Moritz’ Portfolio Review – December (Q4) 2023

Hi team,

Happy New Year to you all! This review is longer than usual because I wanted to spend some time reflecting on the entire year, all my companies, and my thoughts on what to expect in 2024 and the future. So, grab a coffee, and let’s dive in!

A recap of my portfolio in 2023

January: After a tough 2022, my portfolio kicked off with a solid 10% gain, which is encouraging. A reminder that this challenging period will eventually come to an end.

February: Stayed cautious about certain investments, expecting market uncertainties. Had a volatile month with a peak of 25% year-to-date (YTD) return, but ended at 14%.

March: Began investing in toilet paper and dealt with challenges from low guidance (NET), outages (DDOG), and banking concerns (Silicon Valley Bank). Categorized challenges into noise and controllable factors. Wondered why some shifted away from SaaS. Back to 6% YTD.

April: Portfolio experienced a strong decline, resulting in a -4% return for the year. Emphasized the long-term resilience of tech and software. Encouraged others to maintain perspective during tough times.

May: Market overcame pessimism, leading to a significant 32% increase in YTD returns. It was my best month ever, with a 36.4% gain in just one month, which is amazing! Shared personal struggles and stressed the importance of taking a step back to assess the bigger picture.

June: Recognized the positive impact of monthly portfolio reviews, promoting regular performance tracking and in-depth analysis. Encouraged investors to start writing their own portfolio reviews. YTD: 30%.

July: Discussed the importance of holding investments for a longer period, emphasizing a thoughtful, data-driven approach. Mentioned my goal of achieving a 20% annual return over the long term, dealing with challenges, and focusing on high-quality growth companies. Portfolio quietly increased to 35% YTD.

August: Finetuned investing philosophy to prioritize people and leadership in investment decisions. Announced a gradual shift to a different approach over the coming months. 32% YTD.

September: Emphasized the importance of adapting to the changing investing landscape. Switched to a quarterly portfolio review schedule with fewer adjustments, focusing on sustainable, long-term growth. Portfolio remained around the 30s with a 34% YTD gain.

October to December (Q4): After a challenging year in 2022, my portfolio in 2023 ended with 55% YTD. The events of 2023 have once again shown how hard it is to predict the market, which reinforces my choice to always remain fully invested.

Portfolio performance

Timestamp: 12/31/2023

Portfolio allocations

Changes to the portfolio

Positions I trimmed

Bill - BILL

While I find Bill’s large total addressable market (TAM) (over 70 million SMBs worldwide generating $125 trillion of B2B payment volume) and the potential for international expansion intriguing, it is also one of the most challenging for me to understand.

After BILL’s last earnings report, the stock was punished by the market with a 35% drop. Management reduced the fiscal year (FY) guidance due to the challenging macroeconomic environment. Luckily, I had already reduced my position to about 3% before the earnings report, as I was already questioning whether I still saw myself as a long-term shareholder.

I’ve always felt uneasy about fintech stocks, and after the experience with BILL, I might establish a rule to avoid any fintech businesses in the future. There’s just so much disruption and volatility happening in that sector.

However, I didn’t sell any shares after the drop because I believed it went into undervalued territory and still is as of writing this in November. With over 4,000 quarterly expected new customer additions for BILL standalone, the growth of our customer base is still on track. Additionally, the traction of Divvy, now called “Spend & Expend,” looks promising.

Potential catalysts that could send the stock price back up again include:

  • In an improving environment helping TPV-, customer and revenue growth

  • Cross-selling opportunity from standalone customers to Spend & Expend

  • Increase in subscription pricing in FY2025 with BofA

  • Success in international expansion

  • Smart acquisitions

I might hold onto it for a while, but it sits on the bench.

Positions I sold

Nvidia - NVDA

As discussed in an older portfolio review, I considered selling my shares in NVDA, and I ultimately did sell my small position at $462 in early October and used the cash to add to ANET and EVO.

I have great admiration for the company—its leadership, high insider ownership, profitability, and rapid growth. It’s truly a dream investment. However, I have two concerns:

  1. The company’s current market cap is $1.2 trillion, and I question whether it can grow to over $8 trillion in the next 10 years to achieve a 20% compound annual growth rate (CAGR) in stock price appreciation. Even for a 15% CAGR, it would still require a $5 trillion market cap. To put this in perspective, Microsoft and Apple currently have market caps of around $3 trillion. While it’s not entirely unrealistic, there are other alternatives where a 20% CAGR over the next decade seems much more attainable.

  2. Valuation: Looking at the forward price-to-earnings (P/E) ratio of just 24 as of December 15th, NVDA doesn’t appear to be expensive. However, the question is: How sustainable is this growth? Once demand slows down, either the forward P/E ratio will likely increase or the stock price will decrease a lot.

Unlike a typical Software as a Service (SaaS) business with recurring revenues, Nvidia is a cyclical business and still working on developing and ramping up subscription offerings.

Data center revenue exploded to 80% of total revenues during the recent quarters. I have doubts that the recurring revenue segment will fully compensate for the significant drop in growth once most companies have already stocked up on Nvidia’s chips. While it may happen in the future, I anticipate a larger decrease in the stock price once chip demand eases.

Additionally, Nvidia has gained significant attention and is a popular choice among all—retail and institutional—investors at a level that I don’t even see for a SaaS darling like Crowdstrike. This popularity reduces the likelihood of long-term outperformance. (It’s also one of the reasons why I’m not invested in Tesla).

While I recognize the true innovation and disruption that AI brings and believe it will have a significant impact on our lives in the future, I also try to remain grounded. The usage of AI, such as ChatGPT, and the immediate demand for chips has skyrocketed, but I expect this growth to level out. Similar to the early 2000s with the internet, people believed everything had changed. And while it did, it took another decade to truly materialize.

I view AI through a similar lens. It’s promising, in its early stages, and here to stay. However, progress will likely occur at a slower pace than expected. The longer it takes, the more competition will emerge, and it becomes harder to argue that Nvidia is the sole vendor in the market.

Lastly, I keep thinking about Nvidia’s Infiniband technology versus Ethernet. I’m still trying to grasp this topic fully, but I feel that Ethernet might have the advantage when it comes to the best network infrastructure for low inference, which is crucial for AI. Other players in the industry might have an edge in this area, as customers may prefer Ethernet in the future.

For additional context here are the sequential (QoQ) trends:

Total revenue:
Q2 88% → Q3 34% → Q4e 25%(?)

Datacenter revenue:
Q2 140% → Q3 41% → Q4e 28%(?)

