Ben’s Portfolio update end of November 2025
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 YTD Month Jan 9.1% 9.1% Feb 5.0% -3.7% Mar -10.4% -14.7% Apr -2.4% 8.8% May 19.6% 22.6% Jun 32.7% 11.0% Jul 33.5% 0.6% Aug 29.9% -2.7% Sep 34.2% 3.3% Oct 51.0% 12.5% Nov 30.0% -13.9%
These are my current positions:
Nov 2025 Oct 2025 First buy* Cloudflare 21.4% 23.3% 11/2/2020 Nvidia 19.9% 21.1% 5/13/2020 Snowflake 14.0% 14.2% 2/8/2021 Datadog 13.9% 9.8% 5/13/2020 Crowdstrike 11.1% 10.9% 5/13/2020 Axon 6.5% 8.2% 4/2/2024 Samsara 4.1% 4.0% 1/8/2024 Zscaler 3.6% 4.4% 3/4/2021 Monday 2.4% 3.1% 9/13/2021 AppLovin 1.5% 0.0% 11/18/2025 Astera Labs 1.5% 0.0% 11/18/2025 TradeDesk 0.0% 0.9% 5/13/2020
*held through today
Company comments
Cloudflare:
Key insights:
- Delivered a stunning quarter with revenue growth acceleration continuing and will likely continue through Q4.
- Land-and-expand is literally exploding with a record 30k new customers in Q3 (~2x previous record add) and NRR jumping to 119% from 114% in Q2.
- NRR has potential for further expand when the self-serve or lower tiers (driving the high customer volume count) are expanding rapidly as their own inference workloads scale and as the new customers mature into the large customer cohort.
- Large customer growth itself was also impressive as it grew 8% QoQ adding a record 297 large customers (in comparison to 219 adds last Q3 and 185 adds in Q2).
- Even though gross margin pressures continue to create a temporary headwind on the bottom-line, operating expenses grew only 22.3% YoY, while revenue grew 30.7% YoY, resulting in operating margin expansion to 15.3%, up from 14.8% last Q3, net margin expansion to 18.3%, up from 16.9% last Q3 and FCF margin expansion to 13.3%, up from 10.5% last Q3.
Cloudflare reported Fiscal Q3 2025 on 10/30/25. This was an absolutely stunning quarter all the way from top to bottom line! My revenue expectation was handily exceeded with a 13 quarter QoQ revenue growth record of 9.7% and continuing the trend of accelerating YoY growth: Q1: 26.5% → Q2: 27.8% → Q3 → 30.7%. Just look how the raw YoY increase in revenue as a function of time looks like (including my projection for Q4):
The outlook suggests that the accelerating trends will continue through Q4, where I have penciled in 31.3% YoY growth (implying a 2.5% beat in comparison to this Q’s 3.3% beat). They also raised their FY guidance by 1.3%, which was the strongest raise in 13 quarters. This is significant because Cloudflare hasn’t really raised any of their initial FY guides in the last few years, besides Q2’s 1% raise, which this Q3’s raise comes on top of. And they have only beaten any FY guide in the previous two FYs by up to 1.1%. The question now of course is if they can keep this accelerated revenue growth rate into FY 2026. Let’s explore it!
Well, secondary growth metrics would suggest just that: Their land and expand motion is firing on all cylinders with NRR jumping up to 119% (was 110% last Q3 and 114% in Q2). Even more impressive was their new customer acquisition: They managed to add an incredible 29623 customers this quarter (33.4% YoY). How can this not be a typo??? Cloudflare has clearly gone through a sea change with regard to customer acquisition, as seen when we look at their historical adds, which I think not coincidentally started to take off a quarter after the infamous chatGPT moment:
The explosion in customer count is primarily driven by the Workers Developer Platform and AI-native startups. CEO Matthew Prince noted that “80% of the leading AI companies already rely on us.” These customers often start on the self-serve or lower tiers (driving the high volume count) but are expanding rapidly as their own inference workloads scale. This is the developer-led motion returning. Just as AWS won by capturing startups that became giants, Cloudflare is capturing the AI supply chain early. This surge is a leading indicator for future Enterprise growth, as these 29k+ new customers mature into the large customer bucket over the next 12-24 months. Indeed, we already see this process happening from earlier customer adds as NRR has increased to 119% from 110% a year ago.
And while this jump in NRR (also from 114% in Q2) is a major signal it requires context to understand its durability: Management explained that we are now seeing the “consumption phase.” Customers are burning through those prepaid credits faster than expected, which triggers overages or early renewals/expansions. This suggests the 119% isn’t a one-off blip but a release of pent-up value that was hiding in the RPO.
Large customer growth was also impressive: They grew 8% QoQ adding a record 297 large customers. And with land and expand in full motion RPO grew 42.6% YoY and cRPO grew 30.3% YoY.
