We’re convinced the biggest beneficiaries of AI will be enterprise software companies. We’ve come to this conclusion in part from reviewing the earnings calls of 50+ publicly traded SaaS companies like ServiceNow, Netskope, Salesforce, etc. What follows is a summary of AI’s impact on these SaaS businesses, their customers, and their competitive positioning.
We believe this is the most comprehensive analysis of Q1 2026 enterprise software earnings calls published this cycle. Our conclusion is that enterprise software is going to be the biggest beneficiary of the move to agentic AI, by a mile. Big take-aways are below:
SaaS companies are performing nicely. ServiceNow beat expectations and raised guidance. DigitalOcean beat every financial target from the prior call. Zeta had it’s 19th consecutive beat and raise quarter, and closed one of their largest deals on record. Freshworks beat for a 6th consecutive quarter and signed the two largest deals in its history. Procore raised 2026 guidance despite what they described as a “challenging construction market”. Amplitude’s NDR rose to 106% and multiproduct accounts now represent 77% of their ARR, up from 64% a year ago. Datadog’s revenue was up 32% year-over-year, accelerating from 25% in the year-ago quarter; this was also Datadog’s first-ever $1 billion revenue quarter. Cloudflare is tracking north of 46% on a Rule of 40 basis, with visibility to reaching north of 50% the following year. Klaviyo closed its largest number of multi-million dollar ARR deals ever, including one expansion that brought a single customer’s contract to more than $6 million in ARR. Figma’s revenue was $333 million, up 46% year-over-year and accelerating from 40% last quarter and 38% in Q3. Their NDR reached 139%, the highest in over two years, up three percentage points sequentially. JFrog continues to enjoy net dollar retention of 120% and gross retention of 97%. Backblaze’s NRR improving to 110% from 105% and they raised guidance. Fastly’s Q1 2026 call was all records. Dynatrace’s net new ARR grew double-digits for the first time in three years. Braze had it’s 4th straight quarter of growth acceleration which is now 30% YOY, and raised full year guidance. Zscaler had record profitability while growing 25%. SentinelOne had record net new ARR growth. UIPath had revenue acceleration from 14% to 17% thanks to their agentic product launch. MongoDB’s revenue accelerated to 25% YOY, up from 22%, and they raised guidance. Samsara revenue is growing 31% with nearly $2 billion in ARR and profitable, with an acceleration in large customer ARR. ServiceTitan raised guidance. Docusign customers with over $300,000 in ACV grew 12% year-over-year, the first double-digit growth for this cohort in three years. Dollar net retention improved for seven consecutive quarters to over 102%. Rubrik’s full-year guidance was raised across subscription ARR, revenue, margins, and free cash flow. Sailpoint guidance was raised across ARR, revenue, margin, and free cash flow. Asana raised full year guidance. Weave has now meet or exceeded guidance for 17 straight quarters. Sprinklr beat guidance. Netskope’s new logo growth was 60%, the strongest in company history.
SaaS is the infrastructure that underlies AI. A single source of truth, cross-cloud interoperability, and enterprise-grade governance are the three prerequisites that enterprises must have in place before AI agents can operate reliably across their data estate. This is something only enterprise software can provide. For instance according to the ServiceNow CEO, “We are the rules and the rails of business. Don’t fall for the parlor trick that one touch button can replace 22 years of excellence. There is no great control tower, there is no great workflow data fabric, and there is no great integration layer that ties into the hyperscalers, the language models, the systems of record without” ServiceNow. Similarly Cloudflare is processing hundreds of billions of agentic requests per month, so every AI agent calling a tool, querying an API, or taking an action on the internet is generating a request that may pass through Cloudflare’s network.
SaaS companies have agentic product, now. ServiceNow thinks their AI product does $1.5bln in revenue in 2026. DigitalOcean’s AI customer ARR grew 221% to $170 million. Additionally, the AI product is embedded in the software, not just a bolt on. Customers have become smart enough to know the difference. “AI is not causing sales delays — it has become a core requirement in nearly all large enterprise deal discussions. Customers are not asking whether we have AI. They are asking which AI, how it works, and what outcomes it delivers. Freshworks is winning because Freddy AI is embedded in the platform, “not bolted on.” 60% of Figma’s largest customers used their AI product on a weekly basis in Q1, and over 80% of users continued using the Figma Design AI product. PubMatic’s AgenticOS launched months ago. Doximity’s 140 health systems have all purchased Doximity’s Clinical AI Suite. More than 500 customers are deploying Dynatrace’s agentic capabilities. A customer at Braze saw unsubscribes fall 81% and first email opt-outs drop 97% after deploying Braze AI Operator. Another saw app opens rise 284% for specific AI-optimized campaigns. 9100 of Snowflake’s customers are now using their AI offering. Their Cortex Code offering compresses data migration timelines that once required 18 months and 100-person teams into weeks, with one partner reporting 21,000 operations completed in 20 days. SentinelOne’s ARR from non-endpoint products approached 50% for the first time, reflecting growth in AI security, data, and cloud offerings. AI security ARR “nearly doubled again” in Q1. Salesforce’s Agentforce ARR surpassed $1 billion, up 205% year-over-year. ServiceTitan’s customers are seeing a 6-point booking rate improvement, 9-point close rate improvement, 30% higher average ticket, and 50% higher revenue per technician with their agentic offering. Sailpoint is live with Agentic Fabric. Asana expects AI bookings to contribute 15% of net new ARR this year. Over 50% of customers are currently using Weave’s AI tools now.