I may underestimate the durable growth driven by data center demand, and Nvidia might surprise us in the next decade. Actually, I’m reading so many positive pieces about Nvidia’s outlook, perhaps I should just get on board again, instead of writing such a long text about a 1% position I sold.

However, since it was only a “tryout”, I’ve decided not to be on board for now. Investing involves creating assumptions and hypotheses, and then making decisions based on that information. Given my personal investment strategy, I believe there are many more and “easier” opportunities available elsewhere:

Considering the many other opportunities ahead of us that will greatly benefit from AI in the upcoming lifecycle stages of innovation, including the data infrastructure- and application layer that follows the hardware layer contributed by companies like Nvidia, AMD and Arista Networks, you could think of ANET as my “NVDA play,” but with a market cap that is 17 times smaller. And the upcoming opportunities in the data infrastructure and application layers include DDOG, S, CRWD, ZS, NET, SNOW, MDB, (CFLT), and perhaps even MNDY and TTD.

So here I am, watching Nvidia’s stock price skyrocket to the moon—but without me:

Untitled (17)

Confluent - CFLT

Confluent experienced a stock price crash similar to SentinelOne and BILL after lowering their FY guide by 4% from $772M to $769M. I decided to sell a week after earnings. However, I held on to Bill and S for different reasons:

In the case of SentinelOne, I concluded that there was a more immediate upside and disagreed with the exaggerated stock price drop. I believe the business was still on track. As for BILL, I see the investment thesis still unfolding, although I’m considering selling it due to its complexity.

Now, regarding Confluent, there are several reasons why I have decided to move to the sidelines for now:

  1. Confluent still needs to prove if they are truly as mission-critical as they claim. One customer decided to transition to an on-premises solution, which, although not specific to Confluent as management stated, still doesn’t help my confidence.

  2. Management has frequently changed its go-to-market (GTM) strategy. The most recent change is their move to a model where customers pay based on how much they use the product (which I think is a good move). On one hand, I like how quickly they adapt, but on the other hand, I wonder if they should have had a clear strategy from the start.

  3. RPO, cRPO, and customer growth (also large ones) are slowing down. If these metrics had been strong, I would have maintained my position.

  4. Revenue growth is in the low 30s, and next quarter it will likely be in the high 20s. This is significantly lower compared to most other businesses, considering their smaller quarterly revenue base:

    1. CFLT - $200M (32% YoY)
    2. BILL - $305M (33% YoY)
    3. S - $164M (42% YoY)
  5. Lastly, there’s still a high amount of dilution.

While some of these issues overlap with S or BILL, it’s the combination of factors mentioned above that influences my actions differently here.

On a positive note, the trend towards profitability continues to be very good. The non-GAAP net income ($6.3M) was positive for the second quarter in a row, making up 3.2% of revenues. The operating cash flow margin improved to -4.6% compared to -27.5% last year, and the free cash flow shows a similar upward trend. The dollar-based net retention rate (DBNRR) remained high at 130% (although it would be lower if their old method was used).

However, my investment in CFLT was only a small 2% starter position, so I chose to sell it quickly instead of turning it into a full position and diving deeper into the business. Even though I sold CFLT at around $17 and the stock has continued to rise (45% since I sold) as of December 15th, I’m okay with it because I invested most of the cash into TTD and some other positions that have increased by about 25% since then. So not everything is “lost,” and I still stand by my decision.

I will remain open to changing my mind, which might make me consider investing in CFLT again. Some catalysts that could cause this are:

  • Full ramp-up of the consumption-based model

  • Introduction of more products (i.e. Flink)

  • Achieving more certifications like FedRAMP

  • Use cases driven by “AI” which increases the amount of data processed

Positions I increased

Evolution Gaming - EVO

Looking at Evolution Gaming from a short-term perspective, its performance in Q3 2023 was unexciting, despite achieving a record-high 70% EBITDA margin, 60% operating profit margin, and 44% year-over-year growth for operating cash flow per share.

Revenue increased by 20% compared to the previous year, reaching €453M, driven by Evolution’s live segment and expansion in Asia. However, there was a 7% negative impact due to foreign exchange rates. Management highlighted the challenges caused by the slow opening of studios and the need to hire more staff quickly. These issues are expected to be resolved in the coming quarters.

Revenue growth of RNG (Random Number Generator = Slots) slowed down to a negative -1.9% year-over-year (YoY). It is uncertain whether RNG can achieve the originally planned double-digit growth as anticipated by management.

On a positive note, management announced a share buyback program of €400M shortly after earnings and has already executed it three times, purchasing shares worth over €50M. Furthermore, they successfully launched their popular game show called “Crazy Time” in the US, which could potentially boost revenue growth in the US for this quarter.

I maintain a strong conviction in Evolution due to its experienced management, strong competitive position, large Total Addressable Market (estimated at $153 billion by 2030), high insider ownership (12%), high profitability, and zero debt. Several potential catalysts for growth are ahead, including:

  • Resolving studio capacity issues and increasing hiring

  • Achieving success in the RNG segment

  • Launching new successful game IPs in new markets (such as Crazy Time in the US) and driving international expansion

  • The EU market could provide potential upside by 2026 if iGaming regulations are relaxed

  • Making smart decisions regarding capital allocation, such as buybacks and reasonable acquisitions

  • Improvement of the macroeconomic environment

During the quarter, I took advantage of the declining stock price towards €80 multiple times to increase my position.

Arista Networks - ANET

Arista Networks specializes in providing cloud networking solutions, with a focus on high-performance, scalable, and programmable switches for data centers and cloud computing environments, and benefits from the hardware layer in AI. This is because AI workloads are becoming increasingly important among cloud users, there should be faster production upgrade cycles of data center switching products, and they are gaining market share with enterprise customers.

It’s a common misconception that they are “just a hardware company”. In reality, they are a software company with a gross margin of over 60%. Their software solutions improve their customers’ networks’ speed, security, efficiency, and manageability. They have a customer base of over 9000, with a Net Promoter Score (NPS) above 80. Some of their happy customers include Microsoft, Meta, Alphabet, Amazon, Apple, Alibaba, Baidu, and Tencent.

Another aspect that intrigues me is the significant insider ownership of 19% held by highly experienced and trustworthy leadership. They demonstrate strong fundamentals and share goals aligned with shareholders, such as avoiding dilution and executing smart capital allocation decisions like share buybacks. Additionally, Arista has a proven track record of achieving a high Return on Invested Capital (ROIC) of 29%, practices prudent mergers and acquisitions, adopts a long-term mindset, and maintains exceptional profitability with high margins.

