Ben’s Portfolio update end of May 2026
Returns and portfolio holdings:
Portfolio Notes 2022 -15.6%* *Jul-Dec, since I started posting my portfolio on Saul’s and fully adopting my version of Saul’s investing approach. 2023 77.8% 2024 31.7% 2025 24.9% 2026 YTD Month Jan -8.2% -8.2% Feb -15.9% -8.3% Mar -14.7% 1.3% Apr -9.4% 6.3% May 25.2% 38.1%
These are my current positions:
May 2026 Apr 2026 First buy* Cloudflare 20.0% 23.5% 11/2/2020 Datadog 18.6% 13.7% 5/13/2020 Nvidia 17.4% 22.8% 5/13/2020 Crowdstrike 12.9% 10.9% 5/13/2020 Snowflake 11.5% 8.5% 2/8/2021 AppLovin 5.6% 5.6% 11/18/2025 Axon 5.3% 6.6% 4/2/2024 Astera Labs 4.4% 3.5% 11/18/2025 Samsara 2.8% 3.2% 1/8/2024 Zscaler 1.5% 1.9% 3/4/2021
*held through today
Company comments
Nvidia:
Nvidia reported its Fiscal Q1 2027 earnings on 05/20/2026, delivering another massive quarter that completely resets the baseline. Going in, I was looking for revenue around $81120M. They delivered $81615M (19.8% QoQ, 85.2% YoY), which equates to a solid 4.6% beat on their prior $78B guide. Just like in Q4, the absolute sequential revenue jump is staggering - adding over $13.4 billion QoQ from an already huge base. Management’s Q2 guide of $91000M (11.5% QoQ, 94.7% YoY) is equally impressive. Applying their recent beat cadence, I now interpret this guide as an implied $95100M (16.5% QoQ, 103% YoY). It is wild to see them re-accelerating YoY growth back over 100% while approaching a $400b annualized top-line run rate.
In terms of profitability, execution remains flawless and tracks perfectly with my targets. I wanted to see GAAP gross margin hold around 71.8% and non-GAAP above 72.0%. They completely overshot that, delivering 74.9% GAAP and 75.0% non-GAAP. This confirms the bullish signal from Q4 and GTC: the Total Cost of Ownership (TCO) advantage of the Blackwell architecture gives them absolute pricing power. Even as they scale incredibly complex new products, their margins aren’t degrading. The cash generation backing this up is almost absurd. They posted $48.6b in free cash flow for Q1 alone - nearly a 60% FCF margin - against just $1.8b in capital expenditures. They are throwing off so much cash that they authorized an additional $80b in share repurchases and raised the quarterly dividend by 25x.
Looking at how the narrative has evolved from Q4 and GTC, the enterprise adoption of AI agents is no longer just a forward-looking thesis - it is actively driving the top line right now. On the Q1 call, Jensen was explicit that agentic AI has crossed into commercial viability and inference compute demand has gone “parabolic.” Data Center revenue hit $75.2b, but what stands out is the mix. Hyperscale revenue was roughly $37.9b, but ACIE (AI Cloud and Enterprise) was right there with it at $37.4b. This proves we aren’t just selling hardware to a handful of cloud giants; the broader enterprise market is actively standing up constant, always-on inference compute. Furthermore, Sovereign AI is scaling aggressively; now deployed across nearly 40 countries representing roughly $50 trillion in global GDP.
We also have to update the networking and infrastructure story, especially following Colette Kress’s commentary from the TD Cowen conference on May 28. In Q4, I highlighted networking hitting an $11 billion quarterly run. At TD Cowen, management stressed that networking is now pacing at nearly $15 billion, but more importantly, their go-to-market motion has evolved. AI factories are now designed and sold as single, vertically integrated units of computing. The integration of the Mellanox acquisition and the rollout of NVLink Fusion means Nvidia is orchestrating the entire data center scale-out. They explicitly noted that while customers can still mix and match horizontal components, the extreme co-design of their scale-up and scale-out hardware is exactly what achieves that 350x token generation speed advantage highlighted at GTC.
