seats, that recurring revenue dies is the same. And so, how does this narrative get priced into the stock? Well, Salesforce is a longduration software asset. A lot of its valuation depends on cash flows far into the future. That is why the operating numbers can look fine each quarter, but the share price will continue to decline. The component that's capturing this is called the terminal value, and most retail investors, unfortunately, gloss over it.

[00:07:59]Most institutional investors live and die by it. So, let's look into that.

[00:08:03]When a DCF model values a software company, it splits the value into two pieces. The first is the discrete forecast period, typically the next 5 to 10 years, where you can model revenue, margins, and cash flow with some confidence. The second piece is the terminal value, the present value of every cash flow from the end of the discrete forecast out to infinity, assuming the business persists. For a software company growing at high single digits with strong margins, terminal value typically represents 60 to 70% or more of the entire enterprise value. So when the market reprices Salesforce, it's rarely repricing the next four quarters. It's actually repricing the next 40. The question is whether Salesforce is still relevant in 2035, in 2040 when AI agents may do much of what currently requires a human in a CRM seat to do. If the market's answer shifts to greater uncertainty, the terminal value collapses and 60 to 70% of the value comes down with it. And that's what takes the share price down 50% from all-time highs even though the near-term results are fine. To see this visually in the valuation, let's have a look at the base case in my model. We'll start with the equity value diagram. The model splits each case into the discrete forecast, the terminal value, and the net debt adjustment. In the base case, terminal value alone is about 163 billion of the 199 billion total equity value. That's 82% of the value sitting in the longtail. In the bare case, terminal value compresses to $86 billion. The discrete forecast, the part we can actually model with greater certainty, barely changes between the base and bare cases. It's the longtail value that does all the work. So, the four signals I'm about to walk through next are the best tests we have to determine whether the terminal value collapse the market is implying is actually justified. If they hold, the market panicked and the math is wrong.

[00:10:05]However, if they break, the repricing is rational and you don't want to be a buyer at $183 a share. So, these four signals we're going to be covering are one, is demand actually holding up? Two, is AI actually monetizing? Three, is the cash flow real? and four, is the moat holding? Let's take them one at a time.

[00:10:25]Signal one, is demand actually holding up or are seats quietly compressing under the AI narrative. The single best near-term indicator is current remaining performance obligations or CRPO. This is the dollar value of contracts customers have already signed, but where the revenue hasn't yet been recognized. It includes the next 12 months of subscription billings only. Think of it as visibility into next year's revenue locked in and banked. And so why would CRPO be more useful than headline revenue? Well, it's because headline revenue can be inflated by things like acquisitions. Informatica added 4 percentage points to Q4 CRPO and roughly $400 million to Q4 revenue. CRPO has the same problem to a degree, but the contractual nature makes it harder to dress up. If sales teams are losing momentum, CRPO is where you see it first before pipeline weakness shows up in next quarter's revenue. And so what were Salesforce's latest results? Well, we had Q4 FY26 CRPO of $35.1 billion, up 16% year-over-year nominally, 13% in constant currency. Strip out the four points from Informatica and you get organic constant currency CRPO growth of 9%. Total RPO at 72.4 4 billion up 14% and management guided Q1 FY27 CRPO at approximately 14% nominal and 13% constant currency. So they're explicitly telling investors the deceleration from 16 to 14% is expected and largely lapping. That's not particularly aggressive guidance and it's actually showing some grounding from management.

[00:12:05]And so to track which case we're in the CRPO band below 12% growth the bare case starts confirming between 12 and 16% the base case and above 16% we have bull confirmation. The latest print sits inside the bull zone at exactly 16%. To add to this the CFO Robin Washington actually explicitly called out NNAOV acceleration in the second half of FY26 on the Q4 call. NNAOV, net new annual order value, is the leading edge of CRPO. It's the deal flow indicator before it shows up in contracted revenue. So, we have two confirmation sources for this signal. The actual CRPO print and the management commentary on whether NNAOV is still accelerating. And so, when we get that result on May 27th, 2026, that closes a meaningful chunk of this argument. And our signal 2 is the one with the most narrative noise and the cleanest disclosures. It's asking, is Asian Force actually monetizing or is it a feature dressed up as a franchise?