And that trend continued in Q3’2023, which was an outstanding quarter:

  • $1.509 billion of revenue, an increase of 28% YoY

  • 62% GAAP gross margin

  • 46% GAAP-operating income margin

  • 40% operating cash flow margin

  • $1.72 GAAP net income per share, an increase of 52% YoY

Management expects to grow revenues at a low double-digit rate for years to come and anticipates taking half the time it took Arista to achieve the first $5B to reach $10B in revenues.

When I purchased my initial ANET shares, I did not even consider AI. However, I now believe that this trend will contribute to further growth in the upcoming years. It will also provide support that surpasses the cyclical nature of the business, especially regarding the switch upgrade cycles by the cloud titans. Due to this cyclical nature, CEO Jayshree Ullal usually advises investors to take a three-year perspective on earnings rather than solely focusing on a single quarter or year.

Arista’s strong customer base, which includes Microsoft and Meta, is both its biggest risk and its biggest advantage. These companies accounted for 42% of Arista’s total revenue in 2022. On one hand, this highlights the strength of Arista Networks’ offerings. On the other hand, it would be good to see Arista diversify its revenue sources further, and they are actively working towards that goal.

Potential catalysts include:

  • Upgrade cycles might come in faster than expected since AI requires high-speed networks
    • 800Gbps switches
    • 1.6Tbps switches
  • New solutions
    • Core: AI Spine, Enriched EOS stack, Network as a service
    • Adjacent markets: Routing & Campus, Transit WAN Routing, Edge as a Service
    • Software as a service: Deliver Zero Trust Networks (in collaboration with Zscaler), Improve Deployment Designs, Increase Visibility and Observability
  • International expansion (2023: 78.5% Americas)
  • Success in customer diversification (enterprise/campus)
  • Improving macro environment
  • Acquisitions that might be value-adding
  • AI driving needs to upgrade to Arista’s solutions or increase the existing footprint

Every time I check out all those facts about this boring networking company, I can’t resist throwing in a little more cash. (And when it comes to investing, I love the boring stuff. Thanks, Peter Lynch.)

The Trade Desk - TTD

Unlike BILL, The Trade Desk is a company that I find easy to follow. They recently reported a strong performance in the third quarter, with the following highlights:

  • Revenue of $493M, representing a 25% year-over-year increase.
  • Operating cash flow margin of 39% and a free cash flow margin of 37%.
  • GAAP net income of $39M.
  • The repurchase of $90M worth of shares in Q3’2023 to offset dilution.

However, the stock crashed by 30% after hours due to a weaker-than-expected Q4 outlook. In the second week of October, management observed temporary caution from some advertisers, particularly those in industries affected by recent strikes/auto industry. It is worth noting that BILL also reduced its outlook, likely due to similar external factors rather than poor execution, which influenced my decision to hold onto BILL for now.

Here’s why I am not concerned by the weaker-than-expected outlook:

  1. The Trade Desk continues to outperform its competition and the digital advertising industry: While the Trade Desk is growing 25% YoY, the industry is only growing at a 5 to 10% YoY rate, which means that the business is taking market share. Jeff Green anticipates that this outperformance will persist throughout 2024.

  2. The Trade Desk benefits from more deliberate ad spend: With tightening budgets, marketing agencies and brands are more deliberate about their ad spend and prefer to shift it towards advertising that is more ****flexible, and data-driven and offers superior ROI: The Trade Desk.

  3. Innovation coming from AI: The Trade Desk has been investing in meaningful AI use cases for years and expects many new opportunities where they can inject AI. The business owns significant value in data and insights.

  4. CTV continues to be the fastest-growing channel of their business and a key driver of overall omnichannel growth. Executives at every major streaming giant with both an ad-supported and an ad-free tier, (including Disney, Netflix, Paramount, Warner Bros Discovery and NBC Universal) say that total revenue per user is higher on the ad-supported plan than it is on the ad-free plan. That can be a significant catalyst going forward.

  5. Retail media has become one of their business’s fastest-growing areas, and they expect this to continue in 2024.

  6. Global expansion: The Trade Desk made significant and successful investments in its business outside the US over the last several years (e.g. in go-to-market strategy, CTV, retail media).

  7. Rapid adoption of The Trade Desk’s UID2 and EUID technology: UID2 and EUID is increasingly becoming the currency for relevant ads and personalized content that has been adopted by the infrastructure of the internet and nearly all of the biggest content companies in the world. Agencies want to have more control over their campaigns and data.

  8. Upcoming US political election. Since 2016, The Trade Desk has been a vital platform for leading political advertisers. In 2024, they expect to gain more share in this segment and they believe that spending will increase as the year progresses

  9. Profitable growth: TTD is one of the few high-growth technology companies that consistently generates strong adjusted EBITDA, positive GAAP earnings, and free cash flow that has steadily increased over the years.

  10. Decent insider ownership with a 10% stake held primarily by Jeff Green.

  11. A large TAM with an estimated size of $1 trillion.

I couldn’t resist and added to my position after the earnings report, buying more shares at around $60.

“We align our interest with our clients, we maintain our objectivity, and we innovate like hell.” Jeff Green during the Q3’2023 earnings call.

Commentary on my other companies

Crowdstrike - CRWD

CrowdStrike’s Q3 2023 earnings exceeded estimates with 35% revenue growth YoY, raising its FY revenue guide by 0.2% and indicating a further sequential acceleration next quarter (7.9% QoQ). Despite a challenging macro environment, net new ARR accelerated to $233.11M (13.7% QoQ), earning praise from analysts. Management expects net new ARR for the full year to be in line with modest growth, and double-digit YoY growth in the second half, translating to at least $243M net new ARR in Q4 (9.3% QoQ).

The forecast for operating income and net income was raised by 4% for the entire year. Subscription gross margin reached a record high of 80.3%, Free cash flow margin was 30%, and there was a first-time positive GAAP operating income of $3.2M. These factors led to a strong Rule of 66.

Module adoption rates remain strong, with 26% of customers now using 7+ modules (up from 24% last quarter). The gross retention rate remains high (>97%). Lastly, they achieved a significant milestone by surpassing $3B in ARR.

However, the Revenue Performance Obligations (RPO) were a bit weak, resulting in a sequential growth rate of only 2.8%. This can be attributed to the macro environment and budget scrutiny, even in the field of cybersecurity. DBNRR was slightly below the benchmark (<120%) in Q3, as the mix of net new ARR from new customers has exceeded expectations and continued to land bigger deals.