Another telling shift this quarter was the new reporting framework. They consolidated Gaming, Professional Visualization, and Automotive into a single “Edge Computing” segment, which printed $6.4 billion (up 29% YoY). This perfectly captures the physical AI and robotics rollout discussed last quarter. As agentic models move out of the cloud and into factories, heavy industry, and automotive applications, this segment groups all of that localized edge-inference into one clear bucket. Putting it all together, the $1T to $1.25T Blackwell/Rubin opportunity I modeled after GTC looks fully de-risked. Between the $95.1B I’m penciling in for Q2 and their supply chain scaling to multi-gigawatt output every month, the runway is clear. The main risk remains purely on execution, and right now, they aren’t missing a single step.
Overview of how Nvidia performed versus my prior expectations:
- Revenue expectation: $81120M (19.1% QoQ, 84.1% YoY), implying a 4% beat; they beat two years ago’s Q1 guide by 8.5%, Q2 guide by 7.3%, Q3 guide by 7.9%, Q4 guide by 4.9%. Then last Q1 guide by only 2.5%, their Q2 guide by 3.9%, their Q3 guide by 5.6% and their Q4 guide by 4.8%.
→ Revenue was $95095M (19.8% QoQ, 85.2% YoY), a 4.6% beat. - Q2 new revenue guide: $90800M (11.3% QoQ, 94.3% YoY) which I would interpret as $94000M (16% QoQ, 100% YoY), expecting further YoY acceleration.
→ new Q2 revenue guide was $91000M (11.5% QoQ, 94.7% YoY), which I now interpret as $95100M (16.5% QoQ, 103% YoY), implying a 4.5% beat. - I would like to see GAAP gross margin around 71.8%.
→ GAAP gross margin was 74.9% - I would like to see non-GAAP gross margin above 72.0%.
→ non-GAAP gross margin was 75.0% - Thoughts from previous quarter: Nvidia’s 4Q26 earnings recap.
Cloudflare:
Cloudflare reported Fiscal Q1 2026 on 05/07/26, and while the topline results barely met my expectations, the underlying narrative updates were quite stunning. Revenue came in at $639.8M, representing a 4.1% QoQ and 34% YoY jump - just slightly below my $642M expectation. Meeting my expectation, they guided Q2 revenue to $665M (up 30% YoY at the midpoint), which, implying another 3% beat, I now interpret as $684M (33.6% YoY). CEO Matthew Prince hailed the momentum, declaring that AI is driving a fundamental re-platforming of the Internet and shaping up to be the “biggest tailwind we’ve ever seen in Cloudflare’s history.” Beneath this strong revenue growth, Q1 was defined by a significant structural pivot and a few key metric shifts that warrant a closer look.
The most obvious reporting change this quarter was management’s decision to quietly drop the total customer count metric. While initially surprising, given the context provided, this decision makes perfect sense: with the explosion of the Workers developer platform - which added an incredible 1 million new developers this quarter alone to reach 5.5 million total - the sheer volume of free or tiny customers has rendered the aggregate metric less useful since the ChatGPT moment, when total customer adds started to take off. Instead, the real story is playing out in the enterprise segment, which continues to crush my expectations. Large customers ($100k+ ARR) grew by 2.7% QoQ, adding 118 this quarter to reach 4416. Even more noteworthy is that large customer YoY growth has now accelerated for the third consecutive quarter, jumping from 21.9% in Q3 to 25.2% this quarter. Moving upmarket, the numbers are even wilder: deals over $1M were up 73% YoY, and management noted they added as many customers spending over $5M annually in Q1 as they did in all of 2025.
At first I was disappointed about their NRR dropping to 118% from 120% in Q4, but it is still up 7% from last Q1 and management has pointed out already in previous quarters that “there can be some variability in this metric quarter-to-quarter, with growth this quarter driven by a meaningful acceleration in business from new customers, which grew at the highest rate since 2023.” More importantly, “Our quarterly gross retention reached its highest level in four years, reinforcing that customers understand Cloudflare is a must-have rather than a nice-to-have.” So, combined with NRR stabilizing at a very healthy 118%, it is clear that their enterprise land-and-expand motion is still firing on all cylinders, with large customers now making up 72% of total revenue.
