Hyperscalers – Friend or Foe?

The dynamic between the public cloud vendors and independent software providers is evolving quickly. Just a few years ago, the hyperscalers (AWS, Azure, GCP) were rapidly rolling out their own infrastructure services targeted at various layers of the application stack. These included solutions for data processing, security, communications, identity and even observability. The premise was that as enterprises migrated application workloads to the cloud, the hyperscalers might as well try to capture as much spend as possible. These services went far beyond the basic storage and compute offerings of their foundation.

These software infrastructure services were built in rapid succession. If an open source project already addressed a use case, the hyperscaler simply offered a cloud-hosted version of it. This approach generated backlash from the companies that were the primary developers of those projects – notable examples being MongoDB, Elastic, Confluent, Redis and others. When an open source solution wasn’t available, the hyperscalers built their own custom offerings, applying teams of internal developer resources.

The market’s reaction was that competition from the hyperscalers would bury the independents. Many pundits predicted the demise of any software vendor that fell into the product scope of the hyperscalers. An investment thesis for an independent software company was immediately met with the common refrain of “what about competition from AWS, Azure or GCP?” Similarly, for new categories, the common view was that the hyperscalers could marshall their “enormous resources” and crush the start-up.

For a while, this strategy worked effectively, driving incredible growth for the hyperscalers. Analysts and software company CEO’s would closely watch events like AWS re:Invent for the next software category that a hyperscaler was entering. Stocks of public companies would drop immediately on any announcement implying that they were at risk. This was even when the announcement was just a press release reflecting intent or a bare-bones demo of an MVP.

However, as I argued in a prior post, independent software vendors persevered, capitalizing on advantages in focus and nimbleness to bring solutions to market that surpassed the usability and features available from the hyperscaler look-alike products. The independents found software categories that could sustain a stand-alone software service and built tight feedback loops with customers to rapidly iterate to a full product solution. Additionally, these companies were often tracking towards an IPO, offering lucrative stock options to attract top talent.

In defense of the cloud vendors, it quickly became unrealistic for them to field best-of-breed solutions in the hundreds of emerging categories. Too much VC money was funding competitive offerings in niches. As the market for cloud infrastructure services grew into the hundreds of billions of dollars, this became self-fulfilling, where even a percent of the overall market made for a sizable category. The hyperscalers had to start narrowing their field of bets. While they might keep products alive in every category, many of those MVP’s transitioned to life support as independent offerings clearly outperformed them.

Another factor that influenced hyperscaler behavior was the migration to the cloud. While some independents started with software packages that customers could install into their data centers, almost all of them eventually transitioned to a cloud-hosted version of their software. And for that cloud version, they didn’t stand up their own data centers in most cases. Rather, they made the unusual choice of hosting on infrastructure provided by the hyperscalers. This seemed counter-intuitive at the time, as the hyperscalers were also competitors.

However, this choice turned out to be prescient, providing the main catalyst for a new symbiotic relationship with the hyperscalers. As the independents sold more cloud-based software services, the hyperscalers generated revenue from the underlying storage and compute. While this growth threatened the hyperscaler look-alike products on the surface, the hyperscalers acknowledged that those dedicated product teams also incurred high cost. Their operating margin from some software services might be higher by co-selling an independent’s service than supporting an internal competing team.

This perspective makes sense, when we consider that the big revenue opportunity for the hyperscalers still represents the ongoing migration of large enterprise customers from on-premise or private data centers to cloud infrastructure. If partnering with the independent providers helps pull more workloads from on-premise data centers to a cloud-hosted solution, then the hyperscalers should support it. These migrations may not have been possible when competing just on the available feature set or cost of the hyperscaler’s own solution suite.

Additionally, the Big 3 hyperscalers of AWS, Azure and GCP are increasingly competing for these cloud migrations in a “winner takes all” posture relative to one another. As multi-cloud deployments reduce their leverage with customers, each hyperscaler prefers to win the lion’s share of business with an enterprise customer. That is why we see major press releases from any of the cloud vendors when they complete a major deal with a Global 2000 company.

This competitive infighting between the cloud vendors is causing them to re-evaluate their own service offerings and fill product gaps where they are less competitive. If one hyperscaler is weaker than another on big data processing or security, they are increasingly forming a strategic co-sell relationship with an independent to fill that gap. Their goal is to win the majority of cloud infrastructure business from the enterprise over another cloud vendor, even if that means giving up some revenue to the independent for filling a feature gap.

The impact of this trade-off is mitigated further by the fact that most independent providers consume compute and storage resources from the hyperscalers. Even if the independent provider is selected for data warehousing, data streaming or observability, the hyperscaler still generates some easy revenue from the utilization of their platform. And they avoid the cost of managing that particular service.

Because of these factors, we now see the hyperscalers completely reversing their strategy of even a couple years ago. They have created marketplaces that make third party services available to developers on top of their infrastructure. Almost every cloud-hosted independent provider is included in these. In some cases, the hyperscalers take the relationship to the next level and engage in active co-selling with sales teams from the independents. These choices are more strategic in nature, used as a competitive strategy against another hyperscaler or a way to attract customers with a preference for a popular independent solution.

In this post, I will explore this evolving relationship. I will look at some of the historical drivers and what has changed. Then, I will examine a few examples of software segments where independent solutions seem to be thriving and which providers stand to benefit the most. Finally, I will draw some conclusions about how these trends might impact the investment thesis for several of the independent software providers tracked on this blog.

I think this dynamic is worth a blog post because it could provide a significant growth tailwind for select independent software providers in the near term. This magnified effect didn’t exist even 6-12 months ago for some providers. As an example, Snowflake’s contribution from the hyperscalers jumped in the most recent quarter and now makes up over 50% of new product sales. As we investors consider the risk of revenue growth deceleration in the coming quarters for hypergrowth stocks like DDOG, SNOW, MDB, CRWD and others, these new revenue contributions from the hyperscalers could maintain their momentum by introducing new sales channels and lowering customer acquisition costs.

Background

AWS was fully launched in March 2006, as an integrated suite of core services offered as an environment for other developers to build and host their Internet applications. The initial version of AWS consisted of S3 (cloud storage), EC2 (virtualized compute) and SQS (messaging). Amazon began quickly expanding the set of services offered on AWS, trying to keep up with the increasing demands of the developers using it. As an additional driver, by 2010, all of Amazon’s retail sites had migrated to run on AWS. Outside companies began utilizing AWS for hosting as well. Netflix famously started using AWS in 2008 and then declared their intention to go 100% cloud in 2010. They finally completed the cloud migration in 2016.

As you would expect, as usage of AWS grew, it required more and more infrastructure services to address all the common components associated with building modern internet applications. These started as layers around the core compute and storage functions and evolved to over 200 fully featured services today. The pace of these releases has been extraordinary, with the number of new services launched each year accelerating over the first 10 years. Here are rough counts of new service launches by year, taken from Jerry Hargrove’s blog post on AWS History.

There are now so many services, that Amazon had to start grouping them into categories. A user can see all the services by navigating to the AWS home page and hovering over the Product tab to view an expandable menu.

There are now 25 different categories of services. Most categories have many individual service offerings within them. The biggest category is Machine Learning with 32 offerings. This “shelf space” approach to provisioning services makes sense. Any time a developer was seeking a new service to plug into their application, AWS wanted to make theirs the default. This is because AWS could incrementally charge for usage of each service.

This incremental contribution to revenue for AWS became the main driver of the explosion of services. As AWS was becoming an ever larger component of Amazon’s overall revenue and profitability, the AWS team continued pushing to create ever more services to charge for. Many services were introduced in an experimental form. If they received traction, AWS would invest further in building them out. If adoption was low, they would slow down development and promote them less. Interestingly, given the large number of services started, not very many services have been completely shut down. This reflects the fact that keeping a service on life support can be low cost, while still generating some recurring revenue.