With $3.17b in cash, there is ample opportunity for strategic acquisitions. Crowdstrike estimates its TAM to be $100b by CY24 and $225b by CY28, which leaves plenty of room for future growth.

Overall, strong performance! I’m looking forward to the upcoming earnings report, especially the trend on RPO and the yearly update on customer numbers.

SentinelOne - S

SentinelOne’s Q3’2024 earnings report exceeded expectations, beating revenue estimates by 5.3% and showing a 10% increase QoQ. Management raised its FY revenue guide by 2%, projecting $625 million for FY24, a 48% YoY growth. Financial indicators, including gross margins and operating margins, have all been raised, showcasing a commitment to profitability.

Despite a challenging macro environment, they added 350 customers (3.1% QoQ), with a strategic focus on larger enterprises (+66) leading to a 6.6% QoQ increase.

Management anticipates positive Free Cash Flow (FCF) in H2 FY25, signaling a strong outlook for the next fiscal year. They **continue to gain market share against both next-gen and legacy providers.

Additionally, the net new ARR guide for the year has been raised to $200M, up from $195M. If they achieve this target, the net new ARR will be approximately $58M, which is the same as the net new ARR in Q4’23 (last year). I believe they are being cautious and anticipate that the net new ARR will be slightly higher than Q4’23, which was $58.21M.

Acknowledging the challenging threat landscape, SentinelOne plans to transition Singularity Cloud to CNAPP next year.

They announced a new Chief Revenue Officer (CRO) from Elastic.

Despite a high dilution at 5.7% YoY, the pace has slowed to 1.2% QoQ. Management maintained a confident tone, and analysts congratulated its performance. With a strong financial position of $1.1 billion in cash and no debt, SentinelOne is in a solid financial state.

It’s important to note that SentinelOne is not just focused on endpoint security. It’s a platform company that offers a growing range of additional cybersecurity products, such as data lake and cloud security.

In summary, SentinelOne’s Q3’2024 earnings report shows that they are doing well in the cybersecurity market. Their shares have gone up by 20% after-hours and almost 100% in the past 6 months. I’m happy that I kept my investment in S, as explained in my June portfolio review. SentinelOne is still on track to gain market share, along with Crowdstrike and other cybersec companies. For now, I’ve decided to continue being a shareholder.

Zscaler - ZS

What I love about Zscaler is that it is founder-led, with high insider ownership. Zscaler’s CEO, Jay Chaudhry, holds a 39% stake, which aligns management interests with shareholder value. This serves as a powerful incentive for Zscaler’s profitability and success. I believe this factor has contributed to the company’s successful development, as well as its strong Q1’2024 earnings:

Zscaler exceeded expectations with a revenue growth of 40% YoY and raised its FY revenue guide by 1.6%. Due to the challenging macro environment and a sales leadership change, Zscaler remains prudent with its outlook.

Free cash flow margin was an impressive 45.2%, thanks to strong collections from Q4 billings, including a $20M upfront deal in fiscal Q4’23. The combination of growth and cash flow led to an impressive rule of 80 in the quarter.

Customers generating more than $100k ARR came in a bit weak with only 3.8% sequential growth (99 net new adds), which is the lowest Q1 number of customer additions since 2022. Also, customers generating more than $1M ARR were weak with only 4% sequential growth (19 net new adds). Still, they added 14 new logos with more than $1M ARR, which was a Q1 record.

Additionally, 5 customers increased their spending to more than $1M ARR. 50% of new logo customers purchased multiple platform offerings: ZIA, ZPA, and ZXD. Management also announced a record new ACV in the Fed business with over 90% New ACV year-over-year in the Fed vertical.

My take on weaker customer growth: Management emphasized the importance of focusing on generating revenue from their existing customer base. I believe this is a smart move because the existing base provides the easiest and most frictionless way to increase revenue. Why struggle against macro headwinds and waste resources on pursuing large enterprise customers who are still tightening their belts, when there is so much (6x) untapped potential right in front of them?

After stumbling over an analysis of where cybersecurity sales occur for all cybersecurity companies (see image below), it is interesting to note that direct sales growth plummeted from Q4 2022 onwards, while channel growth (resellers, system integrators, and MSSPs) remained steady. This decrease aligns with the timing of weaker customer growth reported by Zscaler, SentinelOne, and other companies. It is attributed to changes in the overall economic environment.

Another interesting fact is that more than 90% of cybersecurity sales happen through partners. It’s not surprising that Zscaler invested in its channel strategy, including training its global sales team to better engage partners, building relationships with distributors, and expanding its channel team to support the growth of the business to $5 billion in ARR.

Similarly, Crowdstrike introduced the Accelerate Partner Program to cover its entire ecosystem, Falcon Complete for Service Providers to offer co-branded or white-labeled managed security services, CrowdStrike MarketPlace for integration partners, and the CrowdCredit reward program.

Dollar based net retention (DBNRR) rate declined from 121% to 120%. However, I’m not concerned because this lagging metric reflects the challenges faced in the past year. I anticipate an increase in this metric over the next 12 months.

The calculated billings guide for the full year remained unchanged, which was below consensus and likely the cause for the after-hours drop of -6%. While management considers it a key business metric, I focus only a little on it. It is more useful for evaluating short-term performance and can be a confusing metric when used to rate long-term performance or the general investment thesis for Zscaler. I’ve shared my thoughts about this topic in the past.

It’s clear why I have a strong conviction in Zscaler: there is substantial potential for durable long-term growth. Zscaler consistently highlights the opportunity to increase revenue by 6 times for each customer. The fact that 40% of the Fortune 500 are already customers demonstrates significant progress, while also suggesting the possibility of further growth through acquiring new customers, especially considering the potential for upselling.

Additionally, I like that Zscaler is expanding into other categories like IT tools with Business Analytics, which expands their Total Addressable Market (TAM). For more info, please refer to Muji’s comments here.

Cloudflare - NET

After a tough year for Cloudflare because of macro and sales execution issues, the company quickly turned things around. Q3’2023 was a great quarter!

Revenue growth increased to 32% compared to the previous year, and even sequential revenue growth accelerated to 8.8%. Cloudflare is on track to generate over $1B for the first time in a full year, which makes it one of the rare companies that can achieve this scale while still growing fast.

Management raised the low end of the full-year revenue guide to $1,287M from $1,286M, which is not much, but a positive sign. After the Q1’2023 guidance massacre, they are unlikely to guide overly optimistic again. They also raised the outlook for operating income for Q4 and the full year, as well as EPS for the full year.