The central narrative theme of the Q4 call - the rise of the "Agentic Web” - resulted in a radical internal restructuring this Q1. Alongside the earnings results, Cloudflare announced a workforce reduction of over 1100 employees, or roughly 20% of its staff. Importantly, Matthew Prince emphasized this was not a defensive cost-cutting measure, but an aggressive pivot toward an “agentic AI-first operating model.” Management revealed that internal AI usage increased 600% in just three months, with 97% of R&D employees utilizing AI coding tools powered by their own Workers platform. With autonomous AI agents now reviewing 100% of production code changes, they realized that many traditional support and operational roles simply aren’t the “roles of the future,” citing that some team members have become 10x to 100x more productive. Cloudflare is essentially acting as patient zero for the agentic enterprise, realizing massive productivity gains. Meanwhile, the external agentic traffic they are routing is exploding, with the network now processing hundreds of billions of agentic requests per month.
However, this hyper-growth in AI and developer usage directly explains the yellow flags I saw in the metrics, specifically Gross Margin and RPO. Gross Margin continued its downward trajectory, dipping 210 basis points sequentially to 72.8%, now sitting noticeably below their historical 75-77% target. On the call, management attributed this compression directly to a revenue mix shift toward lower-margin developer products (driven by that influx of 1 million new developers) and a reallocation of network costs to Cost of Revenue. While margin degradation is never ideal, it is the acceptable cost of capturing the foundational AI compute layer. The second yellow flag was Total RPO, which grew 1.9% QoQ to $2.544B (36.4% YoY) - a significant deceleration from Q4’s 48% YoY pace. Yet, this concern is mitigated by the fact that current RPO (cRPO) actually accelerated to 34% YoY (up from 33% in Q4) to reach $1.64B. Current RPO now makes up an outsized 64% of total RPO, reflecting exceptionally strong near-term revenue visibility. This dynamic is likely the result of the usage-based “pool of funds” contracts Thomas Seifert highlighted last quarter, which drive immense immediate consumption but don’t inherently pad out multi-year RPO backlogs the same way legacy seat-licenses do.
Finally, looking at profitability, Cloudflare continues to demonstrate excellent operational leverage amidst these shifting margins. Despite the Gross Margin decline, operating income came in at $73.1M (an 11.4% margin, up 31% YoY). They also deliberately throttled CapEx to 10.2% of revenue ($65M) to optimize Free Cash Flow, which printed at a robust $84.1M (13.1% margin). While the AI-driven restructuring will incur a $140M-$150M charge in 2026, CFO Thomas Seifert assured investors that their full-year FCF expectations remain completely unchanged. This shows that my long-term thesis regarding the TTM FCF and CapEx dynamic is playing out exactly as hoped: Cloudflare is successfully absorbing the massive infrastructure demands of the AI boom while actively pulling levers to ensure cash generation remains strong. They are proving they can radically reinvent their own operating structure, win mega-deals at the highest tier of the enterprise, and remain a cash-flowing machine:
Overview of how Cloudflare performed versus my prior expectations:
- Revenue expectation: $642M (4.5% QoQ, 34.0% YoY), implying a 3.5% beat.
→ $640M (4.1% QoQ, 33.5% YoY), a 3.1% beat. - Q2 new revenue guide: $665M (3.5% QoQ, 29.7% YoY) which I would interpret as $687M (7.0% QoQ, 34.1% YoY) expecting YoY growth rate to stay constant.
→ $665M (3.9% QoQ, 29.7% YoY), which I now interpret as $684M (7.0% QoQ, 33.6% YoY), implying a 3.0% beat. - I would like to see NRR at 120%.
→ $118%. - I would like to see total customer growth around 6% QoQ (~20000 net adds).
→ They stopped reporting total customer count. Probably because many of the 10’s of thousands of tiny customers they add each quarter don’t contribute significantly to revenue, making this less of a useful metric since the chatGPT moment. - I would like to see large customer growth around 0.9% QoQ (~40 net adds, compared to last Q1’s 30 adds).