With the growing visibility of AWS, other large tech companies saw the potential for cloud hosting and wanted to ensure they weren’t left out. Microsoft announced Azure in 2008 and formally released the service in 2010. Azure now advertises 600 individual services. The Google Cloud Platform (GCP) was initially launched as App Engine in 2008, which offered basic Platform-as-a-Service functionality for running code. It has similarly expanded its suite of services, but in a more methodical way.

According to industry researcher Canalys, Amazon’s AWS accounted for 33% of the market in the most recent quarter. Microsoft Azure had an estimated 22%, while Google Cloud rounded out the top three at 9%. Other cloud providers, including IBM, Oracle, Alibaba and a long tail of smaller companies, accounted for the remaining 36% of the overall cloud infrastructure market. This means that any independent’s solution on the big three providers reaches almost 2/3 of the market.

Most of the growth for Azure and GCP has come at the expense of the smaller providers. Going forward, their growth is also butting up against AWS’ share. In the most recent quarter, AWS delivered nearly 40% y/y revenue growth at a run rate of $71B. Azure and GCP clocked even higher revenue growth in the mid-40% range, but at lower run rates. While cloud migrations and digital transformation is driving an overall expansion of the market, these big three providers are increasingly competing with one another in order to maintain their elevated growth rates.

Additionally, they are targeting the 36% share held by the long tail of small providers. As the big three hyperscalers seek to win the majority of a large enterprise’s cloud migration spend, they can promote the superiority of their product suite across multiple software infrastructure categories. This completeness of offering is difficult for the smaller cloud providers like IBM or Oracle to duplicate. Strategic co-selling relationships with the leading independent software providers compound the disadvantage for the smaller providers. Leading independents get more leverage out of a strategic relationship with AWS than IBM.

Sustained Hypergrowth

At their enormous scale, the cloud vendors are still growing at exceptional rates. This has created an interesting exception to the “law of large numbers”. For typical SaaS companies in areas outside of software infrastructure, we generally see annual revenue growth slow down significantly as the company’s total sales expand. In the past, analysts would model revenue growth deceleration into the 20% range or lower as a software provider surpassed $1B in revenue.

This was born out in actual numbers. For example, Salesforce (CRM) delivered annual revenue growth rates between 50 – 100% while annualized revenue was below $500M. However, once revenue climbed above $1B, revenue growth rates dipped into the 20% range. Granted, this corresponded with the financial crisis, but revenue growth never pushed back above 40% and has settled into the 20-30% range over the last 5 years as total revenue grew past the $10B range. Other large scale software providers, like Workday, exhibit a similar trend.

We can contrast this with the recent performance of the hyperscalers, which are demonstrating much higher growth rates at scale. In Q4 2021, AWS delivered 40% annualized revenue growth on a run rate of $71B. This even represented a slight acceleration over Q3 at 39% y/y growth. Azure delivered 46% annual growth in constant currency with a $40B-$45B run rate. Finally, GCP (which includes GSuite) logged 45% growth at an annualize run rate of $22B.

Investors should note that all three of these cloud providers delivered revenue growth 10-15% higher than Salesforce’s most recent quarter. AWS and Azure did this at much higher revenue run rates. I think this demonstrates the special place that software infrastructure companies occupy. While consumer internet and B2B software companies will wax and wane in rapid cycles, the picks and shovels providers that enable those digital experiences will continue to grow. This is because they benefit regardless of which end user software solution is popular at the time. Each new wave of CRM, e-commerce or social networking tools will utilize their software infrastructure. This allows software infrastructure companies to keep growing proportionally to cloud migration and digital transformation as a whole.

Relationship with the Independent Providers

While independent software companies have made announcements in the past about improving collaboration with the hyperscalers, the trickle turned into a flood over the past 12 months. I don’t recall hearing so many CEO’s refer to the benefits of hyperscaler collaboration, as we did on the series of earnings calls for the most recent quarter. The commentary referred to actual co-selling relationships and revenue generation, versus announcing marketplace listings and hypothesizing about the potential impact. Some providers even shared actual sales metrics through hyperscaler channels.

Adding to the narrative were the distinctions made between the levels of cooperation with the different hyperscalers. Some were called out as being very strong sources of new bookings, with others labelled as zeros. Granted, some of this is driven by competitive jawboning, but the battle lines are clearly being drawn. What is changing is the broader view of the hyperscalers versus the independents. The landscape is no longer binary – one group pitted against the other in a zero sum game. New strategic combinations of participants from both groups are emerging in response to increasing competition between the hyperscalers themselves and their need to pull back share from the long tail of smaller cloud providers.

The cloud vendor most consistently mentioned as a material contributor to sales for the independent software providers in the past quarter was AWS. What is most surprising about this shift is that AWS has historically been the most aggressive competitor, particularly for the independents built on an open source project. Collaboration with AWS at these levels would have been laughable just a few years ago.

Let’s start digging into this trend by examining a few examples. Out of this, I will try to identify a few patterns of relationships. This should help investors with a framework to evaluate the potential impact on their favorite independent software companies. This trend will be important for investors to appreciate going forward, as it is already having material impact on the financial performance of select companies.

Snowflake

The most notable co-selling highlight from Q4 had to be in Snowflake’s report. The scale is staggering. Of the $1.2B in new contract sales in Q4 for Snowflake, over half, or $700M, was co-sold with the hyperscalers. That amount is up 40% sequentially from the $500M in co-selling accumulated over the prior three quarters of the year. That represents an enormous acceleration. The CFO further clarified that the majority was co-sold with AWS, $0 with GCP and the balance with Azure.

AWS has clearly ramped up their sales efforts with Snowflake. Digging deeper, the collaboration has allowed AWS to win business from GCP. At the JMP Technology conference, Snowflake’s CFO discussed many competitive wins over GCP through their co-sell relationship with AWS. While AWS doesn’t win the data warehouse business for their Redshift product offering, they can cross-sell Sagemaker for machine learning and the rest of their data processing and transit services as a bundle. These are deals that AWS might have lost to GCP otherwise.

It’s also important to consider the category of software services. Processing large data sets for analytics and predictions is one of the largest IT spend initiatives for enterprises. I have discussed the explosion in data creation and the potential for enterprises to harness it. Additionally, data storage creates significant gravity for cloud infrastructure customers. Winning big data migration projects is a top priority for each hyperscaler.

The evolution of AWS’ relationship with Snowflake was highlighted in a February article on technology news site CRN. The title in itself underscores the shift in disposition, “AWS And Snowflake: From True Competitors, To Frenemies To…An Alliance”.  For those worried about lingering competition from the hyperscalers, we investors have to update our mental models to refine the broad brush of cloud provider competition as a drag on every independent software provider’s growth potential. The effect is rapidly becoming much more nuanced.

Amazon Web Services’ relationship with data cloud provider Snowflake has followed a dynamic arc that’s seen the two companies evolve from competitors to frenemies to more strategic allies serving joint customers amid a spirit of coopetition. The success of that partnership is illustrated by their joint co-selling goals that have more than doubled year over year since 2020, according to both companies. 

“We continue to blow our metrics out of the water,” Sabina Joseph, AWS’ general manager of technology partners, told CRN.

CRN Article, February 2022

The article points out that this relationship hasn’t always been productive. Back in 2015, the two companies were competing directly on deals. Amazon has their own data warehousing solution, Redshift. They would aggressively push this service to customers and SI partners in a winner-take-all strategy to large data workloads. Wins for Redshift would come at the expense of enterprise workload migrations to Snowflake.

In parallel, Google Cloud emerged with an advanced (many consider better) data warehouse solution in BigQuery buttressed by a broader set of machine learning services. These addressed not just traditional analytics workloads, but moved into forward-looking ML and AI driven predictions and optimizations. Leveraging work from their consumer products, Google has demonstrated real technology strength in these capabilities. Not to be left out, Microsoft Azure developed analytics capabilities through Synapse and a sophisticated AI platform.