Dollar-based net retention rate (DBNRR) increased to 116% from 115%, and I expect it to improve further as the macro environment improves.

During the quarter, Cloudflare added 206 large customers (> $100k,) which is an 8.8% increase compared to the previous quarter. These customers now contribute 65% of total revenue, compared to 61% last year.

Also, Cloudflare’s profitability is on track, with a 20% operating cash flow margin, 10% free cash flow margin, and 17% net income margin.

The quarter would have been even better if dilution had been slightly lower, as it is still growing at 4% compared to the previous year. Additionally, RPO (remaining performance obligation) growth was a bit light at only 4.5% sequentially.

I’m happy with these results and I expect them to continue because Cloudflare is in a good position for the decade, especially if more companies use the edge to reduce interference for AI applications. And it’s reassuring to see Matthew Prince wearing a helmet for skiing since he’s a big part of my investment hypothesis. :skier:

Snowflake - SNOW

Snowflake had a strong Q3’2024, beating and raising expectations:

  • Revenue of $698.5M, an increase of 34% YoY, and 9.1% QoQ

  • Raised full-year (FY) product revenue guide to $2,650M by 1.9%

  • Raised FY non-GAAP product gross margin guide to 77% from 76%

  • Raised FY non-GAAP operating margin guide to 7% from 5%

  • Raised FY Free Cash margin guide to 27% from 25%

  • 70% YoY growth of Free cash flow at $111M

  • Record high GAAP gross margin 69%

  • Record high Non-GAAP gross margin 75%

  • Record high Non-GAAP product gross margin 78%

  • 51% Non-GAAP net income margin

  • 127% Non-GAAP EPS growth

With 370 new customers added during the quarter, the total customer growth remained stable at 4.3% quarter over quarter. However, when adjusting historical numbers, only 2 customers from Forbes 2000 have been acquired, and I am unsure why. On a positive note, they posted 35 customers who generated more than $1M in revenue, showing a strong sequential acceleration of 8.7%.

Furthermore, they posted 4 customers who generated over $5M in trailing twelve months (TTM) product revenue and 2 customers who generated over $10M in TTM product revenue. Management also mentioned that their biggest customers have become stable in terms of optimizations:

  • Consumption growth in September exceeded expectations

  • Snowpark consumption grew 47% QoQ

  • Consumption in October was up 500% since last year

  • 30% of customers used Snowflake to process unstructured data in October

  • Consumption of unstructured data was up 17x YoY

  • Price per credit increased 4% YoY → Increased consumption of higher-priced additions of Snowflake

Remaining Performance Obligations (RPO) slightly accelerated sequentially to 4.5% QoQ, which might be boosted by seasonality and still muted by budget scrutiny.

DBNRR decreased to 135% from 142% (as expected), which mirrors the past 2 years’ challenges (24 months trailing metric). In reality, customer consumption might already be increasing, contrary to what the metric shows.

It’s great to see the main hypothesis about data sharing is materializing: 28% of customers now use data sharing, which is a strong 12.4% increase compared to the previous quarter.

There was also a comment about the mix of hyperscalers and where the consumption comes from. 76% of the consumption comes from AWS, 21% from MS Azure, and 3% from GCP. The reason why GCP has a low share of consumption is because it has higher pricing for customers.

On the product side, they highlighted the “Dynamic Tables” product, which entered public preview earlier this year. It is already being used by 1,500 customers. Also, Snowflake events had 23,000 people attending, which is twice as many as last year, showing a lot of interest in its innovations.

Management mentioned that there will be a lot of migration from on-premises to cloud over the next few years. Also, new solutions will be available in the fourth quarter, which will create significantly more opportunities for expanding workloads:

  • Native apps framework

  • Unistore for transaction processing

  • Snowpark Container Services

  • Apache Iceberg Tables

Based on dilution trends, the total number of weighted-average shares outstanding for the full year decreased from 362 million to 361 million. This decrease is a result of their share buyback program. In Q3, they spent $400 million to buy back 2.6 million shares, and year-to-date, they spent $592 million to buy back 4 million shares at an average price of $147.5 million.

It’s also good to see a data point like Snowflake ranking #1 on the Future 50 list, along with Datadog, Crowdstrike, Cloudflare, and Bill. The Future 50 list was created by the Boston Consulting Group and Fortune in 2017 to evaluate the long-term revenue growth potential of over 1,700 of the world’s largest publicly traded companies.

Overall, excellent. Let It Snow! Let It Snow! Let It Snow! :snowflake:

Datadog - DDOG

After missing their guidance and leaving investors with uncertainties in Q2 2023, many investors sold their shares at the -20% price. Initially, I was also confused. However, after a thorough analysis, I concluded that there is nothing fundamentally wrong with the business.

Fast-forward to Q3’2023, the strong earnings report rewarded my patience:

Revenue was $548 million, with year-over-year growth of 25% and accelerating sequential growth rates of 2.6%→5.8%→7.6%. Management raised the full-year guide by 2% to $2,107 million, and I anticipate that the growth will accelerate compared to the same period last year, starting at around 27% year-over-year in Q1 2024.

Management mentioned that the consumption levels are back to those seen in Q1, and October showed stronger trends than the previous year. Despite this positive performance, they stated that the guide is still conservative and does not reflect the positive trends in October. This is good as it reduces the likelihood of missing the target and allows for further potential gains.

Another highlight was the increase in remaining performance obligations (RPO), which continues to accelerate year-over-year (54% YoY!) and sequentially (16% QoQ!). Customer growth is on track, with 140 net new customers spending more than $1M ARR and a sequential growth rate of 4.7%. This is a significant improvement compared to the negative trend in the previous two quarters (130→80→140). Additionally, management emphasized that they added a record number of new customers exceeding $100k ARR in the quarter.

The dollar-based net retention rate (DBNRR) decreased to 115% from 120%. This is similar to Snowflake’s and other DBNRR metrics, which look at the past. I’m not concerned about it because of the overall economic situation. I’m a little worried about the adoption of modules. The growth rate for customers using 6 or more products slowed down to 2.7% (only 147 customers started using 6+ products) sequentially in comparison to 13.1% (636 customers started using 6+ products) in Q2. I don’t have a clear answer for this, but I will closely watch this metric.

Profitability is on track with $158 million in net income (29% of revenues), $138 million in free cash flow (25% of revenues), and earnings per share of $0.45 (a 96% increase compared to the previous year).

In the future, AI applications could help improve long-term growth. Currently, AI customers, mostly model providers, contribute 2.5% to ARR compared to 2% in Q2. With the increase in AI use cases, there will be many new opportunities for growth. The stock has increased by 40% since Q2.