So, with total customer count growing 33.4% YoY and NRR of 119% further revenue growth acceleration from currently 30.7% YoY is certainly on the table. At least I think we’ll see strong durability at the 30% level for several quarters to come. But surely, they must have paid a high price tag in profitability to achieve that revenue growth acceleration, right?
To start, their gross margin continued to drop 100 basis points to 75.3% (But still within their goal of 75% -77%), due to “paid versus free customer traffic again increased both year-over-year and quarter-to-quarter, resulting in a higher allocation of expenses to cost of goods sold from sales and marketing”. Let me explain this: Counter-intuitively, “paid” traffic is currently hurting margins more than “free” traffic. This is likely due to the heavy compute nature of the new AI/Workers workloads. Unlike static CDN traffic (caching images is cheap), executing AI inference or Workers scripts requires more expensive GPU/CPU compute at the edge. Management is explicitly trading ~100bps of margin to capture the AI inference market share now. They view this as land grabbing the AI infrastructure layer. As they optimize their own hardware stack (Gen X servers), they expect to claw this margin back.
On the other hand, operating expenses grew only 22.3% YoY, while revenue grew 30.7% YoY. And operating expense margins actually contracted to 60% from 64.1% last Q3. So really great operational leverage at play! With that, operating margin expanded to 15.3%, up from 14.8% last Q3, net margin expanded to 18.3%, up from 16.9% last Q3 and FCF margin expanded to 13.3%, up from 10.5% last Q3. With now over $200M in TTM FCF we might actually start to see the beginning of another reversal in FCF versus Capex where FCF could again start taking off, which of course would meaningfully improve any discounted cash flow analysis. (originally discussed here: Ben’s Portfolio update end of May 2024 and here, in May 2024: Cloudflare’s fascinating Q1 2024 earnings - #20 by SlowAndFast).
There were two other narrative updates, post earnings: Just days after earnings, it was announced that CJ Desai (President of Product & Engineering) is leaving to become CEO of MongoDB. This is a significant loss. Desai was viewed as an experienced, results-oriented industry veteran who operationalized the sales force and helped drive the enterprise sales motion. The narrative risk here is whether the new sales discipline is institutionalized enough to survive his exit, or if we will see execution hiccups in FY26. Also, on November 18, Cloudflare suffered a significant outage affecting major services like ChatGPT and X. While these happen to all infrastructure providers, it does temporarily bruise the reliability premium they charge enterprises. However, historically, Cloudflare has used these incidents to publish post-mortems that actually increase engineering trust long-term.
Overview of how Cloudflare performed versus my prior expectations:
- Revenue expectation: $555M (8.3% QoQ, 29.0% YoY), implying a 2.0% beat.
→ $562M (9.7% QoQ, 30.7% YoY), a 3.3% beat. - Q4 new revenue guide: $583M (5% QoQ, 27% YoY) which I would interpret as $594M (7% QoQ, 29% YoY) YoY growth rate to stay roughly at 29%.
→ $589M (4.8% QoQ, 28.1% YoY), which I now interpret as $604M (7.4% QoQ, 31.3% YoY), implying a 2.5% beat. - I would like to see NRR at 114%.
→ NRR was 119%(!). - I would like to see total customer growth around 5% QoQ (~13300 net adds).
→ 11.1% QoQ to 295552 and 29623 net adds(!). Also, YoY customer growth accelerated to 33.4% from 21.7% last Q3. Now compound that with an NRR increase to 119% … - I would like to see large customer growth around 5.5% QoQ (~204 net adds).
→ 8.0% QoQ to 4009 and 297 net adds. - I would like to see RPO grow around 6% QoQ to $2.1b.
→ 8.4% QoQ to $2.14b. - I would like to see cRPO grow around 6% QoQ to $1.38b.
→ 5.1% QoQ to $1.37b. - I would like to see operating income around $84M.
→ $86M and margin expanded to 15.3% from 14.8% last Q3. - I would like to see a FCF margin in-between 7% and 10%.
→ FCF margin was 13.3%. - Thoughts from previous quarter: Ben’s Portfolio update end of July 2025
Axon:
Key insights:
- Axon missed my expectations both on the top and bottom line, moving it down to my lower conviction tier for now.
- The longer term picture of a growth investment through stacking up multiple S-curves is still intact though, evident in bookings, customer metrics, and ecosystem expansion, showing significant acceleration.
- The bottom-line miss though was primarily driven by a combination of tariff related headwinds which were described as a "one-time adjustment”, and continued investment in R&D and S&M.
- ARR also missed my expectations, but still accelerated growth YoY to 41.5% from 39.2% in Q2.
- RPO (39% YoY) and NRR (124%) stayed strong.
- Gross margin story due to the favorable mix-shift toward high-margin software remains intact as Software and Services revenue continued to accelerate YoY to 41.3%, up from 39.1% in Q2 and 38.7% in Q1 and despite that software margin itself dipped slightly due to professional services timing.