[00:13:07]Here's what we know. Agents Force annual recurring revenue or AIRR in Q4 of FY26 was $800 million, up 169% year-over-year. They closed 29,000 deals since launch with deal count up 50% quarterover.

[00:13:23]combined ancient Force and Data 360 ARR was $2.9 billion up over 200%.

[00:13:30]But that headline number includes 1.1 billion from Informatica's contribution.

[00:13:35]So the organic combined agent force plus data 360 is closer to $1.8 billion. Now in this there is an overlooked data point. In the quarter, more than 60% of Agent Force and Data 360 bookings came from existing customers expanding their commitments. 60% of Q4 Agent Force and Data 360 bookings came from existing customers expanding their spend. That matters because it tells us something about the monetization motion. The bare case says AI add-ons defend renewal but don't generate incremental revenue. The data says existing customers are putting incremental dollars down for these AI products on top of their existing subscriptions. That's the definition of additive monetization. It's not defensive. And while it's still early and the number is small, it is a signal in the right direction. Now, I have to be honest about the bare side, too. $800 million of Agent Force ARR against $41.5 billion in total revenue is roughly 2%.

[00:14:39]The growth rate is real. The absolute scale is not yet enough to bend the company level growth rate. For Asian Force AR to materially affect the consolidated growth rate, it needs to scale to several billion. Salesforce's FY30 target of $63 billion implies an 11% compound annual growth rate from FY26. To hit that without seat compression in the legacy clouds, Agent Force ARR has to scale to roughly four to5 billion dollars by FY30. That's a five-fold increase over 4 years from a starting base of 800 million. It's aggressive, but not impossible. The 169% Q4 growth rate gives us that trajectory.

[00:15:24]And then another data point worth tracking that Salesforce reports is a metric they call AWUS, agentic work units, which is their measure of real work an agent did. Cumulative AWS hit 2.4 billion by end of Q4, growing 50% quarterover-arter. They've consumed 19 trillion tokens to date, a 5x year. So you have evidence that agents are being deployed and actually used, not just sold and shelved. In terms of where this agent force ARR sits in each of the cases, we have 0 to900 million in the bare range, 900 to 1.5 billion for base, and 1.5 billion to 2.5 billion for bull.

[00:16:08]The latest is $800 million. So it's sitting in the bare zone. It's $100 million below the base threshold. That sounds bad on first reading until you look at the trajectory. Q3 FY26 ARR was 554 million. Q4 was 800 million. That's 260 million of net new ARR in a single quarter. 48% sequential growth. If the next print on May 27 holds even half that pace, Salesforce crosses into the base zone for the first time. If it holds at the Q3 to Q4 pace, the bull zone comes into view by Q2 or Q3 of FY27. One bare concern that's earning real attention is that Agent Force uses a flex credit consumption pricing model.

[00:16:56]That's good for Salesforce because revenue scales with usage. It's also a reason to watch gross margin carefully.

[00:17:04]If inference costs eat into agent force contribution margin as ARR continues to scale, the headline ARR growth might not flow through into the actual operating profit. Right? So Q4 FY26 GAP operating margin was 20.1%.

[00:17:22]Non-GAAP was 34.1%.

[00:17:25]It is stable for now, but this is a key line to watch for FY27.

[00:17:30]Signal three is the question that gets the least amount of airtime but has the biggest valuation impact. Is the reported free cash flow real or is it being flattered by an accounting tailwind that doesn't actually compound?

[00:17:45]This is in my view the most credible critique of Salesforce at this point in time. Let me walk through it carefully.