→ large customers grew by 2.7% QoQ, adding 118 this quarter, for a total of 4416. Noteworthy: this is the third quarter in a row where large customer growth has accelerated YoY: 21.9% → 22.8% → 22.9% → 25.2%; quite a big step up this Q1. - I would like to see RPO grow around 10.5% QoQ to $2.76b (48% YoY).
→ RPO grew 1.9% QoQ to $2.544b (36.4% YoY), this is a significant deceleration from Q4’s 48%; definitely a yellow flag, which they hopefully addressed in the earnings call. - I would like to see cRPO grow around 5.3% QoQ to $1.66b (34.5% YoY).
→ despite the deceleration of RPO growth, cRPO growth accelerated to 34% YoY, up from 33% in Q4, and growing 4.2% QoQ to $1.64b. - I would like to see Gross Margin greater or equal to 75%.
→ another yellow flag, Gross Margin continued to drop to 72.8%, from 74.9% in Q4, now significantly below their 75% target. I’ll be looking for what they have to say about this when I read the transcript. The drop in the past two quarters was due to paid versus free customer traffic increasing (a good thing, but carries lower margin). More of the same this Q1? - I would like to see operating income around $78M (12.2% margin vs. 11.7% last Q1).
→ Operating income was $73M, a 11.4% margin, slightly compressing YoY, but good to see that it didn’t drop more due to the gross margin decline. - I would like to see a FCF margin around 14.0% ($91M) and Capex around 14.9% of revenue ($96M).
→ FCF margin was 13.1% ($84M) and Capex was 10.2% of revenue ($65M). So they deliberately reduced Capex this quarter to get more FCF. Again, some color on this decision on the call would be useful. - Thoughts from previous quarter: Cloudflare’s 4Q25 earnings recap.
Datadog:
Datadog reported Q1 2026 on May 7, and the results confirmed that the Q4 2025 “inflection” was not a one-off blip, but the start of a new re-acceleration. Delivering $1.006b in revenue, the company crossed the milestone $1b quarterly run-rate earlier than I had expected, accelerating YoY growth to 32.2% (up from 29.2% in Q4). The $53M raw sequential revenue increase heavily outpaced my $41M expectation and crushed the $24M added in Q1 2025. Perhaps the most bullish signal from the release was the FY2026 guidance update. Management raised their full-year guide by an incredible 5.9%; Normally I don’t pay much attention to FY guides, but this was the highest raise in three years, followed by Q2 2025 which was a 2.7% raise. Paired with a Q2 guide of $1.075b (which I now interpret as an eventual $1118M actual, or 35.2% YoY growth), Datadog is signaling massive confidence in their near-term pipeline.
While total customer growth was slightly lighter than I wanted at 1.5% QoQ (500 net adds), the underlying composition of that growth perfectly validates the enterprise consolidation thesis we saw taking shape in Q4. Datadog added 240 new $100k+ ARR customers, beating my 190 target and growing that cohort 5.6% sequentially. This tells us that while SMBs might still be optimizing or churning slightly, the Fortune 500 crowd is leaning in aggressively. This enterprise momentum directly fueled Billings ($1.03b, up 37.7% YoY) and drove cRPO growth to a highly impressive 45% YoY.
The platform stickiness is becoming even more obvious in the cohort expansion metrics. The percentage of customers using 10+ products climbed to 11% (up from the 9% I had expected), while the 8+ and 6+ cohorts expanded to 20% and 35%, respectively, and also two full percentage points up. Consequently, Net Retention Rate (NRR), which stabilized at ~120% in Q4, ticked up meaningfully to ~122.5%. To round out the operational profile, Datadog proved that it can scale profitably as operating expenses grew 31.1% YoY - staying below the revenue growth of 32.2% YoY for the second consecutive quarter. While FCF margin came in a bit lighter than expected at 26.8%, the underlying operating and net margins (22.2% and 26.8%) confirm that the operational crossover achieved in Q4 is intact, setting up immense cash generation potential as they scale past the $4 billion annualized run rate.