Axon reported Fiscal Q3 2025 on 11/4/25. Axon’s third-quarter results presented a complex narrative where near-term profitability and revenue recognition missed my expectations, while long-term strategic momentum - evident in bookings, customer metrics, and ecosystem expansion - accelerated significantly: Bookings growing faster than revenue (39% YoY), ARR growth accelerating (41% YoY) and Software and Services accelerating to 41% YoY. So the business is not just growing; it is widening its moat through high retention (124% NRR) and successfully selling new AI and robotics layers to existing police and enterprise clients. On the other hand, both revenue of $711 million and adjusted EBITDA of $177 million fell short of my $725 million and $190 million targets, respectively. It’s worth pointing out here though that the market hates to see EPS misses. In this case, adjusted EPS came in at $1.17 versus analyst expectations of $1.52-$1.63. The related dip in EBITDA margin should remind us of the overall thesis. Management firmly attributed this dip to a growth investment narrative. The Axon thesis is not one of profitability growth, but primarily one of revenue growth, as the multiple S-curves the company is stacking on top of each other will translate into profitability way down the road.
That said, the $13 million EBITDA shortfall from my expectation was a central theme of the earnings call. Management provided two clear drivers for the margin compression. First, “global tariff-related impacts” on the Connected Devices segment were cited as a primary headwind, which the CFO described as a "one-time adjustment”, suggesting this pressure should ease. Second, and more strategically, the margin dip was a deliberate result of “continued investment in R&D” and “scaling up investment in our products and sales team.” This narrative was heavily reinforced by the quarter’s major strategic announcements, including the acquisitions of Prepared and Carbyne to create the “Axon 911” platform, a move that expands their TAM by $5 billion and signals a clear intention to sacrifice some near-term margin to fuel long-term, high-margin software growth.
The $69 million in net new ARR, a clear miss against my $95 million goal, does initially appear soft. However, management’s commentary shifted the focus to the underlying bookings and long-term customer adoption, which points to lumpiness. This was strongly supported by the $11.4 billion in future contracted bookings (RPO), which was effectively in line with my $11.5 billion estimate, and a robust 124% net revenue retention (NRR). Management highlighted that the “AI Era Plan” is their “fastest-booked Axon software product to date” and that they closed multiple top-10 deals exceeding the $600 per user per month threshold. This suggests that while the recognition of new ARR may be lumpy, the underlying business of selling more software to more users at higher price points is accelerating, not slowing.
Regarding guidance there was no specific commentary to suggest they intend to make up the Q3 revenue delta in Q4. However, they did raise the full-year 2025 revenue outlook to approximately $2.74 billion, a 0.4% increase from their prior guidance. This raise, coupled with the massive RPO and narrative wins in new segments like Corrections (bookings up 2x YoY) and International (a new 9-figure deal in Europe), implies a high degree of confidence that any Q3 “shortfall” was a matter of timing and deal-to-revenue conversion, not a fundamental weakening in demand. With that, I now interpret management’s official Q4 forecast of $750 million to $755 million to come in at around $768 million (8.0% QoQ, 33.5% YoY). The intricate gross margin story - with overall margin expanding sequentially to 62.7% due to the favorable mix-shift toward high-margin software - remains part of the core thesis, even as that software margin itself dipped slightly due to professional services timing. It was great to see, however, that Software and Services revenue continued to accelerate YoY to 41.3%, up from 39.1% in Q2 and 38.7% in Q1.
Overview of how Axon performed versus my prior expectations:
- Revenue expectation: $725M (8.5% QoQ, 33.2% YoY), assuming similar QoQ growth as last Q3.
→ $711M (6.3% QoQ, 30.6% YoY). - Q4 revenue expectation: $765M (5.5% QoQ, 33% YoY), assuming similar QoQ growth as last Q4
→ Their Q4 guide was for revenue of $750M to $755M (5.9% QoQ, 30.8% YoY) at the midpoint), which I now interpret as $768M (8.0% QoQ, 33.5% YoY), and making up for the revenue shortfall this Q. Keep in mind that Axon’s metrics can be much more lumpy than typical SaaS. - I would like to see around $95M net new ARR.
→ $69M net new ARR. - I would like to see RPO around $11.5b.
→ Future contracted bookings (RPO) was $11.4b. - I would like to see NRR around 123-124%.
→ NRR was 124%. - I would like to see gross margin greater than 60.4%.
→ Gross margin expanded QoQ to 62.7% up from 60.4%, but down from 63.2% last Q3. Even though adjusted software and services gross margin dropped to 76.8% from 78.9% in Q2, its significantly faster growth than Connected Devices (52.1%) resulted in the overall QoQ margin expansion. - I would like to see adjusted EBITDA around $190M.
→ adjusted EBITDA was $177M with a 24.9% margin, down from 25.7% in Q2 and 26.7% last Q3. - Detailed thoughts: Ben’s Portfolio update end of August 2025
Monday:
Key insights:
- Monday missed on both my top-line revenue growth expectations and Q4 outlook and also down-revised their FY top-line guide.