These efforts were in response to an acceleration in the migration of enterprise workloads for big data and analytics from on-premise data center solutions to the cloud. Enterprises are moving out of legacy data warehouse solutions, like Teradata, in droves. As more of their data sources, like web applications and SaaS solutions, are cloud-based, it makes sense to centralize their back-end data plane onto the cloud as well. This creates a huge opportunity for the hyperscalers to absorb these migrations. They benefit not just from the core data storage and processing, but all the ancillary services associated with ingesting and prepping that data, as well as the machine learning workloads to generate insights from it.

For AWS, it became much more important to win the bundle of application and data services associated with an enterprise cloud migration over the other cloud providers. This meant bringing in best-of-breed independent service providers to round out their capabilities. Sacrificing Redshift usage for Snowflake was a small concession in order to win the majority of spend from an enterprise, particularly where Azure or GCP would lose.

Plus, cloud-based independent providers rely on hyperscaler resources under the covers anyway. Snowflake consumption generates revenue for AWS through the use of compute and storage in each Snowflake cluster. As part of Snowflake’s IPO in 2020, the S-1 disclosed a commitment to spend $1.2B on AWS infrastructure through 2025. That figure is likely higher at this point.

According to the CRN article, AWS and Snowflake shared about 30 joint customer deals prior to 2019. Now, there are many more. Starting in 2019, AWS realized the opportunity to collaborate with Snowflake to win new enterprise customer business. They built out a joint sales strategy that focused on customer use cases. The foundation was to facilitate the migration of large on-premise data processing workloads to AWS. Snowflake was the center of this, with AWS providing all the surrounding services.

To further increase the value of the bundle, Snowflake invested heavily in technology integrations with supporting AWS services. These included AWS Lambda for serverless compute and AWS PrivateLink, which creates private connections to AWS services for enterprise users. Snowflake now has more than 20 integrations with various AWS services. The most significant is with Sagemaker, AWS’ machine learning capability. Since Snowflake doesn’t directly sell a machine learning product, they are happy to let AWS run those workloads. In fact, at last November’s AWS:reinvent conference, Snowflake was the only partner named as part of the launch of Sagemaker Canvas, a no-code machine learning toolset.

This partnership between AWS and Snowflake grew beyond technical integrations of systems. The two companies now have a formal collaboration agreement, in which both companies have committed to increasing investment in partner sales, marketing and alliance teams. This agreement is made actionable each year by annual plans that identify initiatives and set goals for what the joint go-to-market team wants to achieve. These span technical projects and sales targets, including industry verticals, geographic regions and systems integrators. Last year, industry vertical focused go-to-market efforts included media and advertising and financial services.

Systems Integrator Partnerships

Another testament to the tighter relationship between Snowflake and AWS is reflected in feedback from third-party systems integrators (SIs). The SIs are often brought in to provide consulting and engineering resources to execute on a technology project for an enterprise. This is usually because the enterprise doesn’t have the expertise in-house to complete a large cloud migration. In some cases, the customer relationship can even be owned by the SI.

Deloitte is an example of one of these SIs, with a dedicated Snowflake practice. For enterprises considering a data migration to Snowflake, Deloitte can provide architecture recommendations, migration plans, proprietary tools, and of course labor. Prior to 2019, these SIs were often pressured by the AWS sales team to include AWS products as part of the architecture recommendation for a modern data stack. For data warehousing, this would include Redshift.

Since then, however, AWS increasingly promotes a holistic solution in collaboration with the SIs that includes the best from Snowflake and AWS. The intent is to maximize the result for the joint customer, versus maximizing AWS’ share of customer spend. According to the head of cloud analytics at Deloitte, they share a joint go-to-market program built on AWS and Snowflake with a dozen prioritized customers. This collaboration has accelerated in the last 1-2 years.

“What we have seen over the last probably 18 months is almost like a radical reorientation of AWS’ thinking related to us and Snowflake,” Farrall said. “We now have senior account executives from AWS that will call us up and say, ‘Hey, we think that the right answer for this particular client that we’re all looking at…is Snowflake, AWS and Deloitte. Can we partner together? Can we work together on the architecture? Can we build a business case together for that client?’ At first, we were really quite stunned when we started to get this type of a reach-out.”

Frank Farrall, cloud analytics and AI ecosystems leader at Deloitte

The fact that this collaboration has extended out to the third-party SIs signifies that it goes beyond a light co-selling agreement. AWS is actively trying to identify opportunities to bring Snowflake into customer deals. This is because they realize that customers want best-of-breed solutions. By picking the combined expertise from AWS and Snowflake together, the customer feels they are getting better value than if they took the full bundle from GCP or perhaps Azure.

This acceleration of engagement with AWS is likely to keep pushing revenue growth for Snowflake. Having that joint packaging extend to the SIs makes it more tangible and sustainable. Prior to this enhanced partnership with AWS, Snowflake was already performing well. Going forward, co-selling with AWS provides a new tailwind to sustain Snowflake’s growth.

Retail Data Cloud as Example

Beyond collaborative marketing and sales relationships, investment in product integrations provides further evidence of the intent to work collaboratively. If both Snowflake and AWS didn’t see revenue potential in their sales efforts, why would they both commit additional expense towards technology development?

AWS and Snowflake are now identifying industry segments to target with their joint co-selling and solutions development. An example is emerging in retailer data, where both companies can bring unique capabilities. On March 28th, Snowflake launched the Retail Data Cloud. This new industry cloud facilitates data sharing and aggregation relationships between retailers that use Snowflake, which currently span more than 1,000 companies, including customers like 84.51° (Kroger), Albertsons, Kraft Heinz and Rakuten. The Retail Data Cloud allows retailers, manufacturers, distributors, CPG vendors and technology providers to enhance their own data, access new data streams and share data with partners.

The goal of this data ecosystem is to allow participants to create personalized consumer experiences, lower supply chain costs and optimize their business operations. These functions are accomplished with strong security and governance controls, so that privacy is not compromised or competitive advantage lost. Beyond basic data sharing, participants will be able to utilize industry-specific solutions created by Snowflake’s network of partners. These solutions include examples like standard data models, AI/ML-powered insights and industry compliance programs.

These capabilities are all delivered from the Snowflake platform. As I have discussed previously, these data ecosystems create strong network effects, which draw in new participants as the benefits increase. As part of the press release, Snowflake shared a graph which represents data sharing relationships. Each orange dot represents a customer and each blue line represents a stable data sharing edge. In their Q4 earnings report, Snowflake revealed that 18% of all customers now maintain at least one active data sharing relationship, up from 13% a year ago. I think these industry specific data ecosystems will serve to accelerate the creation of stable data sharing relationships.

As you can see in the graphic and as detailed in the press release, Snowflake is bringing together major players in various segments of the retail industry. Participants span retailers, distributors, the manufacturers and third party data providers. This is a brilliant strategy, as each of these companies maintains an enormous data footprint. I can speak from personal experience, having served as VP of Technology previously at data provider Catalina. Catalina runs software that interfaces with the POS system at most large grocery and drugstore chains, capturing anonymized data about every item in the physical shopping carts at these stores in the U.S. and abroad. As you can imagine, that represents an enormous data set.

Introducing each of these companies to the Snowflake data ecosystem only increases the likelihood that they will migrate more of their data storage and workloads onto the Snowflake data platform. Even if they don’t, the data sharing activity in itself will generate more utilization. This includes consumption of credits to process clean room filtering and aggregation, as well as any analytic workloads resulting from a larger data set.

For each industry participant, the application and data hosting associated with the core data collection and processing systems represents significant spend for whichever hyperscaler hosts them. At Catalina, we maintained a sizable infrastructure footprint on two of the major cloud vendors (can’t share specifics) and were continually evaluating which one to favor. As you can imagine, the competitive dynamic between the big three hyperscalers for this kind of business was significant. Each cloud provider employed their own strategies around bundling, discounting and training to land the lion’s share of our infrastructure spend.

