GCP’s drive for OPEX optimization and low-cost leadership is not a strategy to extract value from enterprise accounts. GCP has focused on a small handful of verticals to create enterprise solutions value. The combination is probably not enough to maintain its third-place public cloud revenue position against aggressive competitors.
Google Cloud revenue was $2.61B in 4Q 2019, up 53% YoY from $1.71B in 4Q 2018, and $2.78B in 1Q 2020, up 52% YoY from $1.83B in 1Q 2019. (Alphabet started breaking out Google Cloud revenue as a separate line item in 4Q 2019, so there’s currently a reporting gap for 2Q and 3Q 2019.)
YoY growth of 53% for 4Q 2019 and 52% for 1Q 2020 sounds impressive, but it doesn’t convey the whole picture.
Google Cloud revenue includes both GCP and G Suite. Alphabet emphasized during its 1Q 2020 earnings call that it saw “significant” growth in GCP, above the YoY average for Google Cloud, while G Suite saw merely “solid” growth.
The challenge for Google Cloud will be to position GCP as an enterprise-class IaaS solution. Its competitors are doing a better job of offering cloud customers easy paths to control, predictability and consistency. But during its 1Q earnings call, Alphabet indicated that GCP will reduce new data center build-out in favor of driving increased existing infrastructure optimization.
As enterprise competition heats up and as pandemic-driven remote work and remote education fuel increased G Suite revenue, mid-2020 Google Cloud revenue growth may shift its mix between GCP and G Suite.
GCP’s competitive challenges are mostly due to data center architecture choices.
Google Cloud Platform (GCP) offers a much simpler menu of Infrastructure-as-a-Service (IaaS) products than the rest of the top clouds, Alibaba Cloud, Amazon Web Services (AWS) and Microsoft Azure. However, underneath that simplicity lies hidden complexity that handicaps GCP in mainstream enterprise IT. GCP started out with fundamentally different choices from its competitors and continues maintain its differentiation.
IaaS architecture walks a tightrope between simplicity and complexity.
Alibaba Cloud, AWS and Azure each give cloud customers a lot of detail up-front, but GCP does not. The impact is that customers may spend more time understanding low-level configuration options or experiencing higher service or cost variability with apps in deployment.
GCP’s relatively simple IaaS service menu stands out. Though GCP’s IaaS complexity more than doubled (128% growth) over the past three quarters, that growth was from a much smaller base (see chart below). Liftr Insights measures IaaS complexity by counting the number of each cloud’s rentable configurations in each of its operating regions. The other three clouds ranged from 35% to 68% total growth (compound monthly growth rates are shown in the chart).
GCP’s IaaS complexity growth was much slower over most of the past few quarters. More than half of GCPs complexity growth over the past three quarters occurred in March 2020, with GCP’s production deployment of two new cost-conscious high-level instance type families.
However, most of GCP’s new type family deployments can be scheduled on GCP’s current infrastructure, so that increase in complexity requires minimal new data center infrastructure build-out.
As of March 2020, the top clouds had similar geographic coverage. GCP had announced several upcoming new regions, which will require new data center build-out (table below).
Alphabet mentioned variants of GCP data center operations “efficiency” nine times during its 1Q 2020 earnings call. It also stated that it would shift the mix of its infrastructure investment in 2020 “with relatively more spend on servers than on data center construction.”
I believe GCP will likely delay deploying some of its announced regions in 2020. There are no publicly committed dates, so some of these regions are likely to slip into 2021.
GCP’s March 2020 type family deployments are great examples of offering more IaaS products that run on existing data center infrastructure to increase efficiency, reduce OPEX and boost utilization rates for GCP’s existing sunk CAPEX.