The Journey from Scale to Hyperscale

The last seven years have been wild.

What used to be a nascent cloud market has turned into the patterns and platforms on which almost every organization is building their future. Gartner has predicted that spending on public cloud services will grow 23.1% in 2021 to total $332.3 billion, up from $270 billion in 2020.??

This has had a profound impact on the world and the services we use every day.? Most of the experiences on your smartphone, on the web or at your favorite local store are partially or completely cloud-enabled.? When you travel, transact, exercise, author or learn, you’re likely directly leveraging the cloud.?

In many ways, cloud computing has moved from scale to hyperscale.

Exciting ideas in reach: the implications of a hyperscale cloud

There are several important implications of a hyperscale cloud. As a CTO at a traditional Fortune 250 company before joining Google seven years ago, leveraging AI primarily meant applying known statistical methods paired with large capitalized investments in computers. The pace was slow and discovery was often “all or nothing.”? Our engineers and scientists rightfully felt frustrated that the frontiers we strove to reach had to be scheduled according to hardware procurement timelines assembled across dozens of IT vendors--in the rare case we could invest enough to reach for those frontiers.?

That put many exciting ideas out of practical reach and disenfranchised quite a few developers, engineers, product leads and executives. But in my time leading the CTO group here at Google, I’ve seen Cloud’s hyperscale unlock completely new applications of AI for our customers in scientific discovery, space exploration, medical treatment formulation, critical system safety, commerce, transportation and public services to name a few. Underneath the AI hood, advances in chip design, networking, data platforms and collaboration platforms have made the tools of a few economically feasible for many.???

But perhaps the most personally challenging implication of this hyperscale is how it forced me and my team to change.??

When cloud was about relocating compute and storage, the blueprint for a CTO organization was a bit more straightforward: hire experts in the key underlying technology areas from a range of industries and geographies and make them available externally and internally for advice to customers and Google product and research teams. We’d learn more about enterprise ambitions, and they’d be able to move faster and avoid important pitfalls. Arguably, six years ago, this was the right approach, and we succeeded for quite a while in that mode.??

An appetite for platforms: a subtle but important shift

Then, about three years ago, a subtle but important shift became more visible. Our customers and partners started asking about how to develop and deploy their own platforms, with cloud as an important substrate. They wanted to move up the value chain and use the cloud as an enabler for better, faster or completely first-of experiences and services. They even started using cloud as a catalyst for new business structures like an all digital insurance syndicate or pharmaceutical spin-out.? I’ve often heard this called digital transformation in shorthand, but in many ways, it was more of a reinvention. Organizations like Albertsons, the Australia Post, Brit Insurance, Deutsche Bank, Etsy, Ford, Iron Mountain, Major League Baseball, Mayo Clinic, Sabre, Shopify, Twitter and the US Air Force all started redefining their industries with cloud computing at the core.

And it didn’t stop there. Increasingly, organizations like those above and our own product teams asked deeper questions like:?

• How would climate change impact our supply chain???
• How would rising interest in data sovereignty impact our ability to leverage machine learning???
• How would new classes of computation threaten the very cryptographic methods that we all depend on???
• How would new forms of computation assure modern and privacy-enabled digital healthcare???

In most cases, there weren’t clear answers. Instead of point advice based on experiences elsewhere, we needed to take a bit more of an explorer’s mindset, methodically testing assumptions with the faith that by working with enough dots, we will start to see meaningful, actionable connections emerge. Those connections would provide the risk-adjusted catalyst for new, broader and arguably more strategic customer and Google technology investments.

As a CTO organization, this meant we had to shift from providing expertise on foundational cloud primitives to more expansive engagement on how cloud could be leveraged with an eye toward more uncertain, but also more profound outcomes. It’s one thing to move applications from one place to another and something completely different when you’re deciding what to assemble and build that might change the growth trajectory of an entire global enterprise. This shift has meant a change in our people and processes and rethinking how our technology comes together from the perspective of our customers.??

Creating and evolving an organization in an unusual period of change

Over the next few months,? I’ll be sharing more about the in-depth process of creating and evolving a CTO organization over the last seven years and the specific changes we’ve made over the past few years to address this unusual period of rapid change and opportunity. My hope is that some of this will be useful to those of you embarking on ambiguous, complex projects where the cloud will play a large role.?

I’ve personally learned a lot about leadership and collaborative innovation, including how to view, analyze and evaluate opportunities; how to tie big aspirational goals to daily structure and incentives in a CTO organization; and how I needed to change my own leadership and mindset. In a way, it will be like me open sourcing Google’s approach to Horizon 2 innovation, the bridge between the truly breakthrough research and the products that form the core of our platforms. Some aspects, I’m sure, will be unique to how Google operates, but I look forward to hearing from you about what you find most interesting and hopefully some hearty debate.?

To make this truly collaborative, if you leave comments on this post, I’ll shape future posts around them.? I’m excited to share a behind the scenes look at “the what” and “the how” we do what we do with all of you.?