And this is the core of the emerging dynamic. While we investors in independent software providers like Snowflake, MongoDB, Datadog, Elastic, Confluent and others focus on the competitive dynamic between each of these and the hyperscalers’ look-alike solutions, we risk missing the larger opportunity. Competition between the big three hyperscalers is even fiercer and constitutes a greater threat for each. Winning a deal for the majority of infrastructure hosting for a Global 2000 company at the expense of another hyperscaler is far more important than having their data warehouse or transactional database solution win business in isolation. This is why we see each hyperscaler trumpet these kinds of encompassing wins in press releases.

So, given that beating another hyperscaler if far more important than beating out an independent software provider, we can start to understand why a hyperscaler would consider collaborating with the independent providers to win deals. Based on my experience at Catalina, AWS or Azure or GCP were far more interested in winning the majority of our infrastructure hosting business over the others. They weren’t concerned if that meant relinquishing the data streaming to Confluent, the data warehousing to Snowflake or the transactional database to MongoDB. They generate some revenue from those third party solutions running on their cloud anyway.

As competition between the hyperscalers intensifies, we have to pay attention to this dynamic. This bias towards cooperation is encouraged by the underlying consumption of compute and storage resources by the independent software providers. It’s no longer a complete loss of the sales value assigned to the independent. Usage by the independents still generates some consumption revenue for the hyperscaler’s lower level compute, storage and network services, assuming the workloads migrate to that hyperscaler. If another hyperscaler wins, they get nothing.

Following Snowflake’s announcement of the Retail Data Cloud, they reinforced their enhanced relationship with AWS. Literally the next day, they announced an additional strategic integration with AWS to improve demand forecasting and delivery for the retail and consumer packaged goods (CPG) industries.

With this announcement, retail brands and CPGs that sell on Amazon will now be able to receive their Amazon PO data natively in Snowflake, leverage brand-level and fulfillment center-level forecasts, and access new data from third-party data providers within the Retail Data Cloud, all on Snowflake’s single, integrated platform. The AWS and Snowflake relationship will empower retail businesses to process, analyze, act on, and syndicate data from a multitude of sources without the delays of traditional methods which require copying and moving data.

Snowflake Press Release, March 2022

If there was any doubt that Snowflake and AWS have formed a more strategic relationship, the timing and positioning of this announcement puts it to rest. AWS is trying to encourage the use of Snowflake for their retail customers, rather than only feeding this data into their data warehouse solution. It also applies to Amazon’s e-commerce solution, which further highlights Amazon’s strategy. They are likely partnering with Snowflake to encourage more CPG companies to utilize Amazon.com for distribution.

The benefit to AWS is clear as well. If we go back to Snowflake’s graphic of the participants in the Retail Data Cloud, we see a number of logos that AWS would love to land for infrastructure services. Snowflake’s network effects created by the data sharing relationships can flow back to AWS. By making it easier to package retail data from AWS and distribute it on the Snowflake Retail Data Cloud, AWS lowers the costs for participants. This provides another benefit to the retail ecosystem participants to shift more software infrastructure to AWS.

The relative revenue numbers for the hyperscalers reinforce this. In Q4, AWS reported revenue of $17.8B, up 39.5% y/y. This translates to a run rate of about $71B. In order to maintain that level of growth in 2022, AWS needs to find around $25B of new revenue. That isn’t going to come from eating into Snowflake’s projected $800M of new revenue in the next 12 months. It has to be driven by winning the majority of cloud infrastructure spend from large customers, at the expense of the other hyperscalers. And, Snowflake utilization generates revenue for AWS anyway. Adding these two factors, we can understand why AWS would radically depart from their previous strategy of competing with the independent providers.

Rather, they are embracing the leading independents, in order to provide potential customers with the best-of-breed suite solutions for all their cloud infrastructure needs. If AWS can assume that they will get the majority of the enterprise customer’s cloud spend, they aren’t worried about giving up smaller slices to the independent providers. Considering the huge spend that AWS needs to capture in 2022, this trickle down to the independents can deliver a material boost to their revenue targets.

MongoDB

MongoDB’s relationship with the hyperscalers has previously been more contentious. This is due to the practice by cloud vendors of offering the open source version of MongoDB as a hosted service on their cloud. This effectively cannibalized MongoDB’s own revenue streams, forcing them to make a license change in 2018 prohibiting this behavior. AWS still offers DocumentDB, but that is pinned to a quite old version of MongoDB.

With the emergence of MongoDB’s cloud-hosted solution, Atlas, the hyperscaler’s perception of MongoDB has improved. Atlas is available on over 80 regions across AWS, Azure and GCP. With Atlas, similar to Snowflake, the hyperscalers generate revenue from the underlying storage and compute, without the overhead of product development resources, customer support and other costs required to maintain their own solution. And, since MongoDB is a transactional database solution, the customer will likely pay for compute resources to run the actual application, even if they push data into MongoDB Atlas instead of DynamoDB or Cosmos.

On MongoDB’s Q4 call in early March, the CEO highlighted their relationships with the hyperscalers.

Raimo Lenschow — Barclays — Analyst

Thank you. Dev, could you speak a little bit to the relationship with the big hyperscalers? It looked like the deals influenced by them kind of shot up quite a bit, like — and obviously, this funny relationship of coopetition. But what you saw — what did you see there in terms of like how we are interacting with you and how that relationship is changing over time? And then, I have one follow-on.

Dev Ittycheria — President and Chief Executive Officer

Sure. The basis of our relation with the cloud providers is really, first and foremost, based on the strong product markets that are of MongoDB. MongoDB is incredibly popular and the pot really spans all major cloud providers I think what we have shown first with Google as we started working with them very closely, given their ambitions to grow their business quickly is that we could partner effectively and help them acquire a lot of new customers, a lot of new workloads onto their platform. This did not go unnoticed by some of the other cloud providers, and we started going deeper with AWS.

As people may remember, in early 2018, AWS introduced a competitor, a clone of MongoDB and know some worries about how that relationship would evolve. And I’m pleased to say that I feel like the relationship has never been stronger. We have deep relationships in the field. We partner more on deals.

And AWS has recognized that MongoDB drives a lot of demand to their platform. And so, the relationship there is very healthy. And we’re also doing a lot of business with Azure. So I would say our win rates are still very high against them when we go head to head against them.

But clearly, they’re good partners, and we’re investing a lot in those relationships.

MongoDB Q4 Earnings Call, march 8, 2022

MongoDB’s relationship with the cloud vendors has more breadth than that of Snowflake. While Snowflake drives the majority co-selling business with AWS and none with GCP, MongoDB is more balanced between the three. The underlying drivers are the same, though. Customers may have a preference for MongoDB’s implementation of the document model and support for an increasing number of workloads. For some companies, MongoDB’s application data platform may be the only database solution they need for application workloads that span multiple data models. This is counter to the offerings from the hyperscalers, which generally address each flavor of data model with an individual product. Readers can review my prior post about MongoDB’s broadening set of workloads and shift in product messaging.

The other driver of enterprises towards MongoDB’s solution is the ability to support multi-cloud deployments. Enterprise IT organizations are sensitive to risks of data lock-in with a particular hyperscaler, knowing that they become increasingly “stuck” as their data footprint on one cloud vendor grows. This inflexibility then reduces their negotiating power and access to the best-of-breed tooling from each cloud vendor. With MongoDB Atlas, data can be shared across multiple cloud vendors, serving application workloads located on each. This mitigates the risk of data lock-in, by allowing data sets to be moved to another cloud vendor if needed.

This flexibility around data location and workloads probably explains the increased interest from the hyperscalers in collaborating with MongoDB. The driver is likely similar to that with Snowflake. The hyperscalers are more interested in winning a major application hosting deal over each other, than having their in-house document-oriented database beat out MongoDB. Co-selling with the MongoDB team may give them an advantage over another hyperscaler. It certainly avoids a disadvantage, if they push an inferior in-house database solution which the customer doesn’t favor.

Similar to Snowflake, MongoDB announced a significant expansion to their relationship with AWS in mid-March. What is interesting are the parallels between the two deals, sharing many of the same drivers. At the core is an effort to get enterprises to migrate software applications off of on-premise data centers to the cloud. The collaboration with MongoDB allows AWS to go-to-market with these customers with a more comprehensive offering. This is aligned against the other cloud vendors, who also want to land these large application workload migrations to the cloud.

