Seven Q1 FY27 prints landed in the eight calendar days from Workday on May 21 to MongoDB on May 28. SaaS, data, identity, hardware, and silicon — five layers of the enterprise stack, seven income statements, one week. The result is the cleanest cross-layer evidence the substitution thesis has produced so far, and it does not point where the original hypothesis framing said it would. The vendors building the substitution layer are mostly capturing it; the market is pricing them on whether their AI revenue line is large enough and credible enough to look like genuine new value rather than a re-labelled extension of the seat business; and the dispersion of stock reactions inside the cohort is doing more analytical work this week than the headline numbers.

What We Covered

The 150th Feature (Tuesday) opened the week with Workday’s Q1 FY27 print, signed off by returning co-founder CEO Aneel Bhusri on his first day back. The headline: GAAP operating margin expanded from 1.8% to 13.3% year over year, agentic AI annualised revenue reached approximately $500 million, the customer count using Workday-built agents doubled quarter-over-quarter to over 4,000, and management raised the FY27 non-GAAP operating margin guide by 50 basis points to 30.5%. The same 10-Q characterised recent quarters as showing moderation of revenue growth from deal scrutiny, lengthened sales cycles in net new opportunities, and reduced growth in headcount-level commitments at renewals of existing customers, with federal-funding-tied verticals (government, higher education, healthcare) explicitly identified — H1’s central seat-compression mechanism in the regulatory record by the most-exposed seat-based incumbent in the category. Bhusri on the call: “The 150th feature in HR or finance is not going to move the needle for our business. The next agentic application will.” Verify weight 4 + complicate weight 3; net very slight verify lean. Strategic refinement: defend/self-fund now distinguishes between layoff-funded variants (Atlassian, Cloudflare, Cisco) and flat-headcount + AI-productivity variants (Workday), with the second variant now documented as working on the same direction as the first.

Three Prints, One Pattern (Thursday) confirmed the bifurcation at Salesforce and Snowflake on the same May 27 print day. Salesforce: Agentforce ARR $1.2 billion (+205% year over year, +50% sequentially from $800 million at the end of January), combined Agentforce + Data 360 ARR ~$3.4 billion, 3.8 billion Agentic Work Units (+111% quarter-over-quarter), 28.6 trillion tokens (+152% Q/Q), revenue $11.13 billion (+13%), GAAP EPS $2.42 (+52%), FY27 guide raised to $45.9–46.2 billion, $25 billion accelerated share repurchase, Benioff naming “Headless 360” as the strategy that expands the addressable market into surfaces the company has never previously monetised. Snowflake: product revenue $1.33 billion (+34%) — the strongest sequential dollar growth in the company’s history per CEO Sridhar Ramaswamy — 13,600+ AI accounts, Snowflake Intelligence accounts doubled Q/Q, Cortex Code in 7,100+ accounts, a new $6 billion AWS pact, and the company positioned explicitly as the “control plane for the Agentic Enterprise.” Cross-vendor synchronisation at this scale across three different vendor architectures — Workday flat-headcount, Salesforce massive ASR + Headless 360, Snowflake consumption + AWS — without offsetting deceleration in any of the three prints is a different magnitude of evidence than any single-vendor print. Complicate weight 4 + verify weight 2; net complicate lean. Strategic refinement: defend/workflow now reads “own the workflow AND the data, and price the agent traffic that traverses them” — the converging playbook of the canonical horizontal SaaS incumbents.

What the Market Added

The hardware and silicon legs of the same week argue against any reading that this is purely a SaaS-monetisation story. Dell reported Q1 FY27 revenue of $43.8 billion on May 28 — up 88% year over year, the fastest growth rate since the company returned to public markets in 2018. Adjusted EPS of $4.86 cleared the $2.91 consensus by 67%. AI Server revenue came in at $16.1 billion recognised against $24.4 billion in orders and a $51.3 billion backlog. The full-year revenue guide was raised by twenty percent to $165–169 billion from a prior $138–142 billion, and AI server revenue guidance for FY27 went from $50 billion to $60 billion in the same release. The data point that does not fit a clean AI-only story sits one line down on the income statement: Traditional Server and Networking revenue grew 92% year over year. The infrastructure refresh is broad. Marvell reported record Q1 FY27 revenue of $2.418 billion (+28% year over year) with non-GAAP gross margin of 58.9% — eighteen percentage points above the silicon-layer margin signature implied by Cisco’s print two weeks ago — and guided Q2 to $2.7 billion (+35% YoY), with full-year FY27 approaching $11 billion and custom silicon for hyperscalers running at $1.5 billion in FY26 with 20%+ growth ahead. The position of the vendor in the AI-capex stack continues to determine which gross-margin regime it operates in, and Marvell’s custom-silicon-into-hyperscaler position is producing nearly twenty points of margin above the networking-into-hyperscaler position Cisco occupies.