AI revenue is still small relative to total revenue. For instance even though Zeta’s new AI release resulted in a 7x increase in agentic interactions, it will be a “minimal” contributor to forecast, for now. Figma launched AI credit monetization only on March 18. Similarly Doximity’s AI suite is not generating enough revenue to push overall growth of the entire business beyond 5% due to large and slow growing legacy revenue. Agentforce while now running at $1bln of ARR is small for Salesforce overall, whose ARR is more like ~$40bln.
Pricing models are shifting, but only because AI workloads are so expensive. Companies like ServiceNow are pricing their agentic product offerings based on consumption, not seats. Procore offers seat based on volumetric pricing based on construction budgets; the latter helps insulate Procore from shrinking seat counts. Zoominfo is moving to a consumption model, but that appears largely due to issues showing the value of their product in a seat based model. Zscaler’s consumption offering has 100% YOYG because AI agents generate machine-to-machine traffic at volumes that dwarf human user sessions, making seat-based pricing disconnected from the value Zscaler provides; metered usage aligns pricing to actual traffic and policy enforcement volume.
That said, consumption based models are resulting in what Snowflake called “sticker shock,” which is a reference to enterprises discovering that AI workloads consume significantly more compute credits than traditional analytics. Consumption based models are also needed to offset the margin pressure from AI workloads: Gitlab gave gross margin guidance of 85–87%, down from 89%, due to AI workloads of customers.
AI is opening new markets for SaaS, expanding the TAM. For instance, when Procore’s AI compresses weeks of manual bidding work into 20 minutes, it is competing for the salary budget of the estimating team. It’s no longer just software. Agents themselves will also become customers; Amplitude launched Agent Analytics specifically for monitoring, evaluating, and optimizing AI agents’ interactions and outcomes. The more AI workloads customers run on Datadog, the more of their AI tools they need to manage them; 56% of their customers now use four or more products (up from 51% a year ago). Waystar thinks it’s TAM is 5x larger with AI, especially as it sees consolidation opportunities. Slop Stopper is DoubleVerify’s AI-powered tool that detects and blocks advertiser placements alongside low-quality AI-generated content. Jfrog’s opportunity is compelling: as AI coding tools dramatically increase the rate at which developers pull open-source dependencies, because AI-generated code tends to rely heavily on external libraries, the attack surface grows proportionally and the value of JFrog’s offering grows with it. Agents magnify the need for Zscaler; according to their CEO “today, users are the weakest link in cybersecurity. But soon, AI agents will be the weakest link. Because they operate at far greater speed and have far less oversight.” Rubrik’s pipeline grew 50% in one quarter thanks to Mythos. Sailpoint says for every human identity an enterprise governs, there can be 100 non-human identities or more.
AI is being integrated into the software offerings. According to ServiceNow’s CEO, “AI, data, security, and governance are now built into every product and package — not a separate purchase. This is a deliberate break from sidecar AI. We’re not bolting intelligence onto disconnected systems. We’re combining context with execution on a single platform.”
Enterprises cannot depend on probabilistic models. Generic AI produces probabilistic outputs without grounding in real enterprise data, real workflows, and real ontologies. Software is the grounding and even then, AI is not to be depended on given it is probabilistic, whereas enterprise customers require deterministic outputs. Alex Karp of Palantir refers to the underlying grounding as “ontology”. Dynatrace references that their offering allows AI agents to “act with confidence to deliver what we refer to as answers, not guesses.” Doximity has been slow to roll out an AI offering because guesses cannot be tolerated in a healthcare setting. Sailpoint’s CEO put it well: “You cannot control what you cannot govern.”