It is a six-year agreement, spanning a broad range of tactics. The deal includes shared sales and marketing efforts, developer relations activities, technology integrations and commercial incentives. MongoDB and AWS will work together to develop joint capabilities for customers around serverless, better use of AWS’ Graviton processors and the AWS Outposts service, which extends AWS’ support for on-premise infrastructure. Finally, MongoDB will expand into more AWS Regions globally and Amazon’s US Public Sector hosting with FedRAMP authorization. The latter will open up MongoDB to more government contracts.

Once again, the benefit for AWS is two-fold. First, any use of Atlas still generates revenue for AWS from the underlying consumption of AWS compute and storage resources. Second, if a customer moves an application workload to the bundled offering, AWS can cross-sell their other services to the customer. The press release references analytics, machine learning and IoT services from AWS. All of these represent utilization separate from MongoDB Atlas.

And, like with Snowflake’s announcement, CRN published an article dedicated to the MongoDB / AWS partnership, entitled “AWS-MongoDB Deal Favors Collaboration Over Competition.” This represents a big reversal from AWS’ posture just 3 years ago, where the relationship was purely combative. While AWS will continue to offer their knock-off solution, DocumentDB, it’s clear that they view the bigger opportunity to be gained from their cooperative alliance with MongoDB.

But Chhabra said that over the last year or two AWS has taken a more agnostic approach to which database it sells to customers and, last year, began bringing more deals – especially within enterprise accounts – to MongoDB.

“So, 2021 was the breakthrough year where our sales force and their sales force really started working together,” Chhabra said. “This is a true go-to-market agreement, which basically marries the two companies together to best serve our customers.”

Alan Chhabra, MongoDB executive vice president of worldwide partners

What is most interesting to me about this quote is the timeline. We know that 2021 was the year that MongoDB began accelerating their revenue growth. Q1 FY2022 (Feb – April 2021) delivered 39.4% annual revenue growth. As the year progressed, revenue growth progressed to 44% in Q2, 50% in Q3 and nearly 56% in Q4. Applying a similar beat to the current quarter estimate (Feb – Apr 2022) implies another quarter of mid-50% growth. Granted, MongoDB had some easier comps coming out of calendar year 2020, but the significant acceleration in 2021 wouldn’t be explained by these. The difference is likely the expanding tailwind from AWS co-selling.

Given that this latest expansion in the AWS relationship was announced on March 15th, a week after the Q4 report, it might portend further upside for this year. The inflection from competitor to collaborator with AWS turns the relationship from a headwind into a tailwind for MongoDB. It will be easier to land new accounts and expand spend on existing ones. It also helps that the focus of the new expanded relationship will be on larger, enterprise accounts. These have been the growth driver for MongoDB over the last few quarters, with their Q4 report showing an acceleration of Direct Sales customer additions.

Google Cloud’s Strategy

What is different for MongoDB than Snowflake, though, is that their deeper co-selling relationships are not exclusive to AWS. MongoDB announced an enhanced partnership with GCP in February 2021. CRN also covered this event in a dedicated article. This relationship shares many of the same aspects as that with AWS. It is multi-year, in this case was a 5 year agreement. It encompasses a broad go-to-market relationship, including listing on the Google Cloud Marketplace with integrated billing. This allows customers to receive a single bill for all Google Cloud services, that includes Atlas. They can also apply spending commitments towards Atlas. These attributes make it easy for large enterprise customers to allocate spend towards Atlas.

Similar to AWS, MongoDB expanded the number of integrations with other Google Cloud services, spanning data pipelines, data warehouse, compute and AI engines. Specific services include BigQuery data warehouse, Pub/Sub messaging, Dataflow data processing, Dataproc managed Hadoop and Spark, TensorFlow machine learning, Cloud Run serverless computing, Cloud Functions serverless execution, App Engine platform-as-a-service and the Google Kubernetes Engine container environment. These integrations all make it easier for developers to share data between these services.

As part of the press release, MongoDB’s SVP of Worldwide Partners mentioned that “Year over year, we’ve seen a tremendous increase in joint customer engagements and adoption” and “Our sales collaboration is gathering serious momentum.” Keeping in mind that this agreement was announced in February 2021, it also aligns with the acceleration in revenue growth for MongoDB over the last four quarters.

The Google Cloud press release provides insight into Google’s motivation for supporting this partnership, versus trying to push their own internal database solution. Like AWS, Google Cloud is targeting enterprises making the transition from on-premise to cloud and promoting the full spectrum of their data services. They view the offering as more compelling for enterprises by packaging in MongoDB Atlas, along with their other services.

Additionally, as organizations across the globe are assessing how to leave their on-premise data centers and modernize their infrastructure by moving to the public cloud, we have made it easier for customers and partners to migrate legacy workloads into MongoDB Atlas on Google Cloud.

We have enabled customers to consume MongoDB Atlas along with other Google Cloud services (such as BigQuery and Pub/Sub) to modernize their database stack when they migrate their infrastructure to the cloud. They can also take advantage of state-of-the-art analytics and machine learning capabilities of Google Cloud. Whether it is an industry-specific use case or a strategic technology initiative like Mainframe Modernization, these solutions are proven to help customers significantly boost developer productivity and reduce total cost of ownership (TCO). We have also worked with MongoDB to integrate our core capabilities in security, AI, ML, monitoring, and logging to provide customers with additional value running Atlas on Google Cloud.

Google Cloud Press RElease, February 2021

One notable difference in Google Cloud’s strategy is their public commitment to supporting open source software. This shift was introduced by Thomas Kurian, the new Google Cloud CEO, in 2019. This marked the beginning of MongoDB’s relationship with Google Cloud, starting with a go-to-market partnership through which MongoDB Atlas was offered through the Google Cloud Platform Marketplace. Google has embraced other open source based companies, like Elastic and Confluent. At the same time, GCP isn’t as receptive to partnerships with closed source software providers, like Snowflake. This is particularly acute where Google feels the category is strategic, like analytics and machine learning.

Google’s strategy to partnering with the independent providers became much clearer on April 5th, when they announced the Data Cloud Alliance. This represents a new initiative to “ensure that global businesses have more seamless access and insights into the data required for digital transformation.” At its core is a commitment to support open standards between participants to facilitate data sharing and end-to-end integrated solutions. The subtext is that this effort seeks to eliminate data silos, created by closed systems like Snowflake and the other cloud providers.

The founding members of the Data Cloud Alliance include the commercial organizations behind many of the popular open source projects within the data ecosystem. Notable data storage and management participants are Confluent, Databricks, Dataiku, Elastic, Fivetran, MongoDB, Neo4j, Redis and Starburst. Databricks and MongoDB leadership even have featured quotes about their participation on the announcement page.

With this strategic alignment, we now better understand Google Cloud’s strategy relative to Snowflake and MongoDB. The Data Cloud Alliance clearly positions GCP as a competitor to Snowflake. This is exacerbated by the fact that Databricks is a founding member of the alliance, which most analysts consider to be Snowflake’s biggest independent competitor. This also provides further justification for AWS to strengthen their co-selling relationship with Snowflake, as a counter to Google for analytics and machine learning workloads.

Relative to these moves, MongoDB may find themselves in a sweet spot. They have managed to build a strong relationship with AWS, providing a bundled solution with other AWS services for application workloads at large enterprises. At the same time, MongoDB’s open source roots allows them to participate in the Data Cloud Alliance and potentially garner cross-sell with Google. This dual benefit would extend to other publicly traded, open source commercial entities who have a relationship with AWS, like Confluent and Elastic to a lesser extent.

As an aside, lest we wonder if Databricks is making out well here, Google announced a new product offering the next day called BigLake. BigLake is an extension of BigQuery technology. It provides a storage engine that allows organizations to access data across data warehouses and data lakes. Data teams can apply fine-grained access control, and accelerate query performance across multi-cloud storage and open formats.