Okta added the cleanest pricing-side data point of the week. The company reported Q1 FY27 revenue and subscription revenue both growing 11% year over year, RPO +16%, free cash flow of $271 million. The numbers themselves are unremarkable in this cohort — the relevant data point is CEO Todd McKinnon’s framing on the call: “AI agents are rapidly becoming a new workforce inside every organization, creating a wave of identities that must be secured and governed alongside human users.” Okta is the canonical per-user-priced vendor in identity — its business model is literally seat licences for human authentication — and the CEO is now publicly positioning agent identities as net-additive demand rather than substitutive. The stock rallied 6.7% on the day. The same week, Evercore ISI downgraded CrowdStrike to in-line and flagged AI disruption risk extending to Zero Trust vendors, triggering CrowdStrike −7% and Palo Alto Networks −5%. The cybersecurity / identity sub-sector now contains a winner (Okta, with a credible agent-additive story) and two losers (CrowdStrike and Palo Alto Networks, where the same framing did not stick) inside one trading week.

MongoDB completes the picture from the other side of the data layer. The company beat both revenue and EPS lines on Thursday — EPS up 32% year over year, revenue up 25%, Q2 guide $729–734 million, FY27 guide $2.92–2.96 billion — and the stock disappointed because investors wanted faster AI acceleration. Same week, Snowflake’s +34% product revenue print was rewarded with the strongest sequential dollar growth in company history; MongoDB’s +25% beat was not. NoSQL document database and cloud data warehouse are different products with different AI workload exposure, and the market this week is selectively pricing the data layer based on whose Cortex-style or Snowflake-Intelligence-style usage curve is most credible at scale.

The Thread

Two weeks ago Borrowed Demand noted that the cloud-AI infrastructure moat was real but conditional on two cash-burning private labs honouring contracts disproportionate to their current revenue. The week before that, The Substitution Line framed the application layer as bifurcating cleanly between layers where AI replaces the work the application performs and layers where customer-specific context cannot be re-derived externally. This week’s seven prints update both readings. The customer-context fortress thesis is verifying at scale, with explicit financial signatures attached: Workday’s 1,154-basis-point operating margin expansion, Salesforce’s Agentforce ARR at $1.2 billion growing 205%, Snowflake’s strongest sequential dollar growth in history, and now the same pattern at Dell ($16.1 billion AI server revenue) and Marvell ($2.418 billion record on 58.9% non-GAAP gross margin). The incumbent-substitution-capture pattern is now documented inside the income statements of seven distinct vendors across five layers in one calendar week, with no offsetting deceleration evidence inside the cohort. The market is rewarding it selectively — Okta over CrowdStrike, Snowflake over MongoDB, Salesforce over a 32% YTD discount that the company is now buying back with $25 billion — which means the bifurcation predicted by H1 is not absent; it is operating at the vendor level inside each layer rather than between layers.

The piece of H1 that this week’s evidence makes hardest to defend is the version of the hypothesis where “software loses value” is read as “incumbent SaaS revenue compresses.” Workday’s 10-Q is the only piece of evidence in the cohort that talks about the seat-compression mechanism in explicit language, and Workday is the only vendor in the cohort whose pricing is literally headcount-tied. Every other vendor in the cohort is reporting that the substitution monetisation is augmenting, not replacing, the existing business — and the market is pricing those vendors on the credibility of their substitution story, not on whether the seat business is shrinking. The hypothesis text still says, accurately, that value expands where customer-specific context cannot be re-derived externally; what this week adds is that the incumbents are the ones holding that context, and the AI-native challenger thesis that would have produced value-transfer signals is, on this evidence, not landing in the public-market data. H1 sits at 53% verified after the week’s posts (live tracker) and the qualitative read is that the hypothesis’s mechanism is verifying but its predicted distribution of value is increasingly mis-specified — value loss is not where the framing assumed it would be.

Sources: Dell Q1 FY27 8-K (SEC); Dell Q1 FY27 earnings call transcript (Motley Fool); Dell +15% AI server rally context (Yahoo Finance); Marvell Q1 FY27 8-K (SEC); Marvell Q1 FY27 preview/post (24/7 Wall St.); Okta Q1 FY27 8-K (SEC); Okta Q1 FY27 earnings snapshot (MSN); TECHi — Okta AI-identity context; MongoDB Q1 FY27 (24/7 Wall St.).