Without software, AI is slop. The CEO of Palantir, Alex Karp, put it well: generic AI without an ontology produces slop. Note we had to look up what “ontology” means and in this case, it’s a structured map of everything that exists and matters inside an enterprise. Every object, every relationship, every rule, and every action, encoded in a way that AI can reason about it with precision. In Karp’s view, only software provides that ontology. Procore makes a similar argument: Procore AI is purpose-built for construction that understands the language and logic of the project — how an RFI connects to a submittal, how a submittal connects to a drawing, how a change order gets approved — operating as a multi-step system that holds context across an entire project lifecycle rather than answering individual questions. A generic LLM doesn’t understand the contractual significance of submittal cross-checking, and cannot reason about the downstream consequences of a change order in the context of a specific project’s schedule and budget. Procore’s reasoning engine can, because it’s trained on data from nearly 3 million active users across hundreds of thousands of real construction projects.
AI needs to deliver ROI, now. Zeta’s CEO said it well: customers no longer want to see “roadmap AI,” rather they want “applied intelligence that works today.” The investment in AI has been significant, and we’re now in the stage of delivery as opposed to promises.
AI requires a clean rollout and real onboarding. You need the infrastructure to roll out AI to the customer successfully. For instance, Zeta’s AI offering was made available to 100% of enterprise clients simultaneously with launch and a dedicated learning and development group has already onboarded the top 30 clients with weekly virtual training, recorded modules, and a planned certification program for client users. It requires real onboarding.
Software companies are putting out more product at a greater velocity. Over half of Freshworks’s code is now originated by AI, enabling shorter development cycles. They’re cutting headcount while raising guidance. Amplitude’s CEO said “over 90% of the code our team ships today is written by AI.” 97% of Cloudflare’s R&D employees use AI coding tools with productivity gains of “two, ten, even 100 times”. Developer productivity at Health Catalyst is up 100%. Ziprecruiter reported they’re primarily using AI to accelerate their product roadmap, not cut costs. ServiceTitan is using AI to “build a software factory where AI agents play a central role in all code development to accelerate velocity.” Docusign says it’s building more product faster than ever.
SMB focused software is very vulnerable. SMB’s are the segment of the market most likely to create homemade software using AI. They have much simpler workflows and low/no corporate requirements. AI becomes an easy substitute. Freshworks is intentionally managing down its SMB portfolio for profitability instead of growth. Sprout Social continued to report deceleration in the sub $30k ACV segment. Dropbox grew only 2% year over year, although paying users grew sequentially for the first time in several quarters. Zoominfo’s guidance was a 4% decline in revenue for the year. That said, there was one SMB brightspot: Weave’s revenue continues to grow for the SMB dental and healthcare practices they serve.
Gross margins are under pressure. Amplitude’s drop in gross margin was due to rising inference costs. Sprinklr reported gross margin pressure this quarter from AI workloads.
The data moats are very real and irreplicable. Health Catalyst’s AI is going to be pulling from their repository of 18 years of data, which is irreplicable. Klaviyo’s infrastructure processes almost 4 billion daily events and signals across 8 billion consumer profiles. Samsara’s AI is based on 25 trillion data points captured annually from physical assets across tens of thousands of customer deployments. That dataset is accumulated from real sensors embedded in real trucks, real construction equipment, real manufacturing facilities, and real utility vehicles operating in real physical environments. That data advantage cannot be replicated by any competitor beginning from scratch regardless of their model quality. Docusign’s AI combines LLM intelligence with orchestration and deep domain expertise from hundreds of millions of consented private agreements. Unity’s game engine runs on billions of devices globally, generating behavioral signals every time a player makes a decision in a game.
AI is only driving up the value of platforms. A new Deloitte study says while AI point products yield a modest 3% increase in ROI, customers deploying an end-to-end AI platform like Docusign’s IAM realize a nearly 30% increase. Similarly Rubrik describes “complementary network effects” whereby every new product they build makes every other product on the platform more powerful.
Our conclusion: the SaaSpocalypse narrative was compelling, but neglected that software is a prerequisite to AI. The data, workflow, and governance layer that enterprise software has been building for decades is a requirement for AI to be effective. The 50+ earnings calls we reviewed this quarter tell a consistent story: AI needs ontology (Palantir), it needs infrastructure (Cloudflare, Snowflake), it needs governance (SailPoint, Rubrik), it needs observability (Datadog, Dynatrace), it needs domain expertise (Procore, Certara, Doximity), and it needs the trust that enterprise SaaS companies have earned. None of those requirements are met by a foundation model built in a garage or by Kevin in IT. All of them are met by incumbent enterprise software. Enterprise software companies are posting record quarters, raising guidance, and building agentic products on top of their existing offerings. We think the best time to be and build an enterprise software company may be right now.
Thank you for your readership. See more blogs and SaaS data at blossomstreetventures.com. Other resources we’ve built for founders include: SoftwareMultiples.com; softwareMRRcalculator.com; FounderInvited.com. Founders are always welcome to reach out to sammy@blossomstreetventures.com as well.