a16z Podcast: Datacenter of the Future

Episode Summary

Executive Summary: Chris Dixon and Peter Levine discuss how mobile and cloud growth are reshaping data centers into software-defined, scale-out systems built from commodity hardware. They argue Facebook and Google now set the infrastructure blueprint, accelerating shifts toward containers, hosted open-source services, and data-center orchestration platforms like Mesosphere, while traditional VM-based and proprietary models fade.

Main Topics: Data centers as the new cloud infrastructure (Priority: 5/5): Levine explains that data centers are the physical backbone of the cloud and are being transformed to support massive mobile and future IoT traffic. Commodity hardware plus software abstraction (Priority: 5/5): The conversation centers on replacing custom enterprise hardware with inexpensive components managed by software layers, enabling scale and flexibility. Facebook and Google as infrastructure leaders (Priority: 5/5): They argue consumer internet giants, not legacy vendors, are driving the most advanced data-center design and networking innovation. Containers vs. virtual machines (Priority: 5/5): They contrast VM-heavy architectures with containers, which reduce overhead and better fit the Linux-dominant, Linux-on-Linux environment. Mesosphere and data-center orchestration (Priority: 4/5): Mesosphere is presented as a platform that aggregates compute, storage, and processes across the data center, acting like an operating system for distributed infrastructure. Hosted open source as a business model (Priority: 4/5): The speakers see hosted services for open-source technologies as the next major commercialization path, citing DigitalOcean and GitHub as examples.

Key Arguments: Mobile usage and later IoT are forcing data centers to scale dramatically, making them more software-driven and efficient. Facebook and Google have pushed back-end infrastructure forward faster than traditional enterprise vendors like Cisco or EMC because they operate at unprecedented user scale. The winning model is commodity hardware plus a software layer, not bespoke integrated hardware systems. Power, cooling, and floor space are the scarcest resources in modern data centers, so architectures must optimize for efficiency. Containers are a superior evolution to virtual machines for Linux-centric environments because they provide deployment portability without the overhead of full OS virtualization. Mesosphere represents a shift from managing individual machines to aggregating resources across an entire data center, analogous to an operating system for the cloud. Open source creates more value when offered as a service, because that captures market value while making deployment easier for users.

Data Points: Example of market share/role: "a quarter, maybe even a third" - Chris Dixon estimates the share of A16Z investments tied to data-center technology. User scale: "billions of users" - Used to describe the scale at which Facebook, Google, and other consumer internet companies operate their infrastructure. Future endpoint scale: "a billion smartphones" - Levine uses this as a visualization for the scale-out data center concept. Investment mix: "a quarter, maybe even a third" - Repeated reference to how much of the firm’s investing is around data-center technology.

Pivotal Quotes: "We like to say the mobile supply chain is eating the data center, right?" — Peter Levine: Explaining how mobile hardware advances are influencing cloud and data-center design. "The data center is a billion smartphones connected together through very sophisticated software." — Peter Levine: A metaphor for the future scale-out data center built from commodity/mobile components. "I think that the notion of having a container and being able to move things ... is really the value." — Peter Levine: Describing why containers are attractive compared with virtual machines.

Implications: The industry is moving toward cloud-native, software-defined infrastructure built on commodity parts. Winners will likely be vendors that abstract complexity, offer hosted open-source services, and help companies manage scale, security, and orchestration across private or public clouds.