Google didn’t go so far as to use the word “Lakehouse” in any of their promotional materials, in case they ruffle feathers over at Databricks. The Databricks Lakehouse Platform is effectively positioned to solve a similar set of problems, allowing enterprises to leverage the best of data warehouses and data lakes in an environment based on open standards and strong governance. Even if Google didn’t use the term Lakehouse, industry news site Protocol did.

Databricks is also an AWS partner, but Snowflake appears to have built a deeper co-selling relationship with AWS. As part of the Google Data Alliance, Databricks will likely garner some co-selling benefit with Google. However, Google’s BigLake product launch clouds this picture a bit. Given the fact that AWS is at least 3x larger than GCP, a stronger relationship with AWS will likely drive more revenue for independent providers like Snowflake.

Getting back to MongoDB, the fact that they enjoy a strong partnership with both AWS and GCP will provide tailwinds going forward. On the Q4 earnings call, MongoDB leadership also mentioned doing “a lot of business” with Azure. In 2019, the two announced the availability of MongoDB on the Azure Marketplace with unified billing.

In this case, Microsoft is cooperating with MongoDB, but doesn’t seem to be actively promoting it. That’s likely an effort to drive business to their in-house NoSQL solution, called Cosmos DB. Cosmos provides open source APIs for MongoDB and Cassandra compatibility. On Microsoft’s recent earnings call, they highlighted the traction of Cosmos DB in their prepared remarks, describing it as “the database of choice for cloud-native app development at any scale.”

Clearing the Competitive Landscape?

Another interesting aspect of these enhanced partnerships between the hyperscalers and independent software providers is that they appear to be anointing leaders in particular categories. Where multiple solutions exist that address the same customer problem, the choice by the hyperscalers of which provider to partner with is likely deliberate and informative. Since the decision to partner at an elevated level involves the commitment of sales, marketing and technical integration resources, the hyperscalers can’t form a strategic go-to-market agreement with every provider.

This enhanced relationship goes beyond simple inclusion in the hyperscaler’s marketplace. That is relatively easy to accommodate and should be table stakes for any independent software provider with a cloud offering. Availability in a hyperscaler’s marketplace does create convenience for developers and can aid in discovery in a limited fashion. However, it represents more of a passive relationship with the hyperscalers, rather than an active one.

Where the real leverage will be realized is for those software providers that strike a broader go-to-market agreement with one or more hyperscalers. These extend beyond simple availability on the hyperscaler platform and listing in the marketplace. They include a commitment to invest resources – in sales programs, incentives and integration with other hyperscaler services. Because of this, the level of scrutiny will be much higher, as the hyperscaler would expect to realize a return on the relationship or consider it as part of a competitive strategy against another hyperscaler.

That makes the distinction of which independent software provider is chosen for the enhanced go-to-market relationship as notable as which ones are not. Logically, the hyperscaler should select the partner that has the largest market share. If they are going to invest resources in the relationship, the partner with the broadest reach will appeal to the most potential customers. However, it can also become a self-fulfilling outcome. By partnering with certain independent software providers, the hyperscalers are concentrating this benefit onto fewer players. This effectively thins the competitive landscape and creates differentiation.

In MongoDB’s case, there are a few commercial offerings that compete for application data workloads to varying degrees. The closest, publicly traded competitor is Couchbase (BASE). Couchbase Cloud is available on AWS, Azure and GCP, including having a marketplace listing on each. The company even highlighted their availability on the Azure Marketplace in a separate press release. However, there are no press releases announcing a more strategic collaboration with the hyperscalers, involving joint co-selling or go-to-market efforts. This level of relationship was struck between MongoDB and AWS and GCP.

This implies that MongoDB will likely benefit from enterprise cloud migration deals that involve the hyperscaler sales teams, over an alternative solution. If anything, MongoDB will be part of the conversation. It’s possible the engineering team at the customer organization already knows what commercial database solution they prefer for their varying workloads. If they don’t, MongoDB has the opportunity to sell them on the benefits of their Application Data Platform as a general purpose database that can address most of their needs. This might come at the expense of not just Couchbase, but other variants like Redis, CockroachDB, InfluxDB or Neo4j. None of these private companies appear to have a strategic partnership in place with the hyperscalers, aside from Google’s recently announced Cloud Data Alliance, in which MongoDB is a featured partner.

Even in the Google Data Alliance, Couchbase is not included. This is in spite of the fact that Couchbase Server has an open source foundation (Membase). Couchbase is available on the GCP Marketplace. But, they didn’t make the cut for the elevated level of partnership with GCP or AWS. This ignores their claims of a superior product architecture to MongoDB, highlighting the importance of developer relations and partnership development.

In addition to mitigating the competitive threat from the hyperscaler’s internal look-alike solutions, these strategic go-to-market relationships also give preference to one independent provider over the others in certain categories. This effect provides another tailwind for those providers selected for this benefit. It doesn’t guarantee that MongoDB will become the transactional data store for every new cloud workload, but it does help grease the tracks and validates them as the leader in the space. As we look at other software infrastructure categories, the singularity of these strategic partnerships should be considered.

Datadog

In some ways, Datadog has had the easiest time in their relationship with the cloud vendors. For the most part, the hyperscalers haven’t developed and promoted robust observability solutions. They each offer bare bones tooling for monitoring usage of compute and storage infrastructure on their cloud, but these offerings haven’t evolved into a full-fledged observability platform with all the bells and whistles. That may have been a deliberate strategy on the part of the hyperscalers, as they focused on services closer to the core of application hosting.

AWS offers CloudWatch, which is buried in its Management and Governance category. CloudWatch collects monitoring and operational data in the form of logs, metrics, and events.  But, it is only integrated with AWS services, rendering it largely useless as a holistic solution for any large enterprise. Interestingly, AWS also offers hosted versions of open source DIY alternatives to Datadog, Grafana, Prometheus and OpenTelemetry.

Microsoft offers Azure Monitor, which is similar to CloudWatch in capabilities. It also integrates with Azure services. Like the offering from AWS, Azure Monitor can feed other popular observability tools, like Datadog. Google Cloud offers an Operations suite which includes solutions for logging, infrastructure metrics and traces. Like the other hyperscalers, this is a basic offering that works primarily with Google Cloud services.

Datadog has been largely left alone from a competitive perspective by the hyperscalers. This creates a much different dynamic than we see with Snowflake and MongoDB. On one hand, the lack of competition has provided Datadog with more opportunity to pursue its TAM without encroachment from the hyperscalers. Datadog has competition from other independent providers, like Dynatrace, Splunk and New Relic, but at least doesn’t have to worry about the double impact of a hyperscaler service offering.

On the other hand, since none of the hyperscalers view observability as a strategic product within their own portfolios, they are less likely to promote it as part of a packaged offering to compete with another hyperscaler. This is different from the strategy that we see AWS employing with Snowflake as an example, where they are using Snowflake as a counter to GCP to win large enterprise cloud migration deals. Datadog provides little leverage in those competitive deals.

That’s not to say that the hyperscalers haven’t formed partnerships with Datadog. They have. But, these are largely focused on including Datadog in their Marketplace and making billing easier. These arrangements benefit Datadog with SMB customers, who likely are seeking solutions in the Marketplace. For large enterprises, though, I don’t see the hyperscaler sales teams promoting their relationship with Datadog as part of a competitive deal to win a large cloud migration deal. Yes, observability is a critical component of a cloud migration, but the hyperscaler sales teams would largely expect the customer to find Datadog on their own.

This isn’t a knock on Datadog. It simply means that the look-forward contribution from the hyperscalers will likely be consistent with past behavior. We won’t see an acceleration in new revenue generated by hyperscaler co-selling, like we can anticipate going forward for Snowflake or MongoDB. For Datadog, this is less the case where a headwind is turning into a tailwind, as there really wasn’t a headwind in the first place.