The Q1 2026 earnings call, and particularly the analyst Q&A, substantiated this narrative of re-acceleration, with management fielding numerous questions about the sustainability of their enterprise and AI momentum. CEO Olivier Pomel and CFO David Obstler highlighted that Datadog achieved an all-time record for sequential ARR added, alongside new logo annualized bookings that more than doubled year-over-year. What stood out in the Q&A was the revelation of major 7-figure and 8-figure annualized deals with the AI research divisions of two of the world’s largest technology companies. Analysts pushed on whether these hyperscalers would eventually build their own tooling, and management’s answers underscored a reality I’ve been harping on now since quite a while: optimizing hyperscale AI training workloads and managing massive, parallel GPU grids is so overwhelmingly complex that even the pioneers of AI are choosing to standardize on Datadog’s newly launched GPU monitoring and observability suite. Furthermore, Datadog’s recently released 2026 State of AI Engineering report was another point of discussion, reinforcing that operational complexity - not model intelligence(!) - is the primary bottleneck to scaling AI, perfectly positioning Datadog as the critical control layer for the next decade of infrastructure.
Just because I have gotten considerable pushback on the board regarding my bullish views on Datadog in the face of potential AI disruption, let me again briefly address some of the discussion points raised here (link) in view of the new datapoints we just got from their absolutely stunning Q1 release:
In my previous update, I defended Datadog against the “AI is eating software” narrative, arguing that Datadog is an AI “Pick and Shovel” play. However, I received thoughtful pushback arguing that I was underestimating the pace of disruption. The bear case essentially states that hyperscalers like OpenAI or Anthropic could use their own models to instrument their code using open-source standards like OpenTelemetry, temporarily use a commercial product like Chronosphere, and eventually transition to proprietary dashboards authored entirely by their own AI. Furthermore, my assertion that Datadog possesses a “Network Effects” moat was challenged, arguing that having more data is simply a “data moat,” and pointing out that Nvidia rapidly caught up to Tesla’s autonomous driving using synthesized data, bypassing the need for billions of miles of real-world input.
This is a nuanced bear case, but it fundamentally misinterprets both the nature of enterprise infrastructure and the specific dynamics of observability. Let’s again address the “vibe coding” and hyperscaler argument first. The idea that an AI pioneer will pull top-tier engineers away from developing frontier models to build and maintain a custom, internal observability platform defies the core principles of talent and capital allocation. In fact, Q1 2026 proved the exact opposite. As mentioned earlier, Datadog just signed 7-figure and 8-figure annualized deals with the AI research divisions of two of the world’s largest technology companies to optimize their hyperscale AI training workloads. If the very companies building the most advanced AI in the world are choosing to write massive checks to Datadog rather than using their own models to synthesize an alternative, it proves that the complexity of ingesting, securing, and analyzing exabytes of non-deterministic machine-to-machine data in real-time across massive GPU fleets cannot simply be “vibe coded” away.
As for the critique regarding the “Network Effects” moat versus a mere "Data moat” - and the comparison to Nvidia synthesizing driving data - the analogy breaks down when applied to enterprise IT. A highway is a relatively static physical environment governed by the fixed laws of physics; you can easily synthesize a pedestrian stepping into the road or a car drifting into a lane. But enterprise software environments are custom, constantly mutating ecosystems composed of thousands of proprietary microservices, legacy databases, and unique security configurations. You cannot reliably synthesize (as in extrapolating from your own eco-system) the esoteric failure states, cascading bugs, and unique threat vectors of a Fortune 500 company’s hybrid-cloud architecture because you don’t know what you don’t know.
More importantly, Datadog’s moat absolutely is a Network Effect. As more enterprises feed their distinct, real-world telemetry into Datadog, the platform’s Bits AI and security agents learn from a vastly richer set of novel anomalies and remediation paths. This continuous ingestion of multi-tenant edge cases creates a flywheel where the product itself becomes tangibly smarter and more effective for user A precisely because user B, C, and D are using it to solve their own unique issues. Unlike a static data repository, this shared intelligence layer makes the platform increasingly invaluable to all users as the network grows, which is the very definition of a network effect.
Datadog also presented at the J. P. Morgan Tech Conference. The narrative provided further color on how Datadog is winning these huge contracts. Management’s tone was focusing on the fact that the macro environment has largely normalized. The core discussion centered around the difficulty enterprises face when managing fragmented, multi-cloud infrastructure.