MongoDB - MDB

After a surprisingly strong Q2’24 with revenue increasing by 40% compared to the previous year, MongoDB had a more typical Q3. However, they still exceeded revenue expectations by 7% ($433M, a 30% increase compared to the previous year) and raised their revenue outlook for the full year by 3% to $1,658M. The revenue from Atlas now accounts for 66% of the total revenue, up from 63% a year ago, and it has grown by 7% compared to the previous quarter.

Profitability is on track:

  • Full-year operating income guide increased by 21%

  • Full-year EPS guide increased by 24%

  • Operating cash flow margin: 9%

  • Free cash flow margin: 8%

  • Non-GAAP operating income: 18%

  • Non-GAAP net income: 18%

I remember that MongoDB, like Cloudflare, initially seemed to have trouble becoming profitable. However, both companies quickly adjusted and started making money when the overall economic situation changed. This shows that their business models are strong and able to grow.

On the other hand, getting new customers, except for large ones, was relatively weak:

  • Added 1,400 new customers in total (3.1% increase from last quarter)

  • Added 1,400 new Atlas customers (3.1% increase from last quarter)

  • Only added 100 new customers through direct sales (1.5% increase from last quarter)

  • Added 117 new customers spending over $100k (6.3% increase from last quarter) - Good

  • The ARR NET retention rate stayed stable at 120% for a long time now

I think the business is not doing anything wrong. It’s just that getting new customers directly is challenging in this situation. This is similar to what I wrote about Zscaler earlier, considering the current situation.

Overall, it was a good quarter. The stock price dropped by -6% after hours because it is already valued high, with a forward EV/S of 15. I don’t expect the stock price to go up significantly until revenue from consumption and getting new customers improve significantly. I believe this will happen when the overall economic situation becomes more favorable.

Monday - MNDY

Monday, the Work OS platform, had a rock-solid Q3’23. They did even better than expected, making $189.2M in revenue (38% more than last year), which was 3% more than their revenue guide. They also raised their revenue guide for the whole year by 1% to $725M.

Profitability remains strong, with a positive GAAP net income of $7M (last year it was -$23M). They had $67M in operating cash flow (35% of revenues) and $20M in free cash flow (11% of revenues). Their non-GAAP EPS increased to $0.64 from $0.05 last year, and their GAAP EPS increased to $0.15 from -$0.51 last year. Their gross margin decreased a little to 89.3% from 89.9% last quarter. It’s something to keep an eye on, but not something to worry about.

Customer growth is strong with 185 net new customers spending more than $50k, up 57% compared to the previous year, and 9.8% compared to the previous period. 59% of Forbes 500 companies are customers, which means there is still potential for more of these large customers as well as significant opportunities for upselling:

One of Monday’s key strengths is their ability to quickly create and distribute new products. For instance, the rollout of mondayDB is progressing faster than expected, and they have introduced a new category called “monday Service” for ticket management, CSAT, Portal, SLA management, and Asset management. Although not available to all customers yet, Monday CRM has gained good traction with 11,025 accounts, a 32% increase from last quarter. Furthermore, monday dev accounts have grown to 1,042, a 33% increase from last quarter. Management aims to make monday CRM and monday dev accessible to all customers by the end of Q1 FY24.

Net retention rates decreased to 110% for all customers, 115% for customers with 10+ seats, and 115% for customers spending more than $50k. However, management mentioned being “encouraged by the signs of stabilization that we witnessed during the most recent quarter”.

Overall, I’m happy to be part of Monday because:

  • TAM: The company has a large and growing market opportunity ($56B).

  • Diversified revenue: They generate revenue from different parts of the world (only 54% from the US).

  • Execution: They excel in various areas, such as culture, go-to-market strategy, research and development, and financials.

  • Insider ownership: The company is led by owner-operators who own 20% of the company.

  • AI: They also have the potential to develop AI solutions that can double revenues by adding value to existing customers.

To learn more about Monday, you can check out their Investor Day 2023 presentation here.

(My thoughts are with all the people affected by the current war in Israel and around the world, which I hope will come to an end soon. It’s impacting 7% of Monday’s global workforce who have been called up for reserve duty.)

New positions

I’ve started four new positions, but I’m not ready to discuss them yet. Each position represents only about 2% of the portfolio. I’ll provide more insights once these positions reach full size, as I may decide not to continue with one of them.

Closing Thoughts

The recent earnings season started with a bang as both CFLT and BILL experienced nearly a 50% drop after earnings, causing me to question my entire portfolio.

Fortunately, since I needed more confidence in those companies, my positions in them were relatively small.

However, the rest of the earnings season turned out to be quite favorable, as many of my largest holdings such as NET, SNOW, DDOG, and CRWD saw double-digit rallies after their earnings reports and cybersecurity has proven to be remarkably resilient.

‘Cybersecurity titans’ have raised their forecasts after beating the best-case scenario last quarter. And the stocks have performed exceptionally well this year: CRWD has seen a 150% year-to-date (YTD) increase, ZS has gained 103%, NET has gained 99%, S has gained 90% YTD, and I would almost include DDOG, touching cybersec, in that group with a 70% YTD increase.

Talking about cybersecurity: I always make a list of potential reasons—thesis breakers— to sell my companies, and “serious security breach” is always one of them. For every company. This shows how important cybersecurity is and continues to be. In addition, there is a growing trend towards using cloud services, and it is crucial to secure these services, especially in the era of artificial intelligence.

The major players in cybersecurity have been incorporating AI for a long time and have already developed their AI solutions. For example, Zscaler has 360, SentinelOne has Purple AI, and Crowdstrike has Charlotte AI. I believe, these AI solutions will contribute to significant growth for all of these companies, as they participate in the AI infrastructure layer throughout their lifecycle.

No single leader holds a double-digit market share. For example, Microsoft, the largest player, had approximately $15B in revenue in 2022, which is still less than 10% of the total market. Palo Alto represents roughly 3%, while CrowdStrike holds about 1.5% etc.

Gartner predicts the information security and risk management market will reach $267.3 billion in 2026 with a CAGR of 11% until 2026.

Moreover, statements like the one below by Palo Alto CEO Nikesh Arora, further reinforce my conviction that there is enough space and runway in the cybersecurity landscape for multiple players.

It’s not a coincidence that almost all of them are reporting strong numbers. That’s why I’m invested in cyber security.

What to expect in 2024?