- Customer metrics on the other hand were very strong and new product traction (now accounting for over 10% of ARR) show that the core thesis is intact.
- Monday’s push up-market ($500k+ customers segment growing 73% YoY) creates a temporary lag between strong bookings and reported top-line growth as large deals are being booked but will be recognized as revenue in future quarters (RPO growth accelerating to 36.3% YoY from 34.9% YoY in Q2).
- Profitability was as expected, displaying operational leverage as total operating expenses grew 21.5% YoY, well below revenue growth of 26.2%. Management, however, pivoted back to S&M spending from last quarter’s focus on R&D spending.
Monday reported fiscal Q3 2025 on 10/10/25. Their Q3 results presented a complex and diverging narrative. While the quarter itself was a picture of operational strength and disciplined execution, a cautious Q4 forecast combined with a down-revised FY top-line guide overshadowed positive standouts like large customer adds. Revenue for Q3 came in at $316.9 million, representing a 26.2% year-over-year increase. This marked a 1.6% beat against their guidance, which was the slimmest beat in the company’s public history. This, combined with a raw sequential revenue increase of $17.9 million (missing my $20.5 million target), resulted in a slight deceleration from Q3’ 26.6% YoY revenue growth. Q4 guidance of $328 million to $330 million fell short of my target for the guide of $332 million and a mixed bag of secondary growth metrics results raises concerns about near-term growth.
Despite the top-line guide, the underlying customer metrics and key performance indicators were relatively strong and beating my expectations in several crucial areas. The company’s strategic push upmarket was the clearest success story. They reported 63,075 customers in the 10+ user cohort, a sequential net add of 1,272 (up from 61,803 in Q2), which comfortably surpassed my expectation of ~1,140 net adds. This strength was even more pronounced in higher-value cohorts: the $50k+ cohort grew by 291 net adds (beating my 222 target) and the $100k+ cohort grew by 131 net adds (beating my 103 target). Underscoring this upmarket momentum, Monday introduced a new $500k+ ARR cohort, which already has 78 customers and is growing at an impressive 73.3% year-over-year.
The product expansion narrative also gained significant traction, providing a significant update. While specific account numbers for CRM, Dev, and Service were not broken out, management delivered a major milestone: new products now account for over 10% of total ARR, achieving the company’s full-year 2025 goal ahead of schedule. This validates the multi-product strategy, with the new Monday Campaigns product, launched in September, already securing over 200 customers. This success, however, helps explain the cautious Q4 guidance. Management commentary indicated that the strategic focus on larger, multi-product enterprise deals inherently involves longer and more complex sales cycles. This can delay the conversion of pipeline into recognized revenue, creating a temporary lag between strong bookings and reported top-line growth. The key metric giving credence to this delay was the introduction of Remaining Performance Obligations (RPO), which grew 36.3% year-over-year to $747 million. This RPO growth not only accelerated from Q2 (34.9%) but is also growing significantly faster than revenue (26.2%). This metric strongly suggests that the pipeline is healthy and that the “longer sales cycle” narrative is credible, as large deals are being booked but will be recognized as revenue in future quarters.
Finally, the quarter’s profitability metrics came in largely as I had expected although under the hood it appears there was a strategic pivot happening in Q3: In my last recap, I noted that Q2 margins contracted specifically because of a spike in R&D investment to 19.8% of revenue, presumably to fund critical AI development. The logical assumption was that this represented a new, elevated baseline. However, Q3 results show a sharp reversal: R&D expenses were pulled back significantly to 18.2% of revenue, dropping in absolute terms as well (from $59.2M to $57.8M). This R&D pullback was not a simple cost-saving measure; it directly funded a re-acceleration in Sales & Marketing, which expanded to 47.9% of revenue (up from 46.6% in Q2). This context is critical: the solid 15.0% operating margin ($47.5 million) and 19.5% net margin were achieved not despite high R&D, but by tactically reallocating capital away from R&D and toward S&M. This trade-off strongly supports the narrative that management is prioritizing the go-to-market motion for its upmarket and multi-product push, which requires a heavier sales lift. Their CFO confirmed this shift: the company is actively “rebalancing investments towards higher ROI areas, such as direct sales and new products.” Here’s a quick breakdown of how they explained this move:
- It was a strategic choice: This was not a sign of R&D weakness but a conscious decision to fuel the go-to-market engine. The increase in Sales & Marketing (S&M) spend was a direct investment in direct sales capabilities.
- It supports the upmarket push: Management explicitly linked this reallocation to their strategy of pursuing larger enterprise accounts and multi-product deals. These larger deals inherently have longer sales cycles, which requires a heavier and more sustained investment in S&M to convert the pipeline.
- It could also explain the smaller beat on revenue: Leadership even tied this rebalancing to the timing effects that led to the lower-than-historic revenue beat. The investment is being made now, but the associated revenue from these longer-cycle deals will be recognized in future quarters.
