However, as industry experience with the cloud matures — and we see a more complete picture of cloud lifecycle on a company’s economics — it’s becoming evident that while cloud clearly delivers on its promise early on in a company’s journey, the pressure it puts on margins can start to outweigh the benefits, as a company scales and growth slows. Because this shift happens later in a company’s life, it is difficult to reverse as it’s a result of years of development focused on new features, and not infrastructure optimization. Hence a rewrite or the significant restructuring needed to dramatically improve efficiency can take years, and is often considered a non-starter.
Now, there is a growing awareness of the long-term cost implications of cloud. As the cost of cloud starts to contribute significantly to the total cost of revenue (COR) or cost of goods sold (COGS), some companies have taken the dramatic step of “repatriating” the majority of workloads (as in the example of Dropbox) or in other cases adopting a hybrid approach (as with CrowdStrike and Zscaler). Those who have done this have reported significant cost savings: In 2017, Dropbox detailed in its S-1 a whopping $75M in cumulative savings over the two years prior to IPO due to their infrastructure optimization overhaul, the majority of which entailed repatriating workloads from public cloud.
To dimensionalize the cost of cloud, and understand the magnitude of potential savings from optimization, let’s start with a more extreme case of large scale cloud repatriation: Dropbox. When the company embarked on its infrastructure optimization initiative in 2016, they saved nearly $75M over two years by shifting the majority of their workloads from public cloud to “lower cost, custom-built infrastructure in co-location facilities” directly leased and operated by Dropbox. Dropbox gross margins increased from 33% to 67% from 2015 to 2017, which they noted was “primarily due to our Infrastructure Optimization and an… increase in our revenue during the period.”
The exact savings obviously varies company, but several experts we spoke to converged on this “formula”: Repatriation results in one-third to one-half the cost of running equivalent workloads in the cloud. Furthermore, a director of engineering at a large consumer internet company found that public cloud list prices can be 10 to 12x the cost of running one’s own data centers. Discounts driven by use-commitments and volume are common in the industry, and can bring this multiple down to single digits, since cloud compute typically drops by ~30-50% with committed use. But AWS still operates at a roughly 30% blended operating margin net of these discounts and an aggressive R&D budget — implying that potential company savings due to repatriation are larger. The performance lift from managing one’s own hardware may drive even further gains.
Actual spend as a percentage of COR is typically even higher than committed spend: A billion dollar private software company told us that their public cloud spend amounted to 81% of COR, and that “cloud spend ranging from 75 to 80% of cost of revenue was common among software companies”. Dullien observed (from his time at both industry leader Google and now Optimyze) that companies are often conservative when sizing cloud commit size, due to fears of being overcommitted on spend, so they commit to only their baseline loads. So, as a rule of thumb, committed spend is often typically ~20% lower than actual spend… elasticity cuts both ways. Some companies we spoke with reported that they exceeded their committed cloud spend forecast by at least 2X.
Among the set of 50 public software companies we analyzed, the average total enterprise value to 2021E gross profit multiple (based on CapIQ at time of publishing) is 24-25X. In other words: For every dollar of gross profit saved, market caps rise on average 24-25X times the net cost savings from cloud repatriation. (Assumes savings are expressed net of depreciation costs incurred from incremental CapEx if relevant).
This means an additional $4B of gross profit can be estimated to yield an additional $100B of market capitalization among these 50 companies alone. Moreover, since using a gross profit multiple (vs. a free cash flow multiple) assumes that incremental gross profit dollars are also associated with certain incremental operating expenditures, this approach may underestimate the impact to market capitalization from the $4B of annual net savings.
Think about repatriation up front. Just because the cloud paradox exists — where cloud is cheaper and better early on and more costly later in a company’s evolution — exists, doesn’t mean a company has to passively accept it without planning for it. Make sure your system architects are aware of the potential for repatriation early on, because by the time cloud costs start to catch up to or even outpace revenue growth, it’s too late. Even modest or more modular architectural investment early on — including architecting to be able to move workloads to the optimal location and not get locked in — reduces the work needed to repatriate workloads in the future. The popularity of Kubernetes and the containerization of software, which makes workloads more portable, was in part a reaction to companies not wanting to be locked into a specific cloud.