[00:17:52]FY26 reported operating cash flow was 15 billion up 15% year-over-year. Free cash flow was 14.4 4 billion up 16%. On the surface, that's a 35% free cash flow margin on $41.5 billion of revenue. By peer group standards, it's elite. Ford EV to free cash flow using current market cap of $174 billion. A net cash adjusted enterprise value works out to roughly 11 times Ford free cash flow.

[00:18:20]That sounds very cheap for a software business growing at 10% given that the peer median is closer to 17. But, and this is the catch the bears keep flagging, FY26 operating cash flow benefited from a $2.9 billion increase in unearned revenue. Unearned revenue is cash that's being collected from customers upfront, but hasn't been recognized as revenue yet. It hits operating cash flow in the year you collect it. It doesn't necessarily reflect the durable earnings power of the business. That $2.9 billion benefit sits inside the operating cash flow line and inflates it relative to what the underlying earnings actually look like.

[00:19:00]If we had to normalize that out, we would land closer to 11.5 billion of owner free cash flow. On that basis, the EV to normalize free cash flow is closer to 14 times. Still cheap relative to peers, but not historically anomalous.

[00:19:14]The market is partly right to be skeptical here. And so under the three cases, the free cash flow margin runs at around 33% in the base, 27% in bare, and 35 12% in the bull. The bare case isn't assuming margins collapse. It assumes the market reprices salesforce towards the normalized cash quality, not the headline cash quality. And that is a defensible bare input. And it's also the input that does most of the work in producing the bare case downside. Now, what does management do with that cash?

[00:19:46]This is where the picture flips back to the bulls. In March of 2026, Salesforce executed the largest accelerated share repurchase or ASR in corporate history, $25 billion. They funded it with 25 billion of senior notes maturing between 2028 and 2066.

[00:20:08]103 million shares were delivered upfront on March 16th. That's roughly 11% of the total share count retired at an average price that is meaningfully below where I think the fair value sits.

[00:20:20]And we'll get to that fair value in the valuation section. And so what does that mean? Well, the signal here is worth sitting on. You see, a management team that thinks the long-term economics of the business have permanently been impaired does not go out and borrow $25 billion at 40-year maturities to retire equity. They preserve flexibility. The fact that this was funded with debt, meaning Salesforce now has a permanent capital structure decision baked into the balance sheet, tells you that Beni off and Robin Washington believe the cash flows are durable enough to service that debt for decades. You can disagree with them, but this is the strongest insider signal you can get short of a personal share purchase. They put the company's balance sheet behind their conviction. Signal 4 looks at whether the moat is holding or if the workflow position is eroding from above and below at the same time. You see there are two distinct angles of attack against Salesforce here. From above we have hyperscalers and data warehouses.

[00:21:20]Snowflake datab bricks they are trying to own the customer data layer directly.

[00:21:26]The argument if customer data sits in Snowflake instead of Salesforce the workflow value Salesforce currently captures shrinks Salesforce becomes a workflow display layer on top of someone else's data. Thin client lower terminal multiple and then from below you have offerings like Microsoft Copilot. In February of 2026 Microsoft disclosed they had 15 million paid co-pilot seats up 160% year-over-year. Microsoft strategy is to bundle AI assistants into the Office 365 procurement process where they already have the relationship. For shallow CRM tasks, drafting follow-up emails, summarizing calls, generating notes, Copilot can do a substantial fraction of what Salesforce seat does.

[00:22:08]If you already pay $30 per user per month for Office, paying another $30 for Copilot is much easier than buying additional SalesCloud seats at $165 per user per month, especially if you're a 50 person sales team that doesn't need the deep enterprise feature set. So, how is Salesforce defending against both attacks? Well, first, data 360 with a zero copy architecture. Salesforce's pitch to the enterprise is that you don't have to move your customer data to Salesforce to use its workflows. They federate it from wherever it lives.