To see this distinction, let’s briefly look at the recent references to collaboration between Datadog and the hyperscalers. On the Q4 earnings call in February, Datadog’s CEO called out the strong relationship with AWS in his prepared remarks. This is a reference to the Global Strategic Partnership announced in early January. Two aspects of that partnership are exciting for Datadog. First, it includes more extensive “joint marketing and co-selling programs” between AWS and Datadog (quote is from SVP at AWS).

We also announced a global strategic partnership with AWS. This is a recognition of our success and growth with AWS and our commitment to further invest to accelerate our joint opportunities. Among the areas of further partnership, we have already integrated Datadog more tightly into the AWS marketplace. We are also working with AWS to build deeper integrations not only for observability, but also for security use cases, and we are also planning to extend our joint go-to-market activities.

Olivier Pomel, Datadog cEO, Q4 EArnings Call

Second, the relationship extends into security use cases. Datadog’s entry into security started about two years ago, but they are already gaining significant customer traction on security product adoption. In the prepared remarks, the CEO highlighted “thousands” of customers using their cloud security products. The inclusion of security use cases in the extended relationship with AWS should help to further accelerate uptake.

In late March, Datadog announced its inclusion as a partner in Microsoft Azure’s Cloud Adoption Framework. The framework provides Microsoft customers with best practices, tools and documentation to help with their migration to the cloud. As part of this, Azure customers can use Datadog observability and security capabilities.

This supplements Datadog’s existing partnership relationship with Microsoft Azure, announced in August 2021. This arrangement added Datadog directly to the Azure portal. It facilitates easier purchasing, configuring, and managing of Datadog services directly inside the tooling that Azure customers to use manage their other Azure services. Customers can purchase Datadog through the portal and consolidate billing. They can install the Datadog agent on Azure hosts quickly and send logs/metrics to Datadog’s platform. Access can be configured through single sign-on and users can view all Azure resources that are being monitored by Datadog. This represents a much deeper level of integration than that available form the Azure Marketplace.

To round out the trifecta, Datadog announced its availability on the Google Cloud Marketplace back in July 2021. This included the standard convenience of easy installation and consolidated billing. It also references deeper integrations between Datadog connectors and various Google Cloud services. Finally, the press release references “extended go-to-market collaboration and deeper sales alignment with Google Cloud and Datadog sales teams.” This implies more than just a listing in the marketplace.

While Datadog is getting the benefit of co-selling relationships with all three of the major cloud providers, we should also examine which of Datadog’s competitors have been elevated to this higher level of partnership. This dampens the impact of Datadog’s relationships somewhat, as a similar benefit appears to extend to other observability providers. This isn’t to say that the collaboration agreements aren’t valuable for Datadog or the other players. They just don’t reflect a significant shift in hyperscaler disposition over the prior 12 months, like we see with Snowflake and MongoDB.

Dynatrace

• AWS. Announced expanded strategic relationship on March 31, 2022. Includes go-to-market and product integrations. Notably, doesn’t call out security features as distinctly as Datadog’s announcment.
• GCP. Announced expanded strategic partnership on October 21, 2021. References collaboration on observability, runtime application security and AIOps.
• Azure. Announced expanded collaboration on September 9, 2021. Includes Azure Marketplace, consolidated billing, Azure Portal access. Reads very similar to the Datadog announcement.

Splunk. I couldn’t find any press releases from Splunk announcing a strategic partnership with any of the cloud vendors. Splunk is available on the various marketplaces for each of AWS, Azure and GCP.

New Relic.

• AWS. Announced an expanded strategic collaboration agreement back in October 2020. Includes product integrations, development and joint go-to-market activities.
• GCP. No press releases for an expanded relationship. They do participate in the marketplace.
• Azure. No press releases for an expanded relationship. They do participate in the marketplace.

Among observability competitors, then, we see some variation in the level of relationship with the cloud providers. But, inclusion and exclusion don’t appear as distinct as for transactional databases and data warehouse. Datadog and Dynatrace seem to have the deepest level of partnership across all three hyperscalers. New Relic has some and Splunk appears to have the least.

These relationships provide tailwinds for Datadog and Dynatrace, as the cloud vendors continue to drive enterprises to cloud migrations and digital transformation. The fact that the hyperscalers largely do not offer a competitive solution in observability has also been an ongoing benefit. However, the contribution of new sales for each provider driven by the hyperscaler relationships is likely steady state, versus the acceleration in co-selling noted by other independents like Snowflake.

Others

I could go on about other independent software providers highlighting an improved relationship in co-selling with the big cloud vendors. The list includes at least Crowdstrike, Elastic, Confluent, Zscaler and likely several more. The key consideration is that this has been broad-based and largely led by AWS. I think we are seeing a significant shift in strategy at AWS, which is undergoing the most notable change over the past 12 months. Given their size in the market, this could provide a major tailwind for those independent providers with the optimal alignment to AWS.

Crowdstrike

I was surprised to hear the reference from Crowdstrike to AWS tailwinds in their most recent earnings call. This hadn’t been a significant topic previously. It likely reflects AWS’ recognition that security is a strategic asset and desire to partner with leading providers to counter offerings from the other hyperscalers. This is targeted at Microsoft Azure, which offers their own endpoint protection solution. Like AWS and Snowflake as a counter to GCP for big data, AWS is partnering with Crowdstrike to beat out Microsoft for deals that involve security as a critical component of an overall cloud hosting package.

One partner, I’d like to highlight is AWS. In fiscal 2022, ending ARR transacted through the AWS marketplace grew more than 100% year over year. Furthermore, CrowdStrike ended the year as one of the top ISV partners by transaction volume on the AWS Marketplace, with partner source deals growing strongly throughout the year. We believe this speaks to the success of our partnership with the world’s largest public cloud provider and highlights the value we can provide to both partners and customers alike.

George Kurtz, Crowdstrike CEO, Q4 Fy2022 Earnings CAll

During their Investor Briefing on April 7th, Crowdstrike leadership provided more detail on growth in ARR generated from the AWS Marketplace. This was part of a record year for partner-sourced revenue for Crowdstrike. ARR from the AWS Marketplace now well exceeds $100M and grew more than 100% over the prior year. This is much faster than the overall growth in Crowdstrike ARR during the same period of 65%, demonstrating that AWS sell-through will exert an upward influence on overall revenue growth.

On the earnings call, the CEO also referred to Google as “another big partner of CrowdStrike.” This relationship may be clouded a little bit by Google’s planned acquisition of Mandiant. That acquisition could be additive for Crowdstrike, or represent a first step by Google to offer more of their own security products, similar to Microsoft’s product strategy.

To round out the large public cloud vendors, Crowdstrike and Microsoft are well-known competitors, often taking public jabs at one another. Therefore, we wouldn’t expect any collaborative selling to occur between Azure and Crowdstrike. It is interesting that in spite of this, Crowdstrike does have a listing on the Azure Marketplace. This highlights the distinction between having a listing on a cloud vendor’s marketplace and actively driving revenue for the independent software provider through joint go-to-market efforts.

The relationship between Crowdstrike and the cloud vendors is similar to that of Snowflake. Crowdstrike competes directly with Microsoft and is used by AWS and GCP to win customer deals over Azure. Snowflake competes with Google BigQuery, and is used by AWS primarily and Microsoft somewhat to win business over GCP.

Confluent

Confluent shares a productive relationship with all three hyperscalers. In these cases, they occupy the primary position in their category of data streaming. The cloud vendors do offer alternative in-house products for data streaming, but it is becoming clear that they realize the incremental benefits to customers of partnering with Confluent on large data migration deals. Additionally, Confluent is a founding member of Google’s new Data Cloud Alliance, providing further credibility for their category leadership.

• AWS (Jan 27, 2022). Marketplace, consolidated billing, multiple service integrations through connectors, co-selling
• Azure (April 5, 2022). Marketplace, consolidated billing, multiple service integrations through connectors, co-selling
• GCP (April 5, 2022). Founding member of the Google Data Cloud Alliance.

Why the Change?

As I have discussed, I think there are several reasons that the hyperscalers have pivoted into cooperation mode with the independent providers.

