Last week I hosted a panel with a team of technology analysts at a global bank along with hundreds of their clients. We talked about a range of topics in the cloud computing space. During the call, I was asked what I thought about Diane Greene’s comments that Google Cloud may be able to catch AWS in five years. I was caught off guard by the question as I hadn’t yet seen the coverage from the Forbes CIO event from the previous day where she made the comments.
— Forbes (@Forbes) April 24, 2017
My reaction was that it would be difficult for GCP to catch AWS in five years from a revenue standpoint. AWS holds a large lead in not only revenue, but my answer focused on a range of commercial metrics like number of clients, number of AWS certified engineers at both clients and partners, scale of partner (software & services) ecosystem and size of sales & client support teams by internal employee count; proxy can also be number of open headcount in these respective areas.
There were a lot of questions that people asked me during this panel. This ‘Google catching AWS’ is the one that I thought the most about after the panel finished. I realized that my answer was less about catching AWS on the technology side and more focused on the people and momentum side.
Assuming there is no large scale security event on AWS, or other Black Swan for Amazon, that causes disruption to AWS organic growth, I think that it is possible for GCP to be in striking distance within five years and potentially be able to overtake AWS in six to eight years.
The below are areas I would focus on to execute a five year game plan for GCP to catch AWS.
There are currently 5,900 open roles at AWS. In order to catch AWS, Google Cloud will have to scale their teams to a similar magnitude of sales, product, services and client support headcount while also maintaining a high hiring bar. Can GCP offer hiring incentives to AWS, and cloud-native, employees to join them at a rate faster, with higher quality, than AWS?
Client & Community Expertise
AWS currently has about 40,000 certified architects and developers worldwide. I have not seen a recent estimate of GCP certified engineers but this number has to come close to parity. Can partners, and potential clients, be incented (cash, Google advertising, rebates, GCP credits, etc.) to achieve these certifications at a rate that is faster than AWS certified engineers over this time period? We wrote here about why we thought the Qwiklabs acquisition would help with engineering certification last year and I would study making more training acquisitions (as well as training company acqui-hires) to further accelerate.
Partner Program Incentives
In addition to incentivizing partners to get certified on the Google Cloud Platform, Google is going to need to scale through partners in general over this time period at a rate faster than AWS. Therefore the third-party partner network of software and services partners is crucial for the acceleration plans of Google Cloud Platform in my opinion.
I wrote earlier this year during Google NEXT that I would also like to see Google Cloud consider an example from its Adsense business with respect to the Google Cloud reseller model. In Google Adsense the majority of the revenue per ad on a website displaying Google Ads is paid to the 3rd party website. In public cloud, the majority of the reseller revenue goes to the public cloud provider and not the cloud services company getting Google’s clients to its cloud. This leaves cloud services as the lion’s share of margin for non-software companies in the partner ecosystem. If you are a services company that has a client that already is committed to going to AWS, what is the incentive to educate the client on choosing GCP instead?
While increasing partner program incentives, Google should also continue to take the lead on broad, partner-enabling platforms such as the Open Service Broker API project. The Open Service Broker API project allows developers, ISVs, and SaaS vendors a single, simple, and elegant way to deliver services to applications running within cloud native platforms such as Cloud Foundry, OpenShift, and Kubernetes. The project includes individuals from Fujitsu, Google, IBM, Pivotal, RedHat and SAP.
Kubernetes is an open-source system for automating deployment, scaling, and management of containerized applications developed by Google and was donated by Google to the Cloud Native Computing Foundation. It groups containers that make up an application into logical units for easy management and discovery. Kubernetes has quickly become the de facto standard for container management. Google should continue educating and helping enterprise clients faster adopt containers and Kubernetes in increasingly efficient ways.
By making Google Container-Optimized OS GA, Google Cloud has enabled enterprises using Google Container Engine to further increase the efficiencies that containers and microservices enable for applications. I would expect continued leadership in this category by Google to help accelerate container adoption in the enterprise as container efficiency becomes more broadly understood and implemented with the tools that Google developed as the standard.
Bonus points for Google to accelerate or automate, directly or indirectly via partner channel, the “containerizing” of existing enterprise applications that are both in AWS as well in data centers that have not yet moved to public cloud.
Machine Learning & Differentiated Offerings
From the article covering the Forbes CIO event above, Diane Greene was “asked for some examples of where Google was pushing its offerings in artificial intelligence and machine learning, Greene invoked several of the company’s acquisitions: DeepMind, acquired in January 2014, and Kaggle, acquired just last month in March 2017. The data science competitions hosted on Kaggle are hoped to give Google an edge, according to Greene, while DeepMind has advanced Google’s capabilities in using neural networks to answer questions too abstract or complex for a simple query or regression analysis.”
Machine Learning and AI have been differentiated elements for GCP since they debuted commercially based on Google’s own use of them internally. We wrote about Evernote’s plans to move to GCP based on Machine Learning last year and some of the growing pains that enterprise software companies have adopting Machine Learning with existing services. Can Google Ventures (now GV in Alphabet terms) seed startups using Google Machine Learning and AI services? Can GCP assist existing software startups with their offerings beyond providing the ML and AI service building blocks?
With the recently launched Tensor Processing Unit (TPU), Google Cloud continues to enhance their platform with offerings that were developed and supported in-house. This will be a key area to continue to differentiate in order to catch AWS as well as increase margins by using your own post-learning, inference chip sets instead of third parties which AWS uses.
One area where I hear a lot of smart people anchor their debate around the moat that AWS has constructed is that data gravity issue. If my company hosts applications and e-commerce platforms on AWS, I am probably collecting a lot of logs and other related data that I am storing in various AWS services. Instead of building my big data or analytics tools on GCP, it makes the most sense to build them on AWS since my data is already there.
Google Search is one of the largest cloud businesses in the world. It contains data on searches that should serve as the basis for services to new Google Cloud clients if they are already advertising on Google Search. I would explore making new Google Cloud services that leverage Google Search data to accelerate growth in services built on Google Cloud.
Hybrid & Multi-Cloud Support
With open source Kubernetes providing the management for containers on-premise, hybrid, or on public cloud infrastructure, Google has provided enterprises and software developers with a multi-platform approach to deploy and manage their workloads. With the broad acknowledgement of hybrid being the reality for enterprises moving to cloud over the next 10 years, it will be important for Google to be proactively providing hybrid workload options for clients.
AWS is betting heavily on VMware Cloud on AWS (our take here), while Microsoft is betting heavily on Azure Stack. Google Cloud published this piece last week describing how a multi-cloud strategy can help organizations leverage strengths of different cloud providers and spread critical workloads. Google Cloud already offers Stackdriver (acquired 2014) to monitor workloads on both AWS and GCP. How else can GCP provide the best support for clients over the next five years on a multi-infrastructure strategy?
Deepmind. Stackdriver. Apigee. Qwiklabs. Kaggle. Orbitera. All smart acquisitions that help drive GCP usage over time in addition to the value they directly provide existing users. What other tools can GCP integrate into its platform to increase usage on the underlying infrastructure that have captive users? We had some ideas here last month for both Google Cloud and Oracle that haven’t really changed.