[00:22:43]Snowflake, data bricks, your data warehouse. In FY26, data 360 ingested 112 trillion records, up 114% year-over-year. 53 trillion of those came through zero copy, up 310% year-over-year. The strategic logic is that if Salesforce makes itself the federation layer for customer context, they don't lose when the warehouse becomes the system of record. They become the customer context layer that sits on top. And just to add some of my personal experience here, the prospect of avoiding a lift and shift scenario between systems is a big deal in large enterprise because it's a pain in the you know what when you have to do it.

[00:23:22]The second defense is workflow depth.

[00:23:25]The bullcase here is straightforward.

[00:23:27]Microsoft Copilot does shallow productivity work well. The deep enterprise work like running a sales territory with 200 reps, managing a multi-stage support escalation that touches billing and contracts and renewals, integrating quote to cash flows that route through Mulesoft and SAP and your ERP that doesn't move to copilot easily. And the more agents you deploy in those deep workflows, the more Salesforce role as the govern execution environment becomes valuable. the moat shifts from human seat UI to a genic workflow context. And so how would we test this? Well, the first signal would be net revenue retention. Now, Salesforce doesn't disclose NRR explicitly, but you can triangulate it from the CRPO build, the absence of a significant churn commentary, and that 60% of existing customer expansion data point from Q4. Those signals are all consistent with retention holding through this AI panic cycle. And then the second signal are the product line growth disclosures under the new FY27 reporting structure. Starting Q1 FY27 in May, Salesforce now splits revenue by salescloud, service cloud. They've got the marketing and commerce, the platform which is their data 360 and industries.

[00:24:38]This is the new disclosure framework and Robin Washington introduced it to give investors clearer visibility into where AI revenue lands. We'll see the first print under that structure on May 27th.

[00:24:50]And I'm getting ahead of myself here, but if we quickly drill down into the cost of capital or the terminal growth element of the valuation, we have things like the risk matrix, the mode scorecard, and the competitive positioning that feed into the valuation assumptions. For the mode scorecard, we have switching costs scored at four out of five with stable trend. Customer lockin is scored at four out of five and stable. AI product differentiation scored at three out of five, which is still unproven. Threat from hyperscalers scored at 2 out of five and elevated.

[00:25:18]and pricing power scored at three out of five and slightly weakening. So ultimately the analysis is concluding that we are seeing stability to slight strengthening at the workflow layer but we are seeing that gradual weakening at the thin seat UI layer. And that is my view too. It matters because it tells us which part of the business to underwrite. For those that are long Salesforce it takes the view that the deep workflow moat compounds faster than the entry seat moat erodess. The market is treating the entire thing as if it's Microsoft attackable. That is just a difference in perception. The market is likely wrong on the depth even if it is right on the breadth. But we can test this signal going forward. If salescloud or service cloud growth specifically drops below 6% on the new FY27 disclosure, the bare case on seat erosion is confirmed. If they hold high single digits while data 360 continues to outgrow the core, we're in the base.

[00:26:14]And if they accelerate as agent force monetizes through existing accounts, we're in bull. So now that we've covered the four signals driving the wedge between the bears and the bulls on Salesforce, let's pull this together from a valuation perspective. Currently, the stock is at $183 a share, market cap at $174 billion. After the ASR completes, share count drops from 957 million to 855 million. Net debt takes off $30 billion post ASR funding and enterprise value is around 204 billion.

[00:26:45]FY26 reported revenue is 41.5 billion.

[00:26:48]Reported free cash flow of 14.4 billion.

[00:26:50]Normalized free cash flow closer to 11.5 billion adjusting for the unearned revenue tailwind we discussed. So now let's look at the three cases. Each one is a calibrated DCF with a fade down approach plus a cross check multiple.

[00:27:04]Let me walk you through each. For the bare case, revenue growth fades from 5% in year 1 towards sub3% terminal.