The story of operating leverage, however, remains firmly intact - total operating expenses grew 21.5% YoY, well below revenue growth of 26.2% - but the source of that leverage this quarter was a deliberate throttle on innovation spending to fuel the sales engine. This, combined with solid cash generation (a 29.1% FCF margin at $92.3M), shows a leadership team actively managing trade-offs to balance profitability with its immediate enterprise growth ambitions.
So, in essence, the Q3 story was one of short-term guidance caution driven by a successful - but temporarily lagging - strategic shift to more significant, long-term enterprise deals, as evidenced by the accelerating RPO and very solid high-value customer additions.
This is how Monday performed versus my prior expectations:
- Revenue expectation: $319M (6.85% QoQ, 27.1% YoY), implying a 2.4% beat.
→ $317M (6.0% QoQ, 26.2% YoY), a 1.6% beat (lowest quarterly revenue guidance beat in their history). - Q4 new revenue guide: $332M (4% QoQ, 24% YoY) which I would interpret as $340M (6.5% QoQ, 27% YoY).
→ $329M (3.8% QoQ, 22.8% YoY), which I now interpret as $337.6M (6.5% QoQ, 26% YoY), implying a 2.6% beat. - I would like to see raw sequential revenue increase around $20.5M. (It then should be in-between $20-21M for Q4).
→ $17.8M raw sequential revenue increase. I am still expecting $20-21M for Q4. - I would like to see more than 62950 customers with 10+ users (~1140 net adds), around 3924 customers in the $50k+ cohort (222 net adds) and around 1575 customers in the $100k+ cohort (103 net adds).
→ 63750 customers with 10+ users (1947 adds), 3993 customers in the $50k+ cohort (291 net adds) and 1603 customers in the $100k+ cohort (131 net adds). They also introduced a new $500k+ cohort with 78 customers currently, growing 73.3% YoY and 10 net adds in Q3. - I would like to see CRM accounts growth around 10% QoQ (to 36827, 3348 net adds).
→ numbers not given. - I would like to see Dev accounts growth around 8.6% QoQ (to 5073, 400 net adds).
→ numbers not given. - I would like to see Service accounts growth around 35% QoQ (to 1361, 353 net adds).
→ numbers not given. They did launch Monday Campaigns in September with currently over 200 customers. - I would like to see that NRR greater or equal to 112%, NRR10+ >=115%, NRR50k >=116% and NRR100k >=117%.
→ total NRR was 111%, NRR10+ was 115%, NRR50k+ was 116% and NRR100k+ was 117%. - I would like to see operating margin around 14.7% (~$47M operating income), net margin around 19% and FCF margin around 30% with FCF of around $96M.
→ OM was 15.0% (47.5M operating income, with operating expenses again growing slower than revenue at 21.5% YoY vs. 26.2% YoY revenue growth), NM was 19.5% and FCF margin was 29.1% with $92.3M free cash flow.
- They introduced RPO as a new metric. RPO YoY growth accelerated to 36.3% (growing faster than revenue), up from 34.9% in Q2, but down from 42% last Q3. - Thoughts from previous Q: Ben’s Portfolio update end of August 2025
Nvidia:
Key insights:
- Nvidia delivered a blowout quarter with YoY revenue growth acceleration from an insanely high base that will likely continue through Q4 (YoY revenue growth: Q2: 55.6% → Q3: 62.5% → Q4: 72%?).
- Management cited visibility into $500B+ of Blackwell/Rubin demand through calendar year 2026, which sets the floor for revenue growth in FY27 to 40% YoY.
- Entered a strategic 10GW partnership with OpenAI (for reference, 10GW is roughly equivalent to the total output of ten contemporary nuclear power plants).
- China has been completely de-risked.
- Gross margin expected to continue to grow and guided for 75% in Q4.
- Beyond FY27, as Blackwell deployments finish, hyperscalers must immediately layer in Rubin to reduce their OpEx (energy costs) ensuring there will be no cyclical drop after Blackwell.
- Physical AI, i.e. robots, autonomous factories, and heavy industry might be the next big wave creating new demand for compute.
Nvidia reported Fiscal Q3 2026 on 11/19/25 and has delivered a blowout quarter that significantly exceeded my expectations, characterized by a decisive transition from the anticipation phase of the Blackwell cycle to full-scale realization. While I anticipated revenue of $55350M (18.4% QoQ, 58% YoY), the company delivered $57006M (22% QoQ, 62.5% YoY), a 5.6% beat of their guidance and in addition to that the forward guidance has been reset aggressively higher, which I now interpret as $67600M (18.6% QoQ, 71.9% YoY), further accelerating YoY revenue growth.