Some people expect a recession to hit and affect the entire market, while others anticipate a continuation of last month’s rally. As for me, I’ve no idea! I read a fantastic book called Stocks for the Long Run by Jeremy Siegel, which presents a wealth of statistical evidence:

Waiting for a recession to be officially declared before deploying cash can be risky. The National Bureau of Economic Research takes its time to confirm, and by then, the market may have already bottomed out, making the recession old news.

For instance, the 2001 recession lasted eight months, starting in March and ending in November, yet its official announcement came later. Similarly, the Great Recession, from December 2007 to June 2009, had its official start declared a year later, and the end date was announced even later. Relying on timely announcements can—I would even say ‘will’—result in missed opportunities and losses.

Instead of worrying about things I can’t influence, here’s my thinking:

Thesis off? Good! Reflect and refine. Market volatility? Great! It’s a buying opportunity. Life throws curveballs? Fantastic! Embrace personal growth. Perspective is key—setbacks are lessons, and challenges are growth opportunities. So, as always:

I focus on what I can control and ride the market coaster if necessary

After experiencing a two-year investing nightmare with a -61% decline in 2022, I’m very happy with 55% YTD return (versus 22% for my benchmark, the accumulating S&P500 ETF) and 17% annualized Time-Weighted Return (TWR) since 2019. Going forward, I’ll continue working towards my goal of achieving a 20% annual return over the next few decades. But can I realistically achieve that goal with the companies I own?

Considering their current market caps and valuations, I’m not sure. However, these are excellent companies with strong leaders, large addressable markets, and favorable trends. I expect most of them to continue growing steadily or even faster. And they must keep delivering, or else it might get tough. One promising sign is that cloud growth has been increasing again since Q3’24 and has shown resilience with 16.4% growth:

My plan is simple: I’ll stay invested in these excellent companies for as long as possible until I see significantly better opportunities or if a specific investment thesis begins to show weaknesses. In that case, I’ll use the funds to invest in one of my newer, smaller, and lower-valued companies. One thing is certain: It’ll be exciting to look back on these thoughts in 2034—10 years from now.

A big thanks to everyone on this board who contributed and helped me improve as an investor. Thank you!

Twitter: @MoritzMDrews

My Previous Portfolio Summaries

Disclaimer: This portfolio summary is for informational purposes only and does not constitute investment advice. Everything expressed here is solely my personal opinion. As I am not a professional, please do not blindly follow my perspectives as they could lead to incorrect conclusions.


This article just came out today:

It claims ethernet grew about 20%-25%, but:

This is a far cry from the 5X increase in InfiniBand networking revenue that Nvidia had in its most recent quarter. For the trailing twelve months, we estimate in our model that Nvidia’s InfiniBand revenues rose by a factor of 3.2X to $5.53 billion. (This includes switches, network interfaces, DPUs, cables, and software, not just switches.) But the datacenter Ethernet switch market is still at an annualized run rate of around $20 billion

Networking revenue in the last quarter was over $2.5B, meaning even if flat from here on out (unlikely), it’s a $10B business for Nvidia in the upcoming year.

BTW, Nvidia beat out Intel and others to acquire Mellanox - the company producing InfiniBand products - back in 2019, closing in 2020 for $6.9B. Smart management there.

Now, true, from this article:

InfiniBand is essentially controlled by a single vendor, and the hyperscalers and cloud builders hate that, and it is not Ethernet, and they hate that, too. They want one protocol with many options in terms of functionality, scale, and price.

And Nvidia has produced ethernet products, such as its “Spectrum-X” line, which is part of the expected upcoming “Ultra Ethernet” standard.

Nvidia contends that the Spectrum-X portfolio can deliver somewhere around 1.6X the performance of traditional datacenter Ethernet running distributed AI workloads, and has said that InfiniBand can deliver another 20 percent performance boost beyond this.

So, anyone that wants to stick to ethernet for compatibility or whatever can still do that through Nvidia with the currently fastest possible speeds. But, those who have the ultimate need for speed will probably continue to opt for InfiniBand, just as they opt for the fastest AI chips from Nvidia.

Forgetting all the tech details, I think the important take-away is that Nvidia continues to give potential customers options. If customers want the best AI/GPU chip today they get the H100. If they need/want traditional compute, get Nvidia’s big “Grace” ARM-based data center CPU. And customers can choose between ultra-fast InfiniBand or ethernet-compatible Spectrum-X products for networking, and still stay within Nvidia.


It’s important imo to include some detail regarding the word “network” in the context of $ANET and/or $NVDA

There are lots of networks in the chain of events starting at an end-user request and ending in a response. For example in the case of a browser residing on a computer inside a typical home residence; to grossly over-simplify and skip over a lot of stuff:

  1. There is a Local Area Network (LAN) physically inside your wireless router: a switch, a router, software etc.
  2. There is a “backbone” and/or “Wide Area Network” (WAN) that connects your wireless router to your ISP
  3. There is a backbone and/or WAN that connects the ISP to a web-proxy device of some sort (Load Balancer /router/Web Application Firewall/probably a combination-device) in the De-Militarized Zone (DMZ) at a dataceter
  4. There is a network that gets the user request from the DMZ to a Virtual machine (VM), physically inside the datacenter e.g the VM that hosts the listening process (HTTPS) for the requested website. This is a datacenter “front end” network; it is typically implemented with Ethernet, and is $ANET’s strength currently.
  5. The middleware handling the relevant business logic eventually makes a request of a database: a relational or non-structured database, or a AI inference capability. In any case, there is a database, and for performance/scalability/redundancy purposes, the database is deployed across multiple physical nodes e.g. a “cluster” of nodes. The database nodes are physically connected by a “back end” network, typically implemented via InfiniBand; this is $NVDA’s strength currently.


Here are some requirements and characteristics of the “front end” network":

  • Needs to connect LOTS of heterogeneous hardware devices
  • Each device can have LOTS of processes listening for requests.
  • …Thus, there is a lot of complexity to stand up, configure, re-configure, maintain, secure and troubleshoot the “front end” network.
  • Requests and responses are aggregated/multiplexed/bundled so that relatively huge numbers of disparate requests from logically separate functions can be sent down the same physical media
  • Latency and throughput are important, but arguably configure-ability, flexibility, visibility, heterogeneity etc, are equally as important; thus, Ethernet is used.


Here are some requirements and characteristics of the “back-end” network:

  • Needs to connect, relatively-speaking, VERY FEW, HOMOGENEOUS hardware devices.
  • Each device has, relatively-speaking, VERY FEW processes listening for requests.
  • Traditionally, once the back-end network is set up, it is rarely touched or changed subsequently.
  • The data sent typically has to do with a VERY SMALL number of specialized functions (e.g. database transactions and/or queries)
  • Latency and throughput are of primary importance; thus, Infiniband is used.
  • The back-end network is dedicated to a relatively small number of hardware devices serving a single dedicated purpose e.g. a database.

