How to get away from financing your competition
Many companies, both small and large, both tech and non-tech sooner or later realize that they are buying services from their competitors effectively quickening their demise.
The major cloud providers Amazon, Google or Microsoft all compete on several hundred verticals. Chances are at least one of them has a product in your industry or will soon.
For example Amazon has business in: Financial Services, Payment (Amazon Pay, Amazon Cash etc), Retail (though WholeFoods acquisition), Pharma (through PillPack), Databases (Redshift, Aurora), Servers (though AWS Outpost) and many other IT related services to name a few.
Why do so many companies continue to give even more financial ammunition to their competitors? Amazon is already a $386bn business. Some friendly coopetition is useful in increasing overall operating efficiency but this is a very lopsided relation here, even between giants.
According to Flexera 97% of companies use at least one cloud and 50% of those use AWS to some degree. Thus chances are a lot of companies are financing their competition, especially tech companies.
The attraction of the cloud
Most companies have the famous "Cloud First" strategy which promises the means to a variety of business benefits:
- Less resource waste through ease of provisioning and de-provisioning of resources.
- Lower time to market due to developers self-serving themselves off compute resources and the use of read-to-use platform services such as databases as a service.
- Opex rather than Capex allows companies to shutdown resources used by failed initiatives.
The danger of financing your competition
AWS has grown by selling to startups. For many years no enterprise would have touched a service with so much downtime and so many usability issues and accidental data loss.
But since startups are no longer needed, Amazon can now afford to shrug off open-source companies and actually compete with them.
AWS's DocumentDB service was created specifically to avoid paying license fees to MongoDB.
While MongoDB is growing revenue with 38% YoY in Q3 2021 ($576.4m) it's operating at a net loss of almost 10% of revenue (-$54.6m).
Amazon is eating into MongoDB's margins simply by competing with them and - ironically - by effectively forcing MongoDB to buy its products due to a network effect: AWS customers already being AWS customers demand data locality hence a service in AWS.
The question is wether MongoDB will be able to support the net loss for long and what would they do when they cannot anymore. Perhaps cut back on sales and marketing and Research and Development.
Use on-prem + cloud burst as an alternative
In an interview for Protocol, Target's CIO McNamara was describing that Target was once all-in in AWS but with the acquisition of Whole Foods, Target decided they need to "take engineering back in-house".
Target is able to run "very efficiently" in their own datacenters for most of the year. By our calculations it's at least 2-6x less expensive than in the cloud.
McNamara talks about seeing "20 times the normal volume" around Thanksgiving and Cyper Monday. Target bursts into Google Cloud and Azure IaaS services during these very short peak periods.
Among other technologies, some of which built in-house, Target uses Kubernetes which acts like a compatibility layer between all environments and allows for easier deployment.
"I can take a workload and I can choose where I want to deploy it. I can choose to deploy it in [Google Cloud Platform]. I can choose to deploy it on Azure, or as I say, I can choose to deploy it on Target assets." says McNamara
This flexibility is key to enable Target to be both very efficient with the use of resources and agile at the same time.
The cloud costs 2x-6x more than on-prem
A TCO comparison between cloud and on-prem for infrastructure services has been done many times and by many companies including Amazon itself via IDC. Although the first page numbers tout 51% lower cost of operations, most of the cost benefits of AWS come from Licensing costs and Staff efficiency increases so not the actual infrastructure costs.
We ran the numbers ourselves. Assuming no licensing costs (we know that Target uses the "open-source version of Kubernetes everywhere") we get between 2.28x to 5.68x more efficient running on-prem than running in AWS.
* We have used colocation prices but we can safely assume that Target's own facility's costs are about the same or less.
The strategy pays off
Target has been clearly been doing well according to the Q4 2020 earnings report.
Target's 2020 sales growth of more than $15 billion was greater than the Company's total sales growth over the prior 11 years.
Comparable sales grew 19.3 percent, reflecting 7.2 percent growth in store comparable sales, and 145 percent growth in digital comparable sales.
Target's digital sales grew by nearly $10 billion in 2020, driven by 235% growth in the Company's same-day services.
So clearly, this strategy has not hindered Target's ability to move fast and innovate. What's also more important is that is has done so by leveraging the strengths of cloud and on-prem in equal measure without being forced to finance the competition.