The narrative has shifted from the awakening of demand seen in FY25 to a productive cycle of execution in FY26, with the Blackwell platform now the primary engine. Management cited visibility into $500B+ of Blackwell/Rubin demand through calendar 2026: “We currently have visibility to a $0.5 trillion in Blackwell and Rubin revenue from the start of this year through the end of calendar year 2026”. Given that their FY starts end of January, I estimate that they will have realized $228B of the $500B+ by the end of this FY. So that leaves $272B+ on the table through the end of calendar 2026 - or almost $300B for FY27, which would correspond to 40% YoY revenue growth as the minimum!
Looking back it is eye opening how the investing narrative has evolved from FY25 to FY26:
In FY25 sovereign AI was described as a concept that had just awakened, with nations realizing the need for domestic infrastructure. This has converted into hard revenue in FY26. Nvidia announced specific massive deployments (e.g., cloud leaders in Japan, India, Indonesia) and a strategic 10GW partnership with OpenAI. So the awakening has turned into a global construction race, where the FY25 narrative focus was on managing the tail-end of Hopper demand while building anticipation for Blackwell. Now, Blackwell is no longer a future promise; it is a reality. CEO Jensen Huang confirmed that “Blackwell sales are off the charts” and that “cloud GPUs are sold out.” The company explicitly stated that Blackwell Ultra is now the leading architecture across all customer categories.
In addition, China has been completely de-risked: While uncertainty around export controls and the H20 chip was a major headwind/topic of concern in FY25, the China risk has been effectively neutralized as a primary narrative driver. H20 revenue was described as “insignificant” in Q3, yet the company still smashed records, proving it can grow explosively without reliance on the Chinese market. Now, despite the Blackwell ramp, Hopper demand remains sustained. Management noted that the H100/H200 platform is still being heavily utilized for inference, mitigating fears of a drop-off in legacy sales before the new chips arrived. Other than that, it is noteworthy that Networking is evolving from a support accessory to a core revenue pillar. It is growing faster than compute itself on a YoY basis, driven by the complexity of connecting massive GPU clusters (like the 100K GPU clusters now being built).
In terms of gross margins, the YoY dip to ~73.5% was largely anticipated due to the costs associated with ramping the complex Blackwell supply chain (which involves more components and intricate packaging) and the return to 75% gross margin guidance for Q4 is a bullish signal that manufacturing yields have stabilized quickly.
The outlook beyond FY2027 (calendar 2027-2028) marks a fundamental shift in Nvidia’s narrative. If FY26/27 is about building the factory (Infrastructure), FY28+ is about automating the world (Industrial AI). While Blackwell is the training engine, Rubin is being positioned as the optimizing engine. Management confirmed Rubin is already in fabs and will be volume-ready by 2H 2026 (Fiscal Q3/Q4 FY27). This matters because Rubin is not just faster; it is designed to be power-efficient enough to allow AI factories to scale without blowing up the world’s energy grid. This addresses the number one bottleneck for hyperscalers: electricity, not silicon. This also ensures there is no cyclical drop after Blackwell. As Blackwell deployments finish, hyperscalers must immediately layer in Rubin to reduce their OpEx (energy costs).
I think what will happen next is to be seen, but I found it interesting that management explicitly linked the future of AI to Physical AI, i.e. robots, autonomous factories, and heavy industry. The pivot here would be that while today, ~80% of revenue comes from Cloud Service Providers (Amazon, Google, etc.), by FY2028, Nvidia expects a massive mix-shift toward Enterprise & Industrial customers. For example, the 10GW OpenAI partnership is a prototype for private AI grids. Expect nations (Sovereign AI) and industrial giants (Siemens, Foxconn, Toyota) to become the new “hyperscalers,” buying billion-dollar clusters directly to run automated supply chains.
How does that future affect Nvidia’s TAM? CFO Colette Kress reiterated the view that $3-4 Trillion in legacy data center infrastructure will be replaced by accelerated compute. The interpretation of this is that if Nvidia maintains even 60% market share (conservative vs. current ~90%) of a ~$3T overhaul, my conservative napkin math tells me that the mathematical floor for their annual revenue sits comfortably above $300B per year for the next 4-5 years (assuming a 10-year super-cycle where the $3-4 Trillion in legacy data center infrastructure will be replaced by accelerated compute, so $500b/year and assuming Nvidia catches 60% of that). Of course, this is super rough, somewhat assumes that the frenzied buying of FY25/26 slows down to a steady “maintenance” pace and doesn’t even include recurring revenue from Nvidia AI Enterprise software, which is growing rapidly. If the Industrial AI phase (robotics) takes off, the TAM could expand further, making $300B the basement, not the ground floor. How much higher than $300B this figure will be is anyone’s guess, but this is just to show that the cyclical bust thesis relies on AI demand drying up; instead, Nvidia suggests AI demand will be broadening from chatbots to physics.
In summary, the Q3 results confirm we are in what people call a super-cycle. The orders visibility of $500B+ implies that Nvidia is effectively booked out for the next 18-24 months. The risk is no longer demand; it is purely execution (supply and energy availability).
