The Past, Present, and Future of Digital Infrastructure with Martin Casado

Key Learnings from:?The Past, Present, and Future of Digital Infrastructure.

Martin is a general partner at Andreessen Horowitz, where he focuses on digital infrastructure. Before joining a16z, Martin pioneered software defined networking and co-founded Nicira, which was bought by VMware for $1.3 billion in 2012. This episode provides an interesting narrative on how we got to where we are with cloud infrastructure and where we might be headed next.

[Learning 1] The Cloud Oligopoly:

Martin believes that the big 3 cloud providers are going to continue to grow nicely and that their margins are going to remain intact. He believes there are still large enterprises that need to adopt the cloud, and there are large workloads from current customers that still need to move from on-premises, and lastly thinks a lot of large customers have sweet heart pricing that won’t get much lower if they add additional workloads (keeping the margins intact as customers grow).

[Learning 2] Primitives Of The Digital World:

Compute, storage, databases and networking are the traditional infrastructure layers. These have all been collapsed into a software layer onto of merchant hardware in the form of public cloud providers. We are now also getting services built onto of this software layer that are higher level abstractions (machine learning workflows, purpose built databases, time series analytics, etc). This is speeding up access to these abstraction layers, such as machine learning that prior was only available to very select organizations (machine learning-as-a-service).

[Learning 3] API First Companies:

When you take a look at any application and you can almost assume that any sub component of that application could become a standalone API company because the size of the software market has become so large. At one point in the history of software, every company had to build each of their own components of their application. The analogous story here would be the automobile as cars gained adoption and market size increased. At a large enough market it make sense to have companies solely focused on building tires rather than the car manufacturers making their own tires. In the software world the "tire companies" are Stripe for payments, Twilio for SMS messaging, AWS instead of racking servers --- as a software developer, you don't need to build these features in-house anymore. This wave of API businesses is speeding up the development cycle for app builders, who now have the public cloud and API companies that give them the building blocks to work with so they can spend more time focusing on the core business and not undifferentiated components. Martin thinks we are pretty early in the API businesses process (especially true if we get into more complex applications that have more sub components).

[Learning 4]: The Future of Open-Source:

Believes that as we move more and more to SaaS, that open-source matters a lot less. For a developer writing an application, you can OSS and operate the software yourself or you can use an API that someone else operates (don’t care if the code is OSS or not). The describes this as open-standards and believes this this is what will stick around (like open-source, this makes it so you are not being locked into 1 vendor, but this is different than open-source itself). The point being if you can give the market a low cost way to leverage your product and flexibility to build around it, you offer some of the benefits developers find in leveraging open-source software.

[Learning 5]: Building a Relationship with Developers:

Developers have become a buying class and they are getting more and more budget. Traditional sales/marketing does not work well for selling to developers. Developers have strong opinions and like to do their own research and testing before talking to sales teams. The best developer focused companies have found out how to build great communities around their developers. You want to build good ecosystems and have a way to get insight into these ecosystems and provide actual value to your open-source users so that you can eventually get to a more traditional sales motion once the user has done their own education (eventually you might need to help with business value assessments, helping with product positioning, finding additional value streams to attach onto, engage your professional services teams, but until then it's great to have customers educate themselves).

[Learning 6]: Backend vs Frontend Developers:

Backend developers are the system folks, they are about building big distributed systems, and less about business logic of the application itself. It’s like building services that the frontend developers leverage. Frontend, are building the UI and logic that is facing the customers. For the most part, there are more frontend developers than backend, and we are seeing more and more that frontend developers are controlling more of the stack and building more of the application (think using database as a service so they don’t need backend as much if more of their system is built by a cloud provider).?

[Learning 7]: Shift in How Software Companies Compete:

It went from who can scale infrastructure properly, build good software, acquire customers, but now with public cloud you remove the infrastructure scaling as a differentiated playing field and even building software has become less differentiated with API first companies removing components of the application to compete on. Now the shift is more focused on how companies leverage data and make decisions with data (how do you match participates, how do you provide insights, etc).

[Learning 8]: Modern AI and ML:

Modern AI and ML is allowing us to take unstructured data and digitize it to then add it to the typical logic of a program. We have not been able to do this for natural language historically (never been able to do this with voice or speech), but there are now companies that are focused on just this type of stuff and many doors are going to open for new companies that were not possible prior. You will be able to build on-top of these companies and their AI / ML (potentially a way to disrupt incumbent software players).