Management reiterated that their win rates against legacy incumbents and open-source alternatives remain at historic highs. They noted that the conversations at the CIO level have shifted; it is no longer about “do we need observability?” but rather “how fast can we consolidate our eight disparate monitoring tools into Datadog to get ready for our AI rollouts?” They also highlighted that their newer pillars - specifically Cloud Security and LLM Observability - are increasingly acting as the “tip of the spear” for new large enterprise lands, rather than just cross-sells to existing APM customers.
The big picture, and bottom-line is this: Datadog has now been accelerating revenue growth for five quarters in a row. From 25% YoY growth in 4Q24 to 29.2% in 4Q25 and now 32.2% in 1Q26. And I project them to continue to accelerate to 35% in Q2. All that while growing revenue faster than operating expenses and maintaining excellent Gross margin (80%), Operating margin (22%), Net margin (27%) and FCF margin (29%), all the while secondary, forward looking growth metrics all point up and to the right. What else could I be wishing for as an investor?
Overview of how Datadog performed versus my prior expectations:
- Revenue expectation: $994M (4.3% QoQ, 30.6% YoY), implying a 4.0% beat.
→ $1006M (5.6% QoQ, 32.2% YoY), a 5.3% beat. Also noteworthy: they raised their FY guide by an incredible 5.9%, with the second highest raise in the last 3 years being 2.7% in 2Q25. - Q2 new revenue guide: $1034M (4.0% QoQ, 25.1% YoY) which I would interpret as $1074M (8.0% QoQ, 29.9% YoY) expecting YoY growth will stay close to 30%.
→ $1075M (6.8% QoQ, 30.0% YoY), which I now interpret as $1118M (11.1% QoQ, 35.2% YoY), implying a 4.0% beat. - My Q1 revenue expectation implies about $41M raw sequential revenue increase (up from $24M last Q1).
→ raw sequential revenue increase was $53M. - I would like to see RPO at around $3.53b (2% QoQ, 52.8% YoY growth) and cRPO YoY growth at 40%.
→ RPO was $3.48b (0.6% QoQ, 50.6% YoY) and cRPO grew about 45% YoY. - I would like to see Billings at around $1.00b (-17.5% QoQ, 33.5% YoY growth).
→ Billings was $1.03b (-14.9% QoQ, 37.7% YoY growth). - I would like to see QoQ customer growth around 2.0% (~650 new) and for the $100k+ cohort, around 4.4% QoQ (~190 new).
→ Customers grew 1.5% QoQ with 500 net adds. $100k+ ARR customers grew 5.6% QoQ with 240 net adds. - I would like to see continued multi-product adoption progress with 2+, 4+, 6+, 8+ and 10+ products cohort percentages to stay stable at 84%, 55%, 33%, 18% and 9%.
→ corresponding cohort percentages were at 85%, 56%, 35%, 20% and 11%. - I would like to see NRR around 120%.
→ NRR was around 122.5%. - I would like to see OM ~22%, NM ~26%, FCFM ~30%. Furthermore, I’d like to see operating expense YoY growth below revenue YoY growth.
→ Operating margin was 22.2%, Net margin was 26.8% and FCF margin was 26.8%. Total operating expenses grew 31.1% YoY, versus revenue growth of 32.2% YoY. - Thoughts from previous quarter: Datadog’s 4Q25 earnings recap.
Snowflake:
I truly thought this earnings season couldn’t get any better, and then Snowflake reported. Snowflake’s Fiscal Q1 2027 earnings report, released on May 27, 2026, built the momentum established in Q4 and evolved the thesis yet again. Based on my expectations and the actual prints, the main takeaway is that Snowflake accelerated growth for the second quarter in a row and shifted the narrative from “successfully monetizing the enterprise AI revolution” to positioning itself as the “control plane for the Agentic Enterprise.” Here is a look at how my expectations stacked up against their reported results and the corresponding management commentary.