This is how Nvidia performed versus my prior expectations:
- Revenue expectation: $55350M (18.4% QoQ, 58% YoY), implying a 2.5% beat; they beat a year ago’s Q1 guide by 8.5%, Q2 guide by 7.3%, Q3 guide by 7.9% and Q4 guide by 4.9%. Then last Q1 guide by only 2.5% and their Q2 guide by 3.9%.
→ $57006M (22% QoQ, 62.5% YoY), a 5.6% beat. - Q4 new revenue guide: $58394M (5.5% QoQ, 48% YoY) which I would interpret as $59778M (8% QoQ, 52% YoY), expecting slight YoY deceleration.
→ The new guide was $65000M (14% QoQ, 65.3% YoY), which I now interpret as $67600M (18.6% QoQ, 71.9% YoY), implying a 4.0% beat. - I would like to see GAAP gross margin above 73.3%.
→ GAAP gross margin was 73.4% - I would like to see non-GAAP gross margin above 73.5%.
→ non-GAAP gross margin was 73.6%
Datadog:
Key insights:
- Fantastic quarter exceeding almost all of my expectations by a significant margin.
- Concern about AI-native customers building their own monitoring solutions was effectively neutralized as segment is fastest growing, with broad adoption of AI monitoring products and extension of OpenAI contract.
- YoY revenue growth continued to accelerate slightly, even after the big jump in Q2 (Q1: 24.6% → Q2: 28.1% → Q3: 28.4% → Q4: 28.6%?).
- RPO growth exploded in Q3 (Q4: 23.4% → Q1: 33.5% → Q2: 35.8% → Q3: 53.3%) with cRPO growth in the low 50’s percent, up from the low 30’s in Q2!!!
- Very strong large customer adds of 210 versus my expectation of 116, multi-product customer adoption increased percentages in all cohorts, by even 2% in 4+, 6+ and 8+, versus the normal 0% to 1% increase and strong NRR of about 120% show their land-and-expand motion is firing on all cylinders.
- Strong profitability with OM (23.4%), NM (28.0%), and FCFM (24.2%) all handily beating or meeting my targets.
Datadog reported Fiscal Q3 2025 on 11/6/25. This was yet another fantastic quarter, again exceeding almost all of my expectations by a significant margin, and, additionally effectively neutralizing the concern about AI-native customers building their own solutions causing a potential revenue shortfall for Datadog. The data showed this segment is, in fact, one of Datadog’s fastest-growing, with broad adoption of its new, specialized AI monitoring products and large customers signing long-term commitment deals:
Strong growth from AI-native customers:
- A top growth driver: Datadog reported that the “AI-native cohort” of customers was a primary driver of its better-than-expected results.
- Significant revenue share: This cohort now represents 12% of the company’s Q3 revenue, a significant increase from 6% in the same quarter last year.
- Broad-based acceleration: Management specifically highlighted that they saw “an acceleration of growth in our AI cohort in Q3 when excluding our largest customer.” This was a crucial detail, suggesting that even if the largest AI customer’s (widely assumed to be OpenAI) spending is lumpy, the rest of the AI-native segment is growing broadly and rapidly. And specifically regarding fears of OpenAI churning: “we extended the contract of our largest AI native customer .”
Product innovation for AI workloads:
Instead of losing customers who are building their own tools, Datadog demonstrated that it is actively and successfully building the specialized tools these AI-native companies need.
- Rapid adoption: The company reported that the number of LLM “spans” (a unit of data) being sent to its platform has “more than quadrupled” in the past month.
- New AI-specific products: Datadog heavily featured its new AI-focused products, including LLM Observability, LLM experiments, and Bits AI agents, all of which are seeing high customer interest and adoption. Over 5,000 customers are already sending AI-related data to the platform (they currently have about 32000 customers).
Large customer commitments:
Analyst questions about the structure of large AI customer contracts revealed that these companies are making long-term commitments to Datadog, not planning to leave.
- High-commitment contracts: CEO Olivier Pomel explained that recent large AI customer deals are structured as “high-commitment contracts.”
- Betting on usage growth: These contracts are designed for growth driven by increased usage of the Datadog platform, signaling that these large AI companies are integrating Datadog as a long-term partner for their scaling needs.
Zooming out further and looking beyond AI native customers, I was happy to see Datadog accelerating in most critical forward-looking metrics. While the headline revenue of $886M was a solid beat against my $883M expectation, the underlying drivers of this performance signaled a powerful update to the company’s growth narrative. The quarter was defined by a massive surge in long-term customer commitments and a dramatic influx of high-value enterprise and AI-native customers, painting a much more bullish picture for 2026 than the Q4 guide might imply on its own.