[00:27:11]Operating margin holds at 30%, no expansion through the cycle. Free cash flow margin compresses from headline 35% towards 27% as the unearned revenue tailwind unwinds and inference costs start to hit. the weighted average cost of capital of 11% with a beta of 1.4 reflecting the higher disruption sensitivity plus a 0.75% company specific risk premium for the AI uncertainty. Terminal value growth at 1.5%. Equity value comes out to $19 billion per share. You land at a mid-range of $127 a share with a low of 104 and high of 155. Either way, you're looking at a downside of roughly 30% to the mid-priced based on the current spot. In the base case, we have revenue growth fading from 9.5% year 1 towards a 4% terminal anchored to the FY27's guidance point of 46 billion and the FY30 target of 63 billion of organic revenue. Operating margin expands modestly from 20.1% gap towards 25% terminal as scale offsets continued AI investment. Free cash flow margin holds around 33%. cost of capital at 9.8%, risk-free rate of 4.4% and a beta of 1.29. Equity risk premium of 4.23% and no company specific premium because the higher beta already captures it. With terminal growth at 2.3%, equity value arrives at $199 billion, giving a share price at the mid of $237 with a range of 201 on the low and 272 on the high, giving an upside of around 30% to the mid. And then finally in the bull case, revenue growth fades from 12% towards 5% terminal. Agent force monetization flows through the upside, putting consolidation growth above the FY30 target. Operating margin reaches 26% terminal. Free cash flow reaches 35.5.

[00:29:02]Your cost of capital at 9.3%, beta drops to 1.18 as the terminal value durability narrative rerates. Terminal growth to 2.8% 8% and we get an equity value of $266 billion. Per share at the mid, you arrive at $327 with a range of 279 on the low and 382 on the high, giving an upside of potentially 78% from the mid.

[00:29:26]Now, these prices aren't single points.

[00:29:29]The cost of capital and terminal growth do most of the work. The bare case, for example, run across a reasonable range sweeps from $105 at 12% cost of capital and 1% terminal growth up to $155 at 10% cost of capital and 2% terminal growth.

[00:29:46]Anchor in the middle, you have 11% cost of capital and 1.5% terminal growth, landing at $126 a share. So at $183, we have 54 of theoretical downside and about $144 of upside. Now, I'll repeat something I've said in my previous episodes. I do not use probability weighted expected returns as a decision tool. I think assigning a 20% to a bear and 30% to a bull is a thumb sucking exercise dressed up as a number. Nobody knows those probabilities, but the calculation is useful. So, if we were to probability weight the cases at 20, 50, and 30, we'd get $224 per share or a 22% premium to the current spot. But that margin of safety all depends on what is happening to that terminal value and underlying assumptions. And this brings us to what I believe is arguably more important than doing these probability weighted expected returns. And that is tracking which case is unfolding. We can look at some of the KPIs each quarter.

[00:30:45]For example, CRPO growth for the bare we have under 12%, base 12 to 16%, bull at 16 plus%. The latest is 16% nominal and 13% constant currency. We have a Q1 FY27 guide of 14% nominal and what would trigger the bare case is if that CRPO fell below 12% for two consecutive quarters. Another KPI would be the agent force ARR. For the bare case, we'd sit under the $900 million base 900 to 1.5 billion and bull 1.5 to 2.5 billion.

[00:31:16]With the latest reported 800 million, we are still in that bare zone, but the trajectory is showing that we will reach the base. Another KPI would be non-GAAP operating margin. For the bear, we'd sit at 30 to 34%, base at 34 to 35%, and bull would be 35% plus. The latest is at 34.3%. The thing to watch is whether Salesforce can move into the high30s without sacrificing growth. And that's the rule of 50 target that they've set.