What’s interesting to me is that $ANET, to this point, has not been able to establish a presence in the back-end network, but THEY HAVE AMBITIONS TO DO SO. They plan on releasing devices that utilize an upgraded version of Ethernet which (…they hope) will beat Infiniband on latency, and also beat Infiniband in terms of flexibility, visibility etc. $ANET is working with their customers, and the relevant standards organization, to define the characteristics of this to-be Ethernet upgrade.

IMO it’s a good example of how leadership at $ANET is constantly looking to expand revenue.

I’m long $ANET and $NVDA; small positions in them both.


We’re getting into some heavy duty tech weeds here - let me try to uplevel the discussion.

First, the “standards organization” you’re talking about is the Ultra Ethernet Consortium, which isn’t a true standards organization run by ISO or ANSI, it’s actually:

a buddy movie collection of Ethernet ASIC suppliers and switch makers who do not really want to cooperate with each other but who are being compelled by the Internet titans and their new AI upstart competition to figure out a way to not only make Ethernet as good as InfiniBand for AI and HPC networking, but make it stretch to the scale they need to operate.

What unites these companies – Broadcom, Cisco Systems, and Hewlett Packard Enterprise for switch ASICs (and soon Marvell we think), Microsoft and Meta Platforms among the titans, and Cisco, HPE, and Arista Networks among the switch makers – is a common enemy: [Nvidia’s] InfiniBand.

The enemy of my enemy is my ally.

And apparently Meta spent $1B for Arista networking in 2022:

I couldn’t get through the whole first article I linked above - it’s well written but extremely tech dense. Here’s how it concludes:

But if the Ultra Ethernet Consortium has it Meta Platforms’ way, Ethernet will be a lot more like InfiniBand and will have multiple suppliers, thus giving all hyperscalers and cloud builders – and ultimately you – more options and more competitive pressure to reduce prices on networking. Don’t expect it to get much below 10 percent of the cost of a cluster, though – not as long as GPUs stay costly. And ironically, as the cost of GPUs falls, the share of the cluster cost that comes from networking will rise, putting even more pressure on InfiniBand.

It is a very good thing for Nvidia right now that it has such high performance GPUs and higher performance InfiniBand networking. Make hay while that AI sun is shining.

Another view, with commentary and less tech, is here:

But, remember, Nvidia has its own fast ethernet products - the Spectrum line. Nvidia’s idea is that full-on AI clusters (think ChatGPT) should use InfiniBand because it’s the fastest; if your data center has a mix of AI and other data center workloads, then Spectrum is Nvidia’s ethernet offering.

I haven’t read anything to suggest that Ultra Ethernet will surpass InfiniBand in speed or latency, just flexibility and compatibility. And so my own view is that while vendors want customers to buy ethernet compatible products, people building large AI clusters (the people buying as many Nvidia H100s as they can get their hands on) almost certainly don’t want networking to be the bottleneck, and Nvidia makes InfiniBand work natively with its chips. If you’re building a multi-tenant cloud, sure, you probably want ethernet, but hey, Nvidia offers that as well.

BTW, I do think Arista has great engineering. I owned shares back in the day. I haven’t looked at the business side of thing in years, though.


Smorg - thanks for the up-leveling!
Very much appreciated and lots of food for thought.

I find it very interesting that $NVDA apparently wants a permanent monopoly in the back-end network space and is squaring off seemingly against everyone else (…not just competitors but also their own customers!) to maintain its current stranglehold.

It’s very profitable (…at least for a while) to have a monopoly, but it seems to me that some of $NVDA’s biggest customers feel strongly that $NVDA is not aligned with their long term interests. For the moment though, seems to me all their customers can do is grumble about it while shoveling truckloads of money over to $NVDA.

$NVDA arguably has a stranglehold not just on the back-end network, but basically the entire GPU hardware and software ecosystem as emphasized repeatedly, over long periods of time, in their CCs. Their tight-single-vendor-integration of the entire stack is what gives them the leverage to keep the Infiniband monopoly.

So who knows, they may be able to maintain their stranglehold (…on the entire GPU ecosystem, not just the back-end network) for years, conceivably decades.


That’s not my view of it, but I don’t have any particular insider or even hands on experience, just an Occam’s Razor view:

I believe Nvidia bought Mellanox in 2019 because it saw InfiniBand as the best way to get networking that was fast enough for the GPUs it wanted to build. At the time, Mellanox’s biggest customers were HP and Dell/EMC, and Oracle was a customer, too.

Nvidia continues to make and sell ethernet networking compatibility and products today, just as Mellanox did before the acquisition. No one buying Nvidia GPU or CPUs is forced to use InfiniBand. But, it’s 20% faster than the fastest ethernet available. Customers grumbling about the fastest networking being InfiniBand, indicate to me that there other issues at play.

The UEC (Ultra Ethernet Consortium) group is working on getting ethernet up to InfiniBand speeds and latency. I don’t have any insight as to when or even whether they’ll be successful, nor whether Nvidia will have new versions of InfiniBand that are even faster. I don’t think Nvidia much cares as long as slow networking doesn’t make faster GPUs a moot product.


I don’t know NVDA from the inside, but I do know another major CPU supplier very well. They had the fastest computing products, but the processing was being limited by slow data transfer in and out of the processor. Other manufacturers had the same issues. Their number-crunchers crunched faster than the fastest data interfaces could move the numbers in and out. They all spent lots of money pushing the data transfer technology forward just so their computing technology could actually be used.

Infiniband was born of a merger of competing solutions specifically directed to high speed computing.

NVDA still has the same problem: Their number-crunchers get faster every generation. They need faster and faster I/O hardware to get their data in and out as fast as possible. Infiniband architecture seems to be good for that application. Whether or not they are trying to secure a monopoly is almost irrelevant. They need the interface speed just to get the maximum performance from their product.


@ibuildthings that sounds reasonable to me. But also…

If $NVDA doesn’t care whether their customers use Infiniband or Ethernet
…given that their customers want Ethernet
…I do find $NVDA’s absence from the UEC interesting.

I’ve been convinced though that the monopoly question isn’t all that material.

More importantly, after reading through the references @Smorgasbord1 provided, I’m convinced that $ANET sees no choice but to go after opportunities in the soon-to-geometrically-expand GPU-back-end-network.