The most significant upside surprise, by far, was the explosion in Remaining Performance Obligations (RPO). My prior expectation of $2.55B (40% YoY) was already strong, but the actual result of $2.79B, growing an incredible 14.8% sequentially and 53.3% YoY, was a blowout. The real story here is the cRPO (current RPO) growth, which accelerated to 52.5% YoY, up from 35.8% in Q2. cRPO, what Datadog management expects to become revenue in the next 12 months, while not directly correlating with revenue growth due to the nature of the consumption business, is now growing at nearly double the pace of current revenue:
Earnings commentary directly attributed this to a significant increase in large, high-commitment contracts from new logos and existing customers. This RPO acceleration is the clearest evidence that the macro-optimization headwinds of the past year are over, and enterprises are now signing larger, longer-term deals.
This RPO surge was fueled by an equally impressive beat in high-value customer acquisition. While total customer growth was slightly lighter than my expectation (600 added vs. 650 expected), this was completely overshadowed by the massive beat in the $100k+ cohort, which added 210 new customers - nearly double my 116 estimate. This is the “why” behind the RPO beat and signals a sharp acceleration in enterprise momentum. Commentary confirmed this was driven by two key forces: first, the AI-native cohort continues to be a major growth engine, now representing 12% of total revenue (up from 6% YoY) and scaling its commitments. Second, management cited a “notable inflection” in usage from non-AI customers, which saw its “highest sequential usage growth in 12 quarters.” This dual-engine growth - a broad-based enterprise recovery combined with AI-driven hypergrowth - is a powerful new narrative.
The land-and-expand model was also in full effect, with multi-product adoption exceeding my “stable” expectations. The fact that all cohorts (2+, 4+, 6+, and 8+) saw their adoption percentages increase (from 83%, 52%, 29% and 14% to 84%, 54%, 31%, and 16% respectively) is a testament to the platform’s stickiness and the success of newer product lines. This was supported by commentary that the Security platform, in particular, saw its ARR growth accelerate to the mid-50s YoY, up from the mid-40s last quarter, successfully pulling customers deeper into the ecosystem. This successful cross-selling, combined with the stable ~120% NRR, is the mechanism that graduates customers into the $100k+ cohort and drives efficient growth.
Finally, the company delivered this acceleration while also flexing its operational leverage, with OM (23.4%), NM (28.0%), and FCFM (24.2%) all handily beating or meeting my targets. This efficient execution only adds to the bullish story. Looking forward, the official Q4 guide to $915M is clearly conservative in light of the quarter’s bookings. The underlying momentum seen in the RPO and cRPO in particular and $100k+ customer blowouts strongly supports my new interpretation of a ~$949M revenue number for Q4 and continued growth strength into FY26, as the business continues firing on all cylinders.
Overview of how Datadog performed versus my prior expectations:
- Revenue expectation: $883M (6.8% QoQ, 28% YoY), implying a 4.0% beat.
→ $886M (7.1% QoQ, 28.4% YoY), a 4.3% beat. - Q4 new revenue guide: $912M (3.3% QoQ, 24% YoY) which I would interpret as $943M (6.8% QoQ, 28% YoY) expecting YoY growth will stay close to 28%.
→ $915M (3.3% QoQ, 24% YoY), which I now interpret as $949M (7.1% QoQ, 28.6% YoY), implying a 3.7% beat. - My Q3 revenue expectation implies about $56M raw sequential revenue increase (up from $45M last Q3).
→ $59M raw sequential revenue increase. - I would like to see RPO at around $2.55b (5% QoQ, 40% YoY growth).
→ RPO was $2.79b, growing an incredible 14.8% QoQ and 53.3% YoY. Even more incredibly, cRPO grew around 52.5% YoY, up from 35.8% in Q2 and 25.5% last Q3. That is much faster growth than revenue at currently 28%. - I would like to see QoQ customer growth around 2.1% (~650 new) and for the $100k+ cohort, around 3% QoQ (~116 new).
→ total customer count increased by 1.9% QoQ (600 new) and $100k+ customer count increased by 5.5% (210 new). - I would like to see continued multi-product adoption progress with 2+, 4+, 6+ and 8+ products cohort percentages to stay stable at 83%, 52%, 29% and 14%.
→ the cohort percentages grew to 84%, 54%, 31% and 16%. - I would like to see NRR around 120%.
→ NRR was around 120%. - I would like to see OM ~20%, NM ~25%, FCFM ~25%.
→ OM was 23.4%, NM was 28.0% and FCF margin was 24.2%. - Thoughts from previous Q: Ben’s Portfolio update end of August 2025
Wrap up
Overall our companies that reported earnings so far are firing on all cylinders, besides Monday and Axon, which had reports that were on the weaker side, but their investment theses are still intact. Nvidia, Cloudflare, Datadog had really amazing results. Especially for the latter two I see a sea change in the investment thesis as both companies successfully start capitalizing on new S-curves that are based on AI, leading to acceleration in revenue growth and likely durability for many quarter’s to come. We also made the decision to exit The TradeDesk after several disappointing earnings reports and reallocated to two new starter positions in AppLovin and Astra Labs.
I am wishing you all a great December!
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