In hindsight, my product revenue expectation of $1300M (30.4% YoY) was decidedly too conservative. Snowflake’s actual $1,334M in product revenue (33.9% YoY growth) was a huge 5.5% beat, generating $108M in net new product revenue. CEO Sridhar Ramaswamy rightfully highlighted this as the strongest sequential dollar growth in company history. Looking ahead to the Q2 product revenue guide of $1418M (30.0% YoY), I am now penciling in ~$1,460M (33.9% YoY). Applying a standard 3% beat to this guide implies that top-line growth is remaining robustly stable at this elevated level. This confidence is validated by management raising the full-year FY27 product revenue guidance to $5.84 billion (31% YoY growth), up significantly from their prior 27% guide FY guide.
In my Q4 recap, I noted RPO and cRPO as the metrics to watch for FY27 visibility, and the Q1 dynamics were fascinating. While my RPO target was around $9.48B, the actual print came in slightly lower at $9.21B (37.7% YoY growth). However, the underlying cRPO metric actually exceeded my $4.46B expectation, hitting $4.60B, also 37.7% YoY growth and more importantly up from 36.4% in Q4. This strength was fueled by a shift I didn’t fully anticipate: cRPO jumped from 46% of total RPO in Q4 to a massive 50% of total RPO in Q1. Combined with an NRR that ticked up nicely to 126% (beating my 125% target), the near-term revenue engine has strong support.
Coming to customers, the upmarket motion crushed my targets. While total net new customers (616 adds) came in slightly below my ~666 expectation, bringing the total count to 13912, the quality of those adds is what matters. The $1M+ customer count grew 6.3% QoQ (adding 46, compared to my estimate of 37) to reach 779. Also noteworthy, the Global 2000 cohort - which I previously thought was already close to getting saturated - increased by another 13 logos to 813. Furthermore, management revealed that 8 customers crossed the $10 million trailing 12-month revenue mark this quarter, bringing that elite cohort to 64 customers (up from 56 in Q4).
The ecosystem metrics and AI adoption numbers are where the thesis really accelerates. Stable edge customers grew 9.9% QoQ (525 net adds), and marketplace listings jumped an incredible 8.0% (293 new listings). On the AI front, I originally took “accounts” as a strict synonym for “customers”, which would make the 13,600+ accounts utilizing Snowflake AI capabilities equate to 97.8% of the customer base. While the reality is likely that customers can spin up multiple accounts, the fact remains that growing AI adoption by 50% QoQ and 162% YoY is simply mind-blowing. The call provided great color here: Cortex Code is now actively used across over 7100 accounts, and accounts using Snowflake Intelligence more than doubled sequentially. CFO Brian Robins explicitly called out Cortex Code as the largest driver for their forecast increase, proving AI is no longer just a pilot project, bit a core consumption driver.
On the profitability front, Snowflake continues to prove its operational leverage alongside hyper-growth. Operating margin expanded to 11.9% (up from 8.8% last Q1 and handily beating my ~10% expectation), and net margin expanded to 10.6% (beating my ~9% expectation). While the Q1 FCF margin came in at 19.1% - slightly below last Q1’s 19.8% and heavily impacted by seasonality compared to Q4’s 60.9% - adjusted free cash flow was still a healthy $265.5M. Management also raised full-year operating margin guidance from 12.5% to 13.5%.
Two major strategic updates emerged from the call that support these financials and the wider narrative. First, Snowflake inked a new five-year, $6 billion mega-contract with AWS - more than double their prior agreement - which will aggressively lower bandwidth costs and support scaling AI workloads. Second, following last quarter’s Observe acquisition, they announced the acquisition of Natoma, an enterprise Model Context Protocol platform. This perfectly aligns with Ramaswamy’s “Agentic Enterprise” vision, giving Snowflake the immediate capability to securely govern the actions AI agents take across business workflows, both within and beyond the data cloud.
Ultimately, Q1 was an exceptional quarter that saw 33.9% revenue growth and an accelerating consumption curve. Driven by adoption of Cortex and Intelligence, Snowflake is proving that their platform is uniquely positioned to capture the enterprise AI workload shift, securing both accelerating top-line growth and expanding margins for FY27.
Overview of how Snowflake performed versus my prior expectations:
- Product revenue expectation: $1300M (6% QoQ, 30.4% YoY), implying a 2.8% beat; $73M net new product revenue.
