No More Hustleporn: How to Understand AWS

Tweet by Nayut Sitachitt

I manage ESA Global Value Fund. I’m here to learn and share what I’ve learned.

My tweets are not investment advice.

1/ Understanding AWS - A perspective from a manager who has $AMZN as one of the largest positions in the portfolio for over 7 years.

In the thread, I will also discuss $MSFT's Azure, $GOOG's GCP. Let's dive in.

2/ Look at the Future Not the Past - Some tend to focus on current IT spending and how much will migrate to the cloud. That understates the cloud's potential. It’s the new workloads that don’t yet exist today that will power the growth of the cloud.

3/ Revenue Potential - If every company is increasingly turning into a software company, and their incremental spending is on the cloud, one can look at AWS as a business that takes an increasing share of revenue across an increasingly large number of companies.

4/ It gets a small cut of all the future digital economic activities, and it faces much less competitive constraints than the retail side of Amazon. AWS will most certainly be a multi-trillion revenue business over the longer term due to the dynamics above.

5/ Workload Type - I put workloads into three buckets 1) Migration: on-prem workloads that move over to the cloud. 2) 3rd-Party SaaS Workloads: workloads of SaaS companies that build on top of a cloud platform. SaaS companies have direct customer relationships.

6/ 3) Net-New Build: cases where larger enterprises built their own solutions on top of a cloud platform, e.g., Netflix and other companies building on top of AWS. It’s important to break them into the above as historical and future demand differs in each case.

7/ Azure - Azure receives a lot of migration workloads from companies running Microsoft on-prem ecosystem (1). Once migrated to the cloud, these companies then build on top of Azure (3).

8/ Azure customers are more likely to consume other Microsoft services when they build on the cloud. This means Microsoft will generally have a higher wallet share of its customers compared to AWS as companies move from one Microsoft ecosystem to another.

9/ AWS - It doesn’t have an on-prem ecosystem like Microsoft. The migration loads tend to be companies that use open-source, e.g., Linux, and they are more likely to continue to use open source and 3rd-party services once they are on AWS.

10/ Despite that, its strength in net-new workloads (2,3) is what makes it so valuable. AWS is a preferred platform for developers. Most SaaS, tech companies, and enterprises that want the best ecosystem to really build use AWS.

11/ Migration (1) - Migration workloads are a bigger growth driver for Azure in the past and in the near future, but they'll slow drastically over the longer term. This is partly why we see the growth rates of AWS and Azure converging.

12/ Microsoft understands this well, which is why it bought GitHub and took action to try to attract developers to its platform. But being in the best ecosystem matters a lot more to developers and tech-forward organizations than to average enterprises.

13/ Ecosystem Advantage - AWS ecosystem has more features, more developers on it, and more available 3rd-party resources on the web. For companies, it means a larger talent pool. For software engineers, it means more job opportunities. This is hard for Azure or GCP to match.

14/ SaaS Workloads (2) - This is likely to be the fastest-growing market over the longer term, and they’re largely on AWS. The talent shortage makes it hard for average enterprises to hire good tech talent to build own solutions. SaaS companies of the future will fill this gap.

15/ Net-New Build (3) - A large driver of growth of this market will likely be tech companies that can continue to hire talent. Some larger enterprises will, too, be a driver. As the home of many tech companies, AWS will benefit significantly from this.

16/ Lock-in - The moat of AWS strengthens as its business grows. The more a customer builds on AWS, the higher the lock-in. Over time, the customer is shifting from trying to avoid the lock-in to just accept it so they can take full advantage of all the services offered.

17/ This is a win for the customer and an even bigger win for AWS. If you read the 10-Ks of many AWS customers, they highlight the extent of the lock-in and read almost like a bull case for AWS.

18/ Margins - Due to the high lock-in, the ability for AWS to dictate its margin is very high. I would argue that cloud providers' lock-in is higher than any other form of tech lock-in. However, at growth stage, it doesn't make sense to exercise full pricing power.

19/ AWS can essentially charge whatever it wants without seeing much churn given the deep lock-in above. The longer-term margin is not capped by the business model itself but will likely be constrained by regulations if the company abuses its power.

20/ Durability - IaaS is more durable than PaaS/SaaS. Snowflake may be the best at what it does now, but many are coming after its market. What one can count on, however, is that both the incumbent and the challengers will continue to buy compute and storage from IaaS providers.

21/ Customer Acquisition - GCP losses understate how high CAC is for a challenger because when it buys or invests in companies to get cloud contracts, those deals are CACs that aren't being expensed. AWS has the advantage of being the default choice and pays far less CAC.

22/ SaaS built on AWS also spends a great deal of money to help AWS acquire customers. SaaS companies hold the relationship, but AWS gets a cut for the IaaS consumptions, all while putting in no effort on customer acquisition.

23/ Life Time Value - Hyperscalers' customer LTV is the highest LTV found in any industry. That's why companies go as far as buying shares in other companies to win cloud contracts. Securing them now means getting an increasing slice of their IT spend to perpetuity.24/ 30,000 - That’s the number of job openings at AWS, and it's roughly equal to the openings at the rest of big tech combined. Amazon doesn’t have enough people to do everything it wants to do, and this opens up opportunities for other companies to step in.

25/ AWS must prioritize, and thus far, it has put more focus on areas directly related to compute and storage e.g., Lambda, Graviton etc. It’s struggling to keep up on things where smaller and more focused companies leverage IaaS from AWS to offer competing services.

26/ Amazon's poor reputation as an employer provides a massive opportunity for many startups and smaller companies. It still works out okay for AWS as it still gets a cut when those services are consumed within AWS.

27/ First & Best Customer - The 3 hyperscalers have themselves as their first and largest customers. Their other businesses help fund the growth and give their clouds added scale. This shows how some of the best opportunities aren't ones that startups are best suited to pursue.

End/ Although I've been using Twitter for a long time, I've been writing tweets for only 3 weeks. My current reach is limited. If you like my writing, please consider sharing it, so it reaches more people.