The “AI is going to eat SaaS” thesis ran into a tape it could not argue with on Wednesday. Datadog reported Q1 revenue of $1.006 billion, up 32 percent year over year, generated $289 million of free cash flow, told analysts AI workloads are driving demand rather than cannibalizing it, and watched its stock surge 31 percent. Fortinet beat estimates, lifted full-year guidance on cybersecurity strength, and added 23 percent. DoorDash popped close to 10 percent on a 33 percent revenue jump tied to its Deliveroo integration. Different end markets, same lesson: the AI build-out is feeding the software stack, not eating it, and the public market is finally pricing that in.
The Datadog Print Was The Whole Argument
Datadog’s number is the cleanest rebuttal of the AI-kills-SaaS thesis the year has produced. Revenue at $1.006 billion crossed a billion for the first time in a quarter and grew 32 percent, the highest growth rate since the post-2022 normalization. Free cash flow of $289 million translated to a 29 percent free cash flow margin. Net new customers, dollar-based net retention, and platform attach all came in ahead of model. The pattern that mattered most: large customer cohorts are paying Datadog to instrument the AI workloads they are deploying inside their own platforms.
CEO Olivier Pomel was direct on the call, per the company’s own Q1 investor relations release. AI workloads are more complex than the cloud workloads they augment, with longer dependency chains, GPU and accelerator instrumentation, aggressive autoscaling, and higher security attack surface. Each is exactly what Datadog’s observability platform exists to handle. The AI buildout is not a substitute for observability. It is a forcing function for more of it.
That is the story Wall Street had been afraid to underwrite. From mid-2024 through early 2026, the dominant cocktail-party thesis among growth investors was that foundation model agents would compress the SaaS stack into a few horizontal layers, with companies like Datadog rendered redundant. Q1 was the first quarter in which Datadog’s actual customer behavior produced the opposite read at scale. The 31 percent move on the print is the market revising the thesis in real time.
Fortinet And The Cybersecurity Tailwind That Has Not Stopped
Fortinet’s quarter is the supporting evidence. The company beat top and bottom line and raised full-year guidance, with strength concentrated in OT security, secure SD-WAN, and the cloud-delivered SASE bundle. Why this matters in the context of AI: every additional AI workload a customer puts into production multiplies the attack surface that has to be defended. Inference endpoints, model APIs, retrieval-augmented databases, vector stores, and agent orchestration layers each create new vectors that did not exist in the pre-2023 stack.
Fortinet’s earnings argued that customers are not absorbing that surface with internal teams or with horizontal AI tools. They are buying more from the existing security vendor bench, and Fortinet specifically has captured a disproportionate share of the AI-driven security demand because of its hardware-software-software-service integration. The 23 percent move on the print suggests the market had under-allocated capital to the same thesis that drove Datadog. Both companies are the picks-and-shovels exposure to the AI capex cycle, but at the operational layer rather than the silicon layer.
DoorDash Is The Reminder That AI Is Not Eating Every Vertical
DoorDash’s almost-10 percent move on a 33 percent revenue jump is a different signal but it lands in the same week. The Deliveroo integration is closing faster than the model expected, international order growth is broad-based, and the company’s read on logistics demand suggests the consumer is not retrenching. AI gets some credit here on the operations side, where DoorDash has been quietly shipping ML-driven dispatch and demand forecasting that is improving unit economics quarter over quarter. But the simpler read is that the company is executing through a moment when the software-platform thesis was supposed to be in trouble and the underlying business is getting more resilient, not less.
For investors watching the SaaS multiple compression of the past 18 months, this is the first earnings stretch where the underlying business numbers actively contradict the bear case. Public software has been priced as if the AI cycle was deflationary. The Q1 prints from Datadog, Fortinet, and DoorDash say it is neutral or actively additive depending on the company.
Why The Bear Thesis Always Was Selectively True
The AI-kills-SaaS argument was never wrong in its entirety. It was right about a specific category. Legacy creative tools, marketing automation suites that look like they were designed for 2014, and any vertical SaaS company whose moat depended on UI craft rather than data integration are genuinely facing a cliff. We covered the most public version of this when Adobe’s CEO resignation laid bare the AI identity crisis inside Big Software earlier this year. That story is still real, and it will keep producing casualties.
What this week’s tape clarifies is that the cliff is not the same shape across the whole SaaS universe. Observability, security, infrastructure, payments, and platform companies that sit beneath the application layer are seeing the opposite trajectory. The customer doing AI is buying more of their software, not less. The companies that get hurt are the ones whose product was already replaceable by a sufficiently good prompt. The companies that get fed are the ones whose product becomes load-bearing the moment AI-generated code, agents, and inference endpoints land in production.
The Investor Read
If you are positioning a public-market software book in the second half of 2026, the Q1 prints reset the framework. Owning the picks-and-shovels of AI is not just a silicon trade. The same thesis that drove AMD’s data center number this week, that hyperscalers and large enterprises will keep spending into the build-out, expresses inside the software stack as observability spend, security spend, and platform spend. Datadog, Fortinet, and adjacent names like Cloudflare, Crowdstrike, and Snowflake are the natural beneficiaries.
The risk in the trade is not whether AI eats SaaS. It is whether AI capex itself slows. According to Bloomberg’s read on the megacap earnings stretch, hyperscalers committed more than $130 billion in AI capex in Q1 alone. As long as that number keeps compounding, the demand layer that Datadog and Fortinet are tapping continues to widen. If hyperscaler capex flattens or contracts in 2027, the same trade compresses fast.
What To Watch From Here
Three reads matter for the rest of the year. First, the next observability and security earnings stretch, especially the Snowflake, Cloudflare, and Crowdstrike prints, will tell you whether the Datadog and Fortinet beats were idiosyncratic or sector wide. Second, watch attach rates inside the hyperscalers. AWS, Azure, and Google Cloud have their own observability and security products, and a competitive read will surface in attach commentary on Q2 and Q3 calls. Third, watch the AI-native software bench. Mistral, Anthropic, and OpenAI are pushing toward platform layers that overlap with the same workloads Datadog and Fortinet sell into, and whether the AI labs build, buy, or partner is unsettled.
For now, the headline is simpler than the cocktail-party thesis suggested. The companies that build the floor under AI workloads are the ones being rewarded by the market right now. The crowd that has been calling for SaaS’s funeral has just had a hard quarter. Datadog, Fortinet, and DoorDash all printed numbers that say the AI cycle is feeding the software stack, not displacing it, and the tape is finally pricing that in.