Shared Benefits. As the independent providers build their software solutions on the cloud, they are choosing to locate them on an existing hyperscaler’s infrastructure. The larger independents run on at least AWS, Azure and GCP. This makes sense, as they can reach over 60% of the market by targeting these providers. The benefit to the hyperscaler is that they generate revenue from the independent’s underlying consumption of compute, network and storage resources. As the independent makes money, the hyperscaler makes money.

Additionally, this revenue is low cost for the hyperscaler. The independent is doing most of the work. They build and maintain the software. They provide the customer service, documentation and onboarding. The hyperscaler may expend some sales resources to help close the deal, but this is a minimal lift and is sometimes bundled with other services exclusive to the cloud vendor.

Cross-sell. When the hyperscaler brings an independent into a new deal, there are often a slew of other services they are selling the customer besides the slice addressed by the independent. Snowflake may handle the data warehouse, but Amazon Sagemaker processes the machine learning jobs. Datadog may be the observability solution, but AWS EC2 hosts the application code. Same scenario applies for MongoDB as the database solution. In these cases, the hyperscaler is bringing the customer a best-of-breed bundle, of which they are still getting the lion’s share of spend.

Competition between Hyperscalers. While AWS is still larger than Azure and GCP, over the past couple of years, the smaller two have been growing their revenue at a faster rate. Some analysts have speculated that Azure might pass AWS in size eventually. Not to be outdone, GCP has been employing very aggressive tactics like discounting and bundling to win business.

This competitive pressure may well be the catalyst behind AWS’ about face. At an analyst conference, the CFO from Snowflake discussed how AWS and Snowflake together have won hundreds of deals over Google BigQuery. As I have discussed, AWS benefits from this collaboration on two levels. They generate revenue from Snowflake’s use of compete and storage. Second, they sell other services, like Sagemaker or data pipeline solutions around the data warehouse. This compromise is far better than $0 revenue, if they lose the entire customer deal to GCP.

This need to win large enterprise deals not only pits the hyperscalers against one another, but also positions them against the long tail of smaller cloud providers that make up the other 36% of the cloud infrastructure market. In order for AWS, Azure and GCP to maintain revenue growth of 40% or more, they have to pull back some share from these smaller providers. The fact that the major independent service providers generally host on the big three hyperscalers exclusively further exacerbates this dynamic.

Types of Relationships

Looking at the examples with the companies discussed, we are seeing three types of relationships with the hyperscalers emerge. I think these will have varying impact on revenue growth for the independent providers going forward. By considering how any independent software provider falls into these categories, investors can discern how strong the tailwind will be for revenue growth going forward.

Competitive Wedge

Description: This is the case where one of the cloud vendors forms a strategic partnership with an independent provider to form a combined solution to take to market. This is positioned against another cloud vendor’s solution in order to win the majority of cloud spend from a large enterprise customer. In this case, the cloud vendor is sacrificing the sale of their in-house solution to the independent’s benefit, in order to win the larger deal. The cloud vendor still benefits from the underlying compute/storage utilization of the independent’s platform.

Examples: Snowflake and AWS versus Google Cloud Platform. Crowdstrike and AWS versus Microsoft.

Expected Effect: Strong. I think the impact of a competitive wedge relationship, particularly with AWS, will have the most near term net benefit in incremental revenue growth for the independent provider. We see this clearly with Snowflake, where over half of their new bookings were generated through cloud vendor co-selling. AWS sourced the majority of that.

Anointed Leader

Description: In this case, the cloud vendors offer a competitive product, but consider the strategic relationship with the independent as an opportunity to appeal to a broader set of enterprise customers. They may use the co-selling relationship as an opportunity to stitch together a more appealing package of services for the customer, leveraging an existing preference by the customer for the independent’s solution. The independent will also generally have a relationship with all the cloud vendors, or at least lacks an adversarial relationship with one of them.

Because the cloud vendor is somewhat dependent on the popularity of the independent provider, they are more selective in the relationships they form. They may pick only one provider in a category, which effectively crowds out the smaller competitors of the anointed independent. Those smaller competitors may have a marketplace listing, but aren’t top of mind with the cloud vendors’ sales teams. This effectively narrows the competitive field in that category, relative to other commercial or even open source alternatives.

Examples: MongoDB, Confluent

Expected Effect: Medium. Because the cloud vendors aren’t using their relationship with the independent providers as a competitive wedge, they won’t actively drive as much business. But the scope of impact extends across all cloud vendors. The independent also receives the benefit of being designated the winner in their space. The cloud vendors’ decision to partner with a particular independent provider over the others may signal to customers some sort of perceived advantage or sustainability with that provider. While this relationship may not drive as much direct revenue as the competitive wedge, it narrows the competitive field to a single provider.

Collective Tailwinds

Description: This is the case where none of the cloud vendors offer a material competitive product and they don’t consider the partnership to be a driver of differentiation from the other cloud vendors. The cloud vendors form partnerships with multiple providers in the category, not really narrowing the field for customers.

Examples: Datadog

Expected Effect: Low. While it’s great that Datadog and Dynatrace have a strategic collaboration in place with the hyperscalers, that relationship doesn’t appear that it will attribute incremental gain to one over the other. Each does benefit from the fact that the cloud vendors don’t offer a competitive product and partner in order to provide their customers with a holistic offering. The cloud vendors also realize the compute/storage consumption from those vendors who are cloud-based, reducing the incentive for them to develop a directly competing product. The tailwinds to Datadog and Dynatrace from cross-sell with the hyperscalers are a net benefit to them, but the trajectory of this impact hasn’t changed over the last 12 months. This is different from the rapid and concentrated shift in disposition for other independents, like we are seeing with the competitive wedge beneficiaries with AWS in particular.

Investor Take-aways

I think the shift from competition to cooperation with the hyperscalers is one of the most important changes to take place for the independent software providers in some time. Investors should understand the potential benefits of this pivot, as it stands to influence the growth trajectory for several independent software providers. As we consider their durability of revenue growth going forward, the relationship with the hyperscalers for some is transitioning from a revenue headwind to a tailwind.

In the past, as hyperscalers introduced competing products and bundled them into deals, we could expect some percentage of business to be peeled away from the independents. This wouldn’t represent a huge drain, but always provided a headwind on both growth and costs. Independents might lose 10-20% of potential revenue a year due to hyperscaler encroachment. Additionally, sales teams would require more investment to counteract these deal risks and ensure that potential customers fully understood the superiority of the independent company’s offerings.

With collaboration, these drags on growth and profitability reverse. Now, independents might expect to win more deals through co-selling with the hyperscalers. Cost of sales also diminishes, as the hyperscaler is bringing the independent customer leads and providing promotional support on their marketplaces. These factors produce a tailwind. High revenue growth will be easier to sustain, at least in the near term. And independents can either pocket the sales efficiencies or redirect those funds towards customer expansion initiatives in other areas.

This comes at a time when the high growth software infrastructure companies are being scrutinized for both durability of revenue growth going forward and profitability. The emergence of strong co-selling relationships with the hyperscalers can help avoid deceleration of revenue growth or erosion of operating margins. In the near term, it might even generate a little acceleration.

With these effects in mind, I have examined in depth the evolving relationship between several independent providers and the hyperscalers. This analysis elicited a framework for considering the potential impact to the independent software providers across three rough groupings. Depending on where an independent providers falls, investors can anticipate the potential impact to their revenue growth going forward.

This effect is particularly acute where the hyperscaler is using the independent provider as a competitive wedge to win enterprise business over other cloud vendors. This shift in bias for AWS in particular over the past 12 months is noteworthy. The revenue impact has inflected from a negative to a positive very quickly. As we consider the durability or even acceleration of revenue growth going forward for select providers, this inflection in revenue contribution could be material.

NOTE: This article does not represent investment advice and is solely the author’s opinion for managing his own investment portfolio. Readers are expected to perform their own due diligence before making investment decisions. Please see the Disclaimer for more detail.