[00:31:42]And then another KPI would be the salescloud and service cloud growth, the new seat durability test, which we can now track under the new FY27 disclosures. below 6% the bare case on seat compression is confirmed. High single digits we're in base and if it's accelerates we are in bull. But those aren't the only things to watch. We also have some of the leading indicators which we've modeled in here. For example, we've got things like sales and service cloud growth, platform and data 360 growth, gross margin trajectory, R&D intensity, AI revenue disclosure cadence, share count, public customer references, agent force related deal references. Each card has a bull outcome, a bare outcome, and the valuation input it affects. And so now that we know what to watch, here's when we get the answers. First is May 27th, 2026 Q1 FY27 earnings. First checkpoint.

[00:32:33]CRPO and Agent Force ARR are the two numbers that matter most. If CRPO clears 14% and Agent Force AR clears 1 billion, the base case is intact and the bull case has a path. If CRPO drops below 13% or Agent Force ARR fails to scale past $900 million, the bare case starts confirming. The next checkpoint is August of 2026 for Q2 earnings. There we have two consecutive quarters of CRPO data. The trend either confirms or breaks. SalesCloud and ServiceCloud growth visible under the new disclosure structure for the second time, removing the compbased explanation if either is decelerating. Checkpoint.3 is November 2026 where we have Q3 earnings. By this point, FY27 is mostly visible. The framework for FY28 is getting set up.

[00:33:23]Investor day commentary on agent force monetization pacing should also be available. And then checkpoint 4 would be Feb 2027 where we get Q4 full year 27 results plus the FY28 guidance. The full thesis test. If FY28 guide comes in at 11% plus revenue growth with operating margin expanding to 20 plus% we have base to bull confirmation below 9% growth or margins are flat. Base case scenario needs reassessment. Now let me just zoom out to the question that ultimately decides the terminal value debate. The cost of intelligence has dropped 99% in 2 and 1/2 years. Tokens are commoditizing. Foundation models are commoditizing. When a resource commoditizes, economic value migrates to whatever stays scarce and complimentary.

[00:34:13]In this world, the bears say that customer data can move to snowflake.

[00:34:17]Workflows can be replicated by AI agents. The seatbased UI is going to compress. Salesforce is left with a slowly eroding application franchise and a defensive AI product that protects renewal but does not grow. While the bulls will argue that the customer relationship is the scarce asset, not the data underneath it. The workflows that touch revenue, support, billing, contract, renewal, those carry compliance, audit, identity, integration, depth that does not migrate to Snowflake. They migrate to wherever the workflow lives. Salesforce has spent 25 years embedding those workflows in 90% of the Fortune 500. The more agents you deploy, the more important the governed execution layer becomes. In this world, agents do the work.

[00:35:10]Salesforce keeps the rent. And now, based on my experience working in large enterprises, I likely lean towards the bull view, but I hold it with appropriate uncertainty. What is not real in my view is the proposition that all of this collapses the workflow layer where salescloud and service cloud actually run. That's the part where 25 years of enterprise integration and embedded processes compound. And you can disagree, but at least we have a framework like this that allows us to test and monitor going forward. Now, full disclosure, at the time of recording, I do not hold a position in Salesforce. For me, it comes down to the terminal value risk that I can't resolve with the data we have today. And the terminal value question reduces to one binary. Do agents commoditize the workflow layer over a 10-year horizon, or do they make a governed execution environment more valuable than the human seat era? The bullcase asks us to believe that agents need governance more than humans did. I think that's plausible, but it's not yet proven. The Microsoft co-pilot threat at the entry seat layer is real. The Snowflake and data bricks pull on data control is real. both compress where Salesforce currently captures rent over a decade, even if the next four quarters are fine.

[00:36:29]And so those are my thoughts on the Salesforce situation. This is not financial advice. Always do your own research. But I really enjoyed being able to present this analysis. If you found this kind of breakdown useful, the like button really helps the channel and consider subscribing because I'm really keen to see if there's demand for this kind of analysis. I'm currently working on more episodes covering these SAS and data companies as we all watch this AI experiment unfold in real time. I'll see you in the next one.