Hidden among the numbers revealed when technology startups go public for the first time, there’s been an interesting trend in recent years: the majority of startups about to go public are entering into exclusive, multiyear contracts with a single cloud provider — locking in cost savings, but also, the company’s business for the long haul.
When Snap, the company behind Snapchat, went public in 2017, we learned that it had entered into an agreement with Google Cloud to use a minimum of $2 billion in services over five years. Lyft’s IPO in April revealed a $200 million commitment with Amazon Web Services through 2021. Pinterest, which will IPO in the coming weeks, committed $750 million of spending for Amazon Web Services as well. Uber and Slack’s S1 filings contained similar agreements as well.
But, there’s an elephant in the room: signing these large commitments to a cloud platform comes with the expectation of using a single cloud provider, regardless of the product or developer’s need.
Why are all of these enormous startups locking themselves into a single cloud for the long run? Cost savings — and ensuring that the cost of running their platforms is predictable enough for public investors to feel confident investing their money. But, there’s an elephant in the room: signing these large commitments to a cloud platform comes with a second expectation: using a single cloud provider, regardless of the product or developer’s need.
In the search for cost savings in modern cloud, we’ve lost sight of an important and difficult to measure problem: developer productivity. Startups thrive on taking advantage of the best-in-breed tools, hiring developers that help them choose tooling that gives them an edge over incumbents and emerging competitors, but single-cloud commitments mean losing access to many of those tools.
Locking into a single cloud has other costs, too: attracting engineering talent is harder, due to a narrower set of languages and stacks, application performance can suffer and interoperability with third parties can become limited. The promise of the cloud over traditional data centers was freedom of choice, which sped up implementation of new ideas dramatically as developers experimented with tooling, API-as-a-Service, and other emerging ideas — so why artificially limit that again?
The technology landscape for developing modern applications is moving fast. A large advantage comes with a multicloud strategy: you’re ready and able to make the jump to new technologies and platforms as they become available and viable for your business, without lag.
A great example of this can be found in the current fight for dominance in machine learning tooling. Both Amazon and Google offer a range of tools for companies leveraging machine learning, but Google Cloud has a distinct edge with custom hardware designed for AI workloads which helps complete jobs faster. If you were locked into a contract with Amazon and didn’t have a service that competes in that space, your team couldn’t take advantage of that innovation.
We can look to the leaders of the industry today to see multi-cloud strategies that work. Apple, for example, has operated iCloud across both Google Cloud and AWS since at least 2016, using services from both platforms to ensure it gets access to the best tools and costs, while avoiding backing itself into a situation where it has all of its bets on a single external provider.
Netflix has developed powerful internal tooling like Spinnaker, which allows its teams to leverage multiple cloud platforms transparently. While Netflix remains the poster child of AWS, its platform-agnostic tooling encourages developers to experiment with new services and cloud technologies without introducing friction.
Netlify, a company that specializes in static-site hosting at scale, hosts across Google Cloud (GCP), AWS, and Rackspace Cloud. It moved from a single provider to leveraging three to “make sure that in the face of a major outage from an underlying provider, anywhere in our network, Netlify’s service would continue with minimal interruption”
Chief Technology Officers (CTOs) are moving toward cloud-agnostic technology stacks which can run anywhere — the first step towards being able to operate on any cloud — containerization has become the standard first step, followed by orchestration tooling such Kubernetes to add an abstraction layer above the platform you’re deployed to. At Google Cloud Next, the company even unveiled its own first-party tool for running workloads seamlessly across multiple clouds simultaneously, regardless of whether it’s Amazon, GCP and Azure. Multicloud is truly already arriving.
These innovations make it crazy to ignore multicloud. By choosing flexibility, you gain access to the best price available as “other” clouds rise in popularity, which forces the big three to compete. Looking at the market share of the public cloud today, “other” clouds make up over 30% of active usage because they offer something different from the large incumbents.
A great example: clouds like BackBlaze’s B2 data storage or DigitalOcean’s Kubernetes platform are built by more nimble companies that ship features faster and compete aggressively on cost, providing unique features along with higher levels of support than the bigger platforms.
Tools like Manifold connect with all of the services a startup might use — from Mailgun to LogDNA — allowing developers to drop in the platforms they prefer to use, without worrying about getting access or IT to negotiate a contract.
This provides a central stop for developers to jump off from, circumventing vendor lock-ins and providing instant tooling access. It also helps avoid “shadow IT” — where developers simply hide their use of unsanctioned tooling — before it occurs and preserves the development team’s ergonomics: they’re never blocked by red tape or cost controls.
The developer is given the flexibility of the best tools for the job, while the chief financial officer and CTO maintain broad visibility into what’s being used in the organization; a rare win-win scenario. Centralized management like this was only thought of when using a single cloud platform — but it’s an important part of a cross-cloud strategy, enabling properly governed infrastructure regardless of who’s actually billing you, and removing friction from the process of implementing new services.
It’s clear that there’s a strong desire among late-stage startups and executive teams to rein in their cloud costs, but focusing on cost alone loses sight of what made those businesses competitive in the first place: being early to market, and having the flexibility to jump onto new tooling that provides an edge.
Today, developers are spending up to 90% of their time maintaining software rather than innovating or building something new. There is a clear opportunity for gaining an edge through enhanced developer productivity by breaking the trend, and focusing on a multicloud strategy that provides developers with best-of-breed cloud services that enable them to use the best tools for the job.
Multicloud doesn’t need to increase complexity as a startup scales and can be just as cost-effective as a single-cloud strategy. But, most of all, your developers will thank you by sticking around and getting more done.
Image via Pixabay.