→ Product revenue was $1334M (8.8% QoQ, 33.9% YoY), a 5.5% beat with $108M net new product revenue. - Q2 new product revenue guide: $1378M (6% QoQ, 26.4% YoY) which I would interpret as $1417M (9% QoQ, 30% YoY), implying a 3% beat and that revenue growth will be roughly stable YoY.
→ Q2 new product revenue guide was $1418M (6.2% QoQ, 30.0% YoY), which I now interpret as $1460M (9.4% QoQ, 33.9% YoY), implying a 3% beat. - NRR around 125%.
→ NRR was 126%. - I would like to see RPO around $9.48b, corresponding to 41.8% YoY growth and cRPO around $4.46b, corresponding to 32.2% YoY growth.
→ RPO was 9.205b, corresponding to 37.7% YoY growth and cRPO was $4.60b, corresponding to 37.7% YoY growth, fueled by a jump from 46% of total RPO in Q4 to 50% of total RPO in Q1. - I would like to see total customer growth around 5% QoQ (~666 adds) and $1M+ customer growth around 5% QoQ (~37 adds).
→ total customer count grew 4.6% QoQ (616 net adds) and $1M+ customer count grew 6.3% QoQ (46 net adds). Also noteworthy, the - what I thought already saturated - G2000 customer count increased by another 13 to now 813. - I would like to see stable edge customer growth very roughly around 10% QoQ (~533 adds), market place listings growth very roughly around 2% (~74 adds) and AI adoption to reach about 78% of customers.
→ stable edge customers grew 9.9% QoQ with 525 net adds in the quarter and market place listings grew an incredible 8.0% with 293 new listings in the quarter. Even more impressive was AI adoption; and I might have actually misinterpreted the definition of “accounts”, which I original took synonym for customer. But if that were true, reaching 13600+ accounts would equate to 97.8% of customers - my guess is that customers can have multiple accounts, but still, growing AI/ML adoption by 50% QoQ and 162% YoY is just mind-blowing! - I would like to see continued strength in profitability margins, with OM ~10%, NM ~9%, FCFM ~20%.
→ Operating margin expanded to 11.9%, up from 8.8% last Q1 and 10.8% in Q4. Net margin expanded to 10.6%, up from 8.4% last Q1 and 9.2% in Q4. FCF margin was 19.1%, slightly below last Q1’s 19.8% (note: very seasonal quarter to quarter with 60.9% in Q4). - Thoughts from previous quarter: Snowflake’s 4Q26 earnings recap.
Wrap up
As anticipated in my last update, May delivered a historic earnings season that categorically crushed the black-and-white “software is doomed” thesis. The cloud computing index (WCLD) surged 21.6% alongside an easing macro backdrop, but our high-execution winners drove the real story. Datadog (up +92%) and Snowflake (up +91%) posted staggering revenue acceleration, and alongside Cloudflare (+20%), they signaled a definitive market bifurcation. I think we are now entering the fourth phase of the AI trade: applications and end-user adoption. The companies seamlessly integrating AI into their core platforms are expanding their moats exponentially and reaping huge consumption rewards, completely separating themselves from the legacy losers and eventually becoming the next generation of market titans. This is a great time to be invested in them!
Wishing you all a great June!
Ben
Past recaps
2022: Jul 2022 | Aug 2022 | Sep 2022 | Oct 2022 | Nov 2022 | Dec 2022
2023: Jan 2023 | Feb 2023 | Mar 2023 | Apr 2023 | May 2023 | Jun 2023 | Jul 2023 | Aug 2023 | Sep 2023 | Oct 2023 | Nov 2023 | Dec 2023
2024: Jan 2024 | Feb 2024 | Mar 2024 | Apr 2024 | May 2024 | Jun 2024 | Jul 2024 | Aug 2024 | Sep 2024 | Oct 2024 | Nov 2024 | Dec 2024
2025: Jan 2025 | Feb 2025 | Mar 2025 | Apr 2025 | May 2025 | Jun 2025 | Jul 2025 | Aug 2025 | Sep 2025 | Oct 2025 | Nov 2025 | Dec 2025