SaaS leaders are moving past the “all-in on public cloud” era, and in its place hybrid SaaS architectures are taking prime position.
Following the cloud boom, cloud primacy became deeply rooted in the business consciousness. And for cloud-native industries like SaaS, born into the cloud bubble, this went unquestioned for years.
But, in time, even cloud-natives have realized that relying entirely on hyperscale cloud comes with downsides such as unpredictable spend, resource contention, redundancy shortfalls and limited hardware control.
And so, what started as a trickle of early adoptees selectively repatriating workloads out of hyperscale cloud has become an industry-wide phenomenon.
Buoyed by major wins (well documented by early trailblazers including 37Signals and Dropbox), SaaS teams are choosing hybrid infrastructures in a bid to right-size costs, boost performance and regain a greater level of control.
In other words, the ‘cloud-only’ mentality is fast being replaced by ‘cloud-also’.
Hybrid SaaS infrastructure is a SaaS hosting strategy that involves dividing workloads across multiple compute types and providers. A hybrid SaaS architecture can include any combination of public or private hyperscale cloud solutions and bare metal environments (including bare metal cloud hosting, colocation or on-premises deployments).
As hybrid infrastructure strategies become mainstream, SaaS organizations are realizing that going hybrid is more than just a strategy for avoiding cloud vendor lock-in (though this remains a significant consideration). It’s a strategy that brings cost, performance and redundancy wins – not to mention leverage to pivot with evolving data sovereignty laws.
Offering instant elasticity, short contracts and free credits at sign up, hyperscale cloud can be incredibly convenient in the first instance. But it comes with a complexity tax down the line.
A recent Flexera report shows that 84% of organizations say that managing cloud spend is their top cloud challenge. For SaaS teams in this situation, selectively repatriating workloads away from hyperscale cloud and onto more cost-efficient bare metal is helping reset the unit economics.
SaaS web software company, 37signals is the most well cited example. In 2022 the company announced that it would be migrating its project management platform, Basecamp, and email service, Hey, off AWS and GCP and into their own data centers. Since then, CTO and co-owner David Heinemeier Hansson has reported the company’s experience and accumulated savings via his blog and on social media.
In a recent post outlining the company’s cloud spend between 2017 and 2025, Hansson shares: “We squandered well over ten million dollars between then and when we pulled the big plug in 2023.”
For 37signals, the final piece of the repatriation puzzle involved repatriating a few remaining workloads that had still been running on AWS S3, due to a four-year contract lock-in. The move was completed in October 2025 rendering the company 100% cloud-free and anticipating accumulated savings in the region of $10 million.
For Hansson, working towards total repatriation has become something of an anti-cloud manifesto. But not every organization will want (or be able) to execute a repatriation of this scale – and that’s okay.
What we can take from 37signals is solid evidence that even partial repatriation can result in a significant reduction in infrastructure costs. In other words, hybrid SaaS solutions that combine cloud and bare metal effectively lower the steady state cost of predictable workloads whilst preserving burst capacity in cloud.
Relying completely on multi-tenant cloud environments poses several challenges: performance degradation from resource contention, limited hardware-level control and minimal customization over CPU, RAM, storage and network paths.
You’re left with an environment that is highly elastic but vulnerable to performance trade-offs and is near impossible to optimize.
For SaaS teams, these challenges can quickly translate into more acute problems. High levels of tail-latency for databases, caches and large analytics queries are often caused by network congestion, server overload, system bottlenecks and ‘cold starts’ in serverless environments. And when these problems are felt consistently, it’s not just a nuisance but damaging for business. Timeouts, slow page load speeds, and support ticket spikes soon translate into customer dissatisfaction and churn that damages the bottom-line.
The good news is that these challenges can be mitigated by migrating performance-sensitive workloads to single-tenanted bare metal environments. In single-tenanted environments you don’t have to compete with other businesses over finite hardware resources (a common cause of system slowdowns). And as the server is yours alone, there’s greater opportunity for performance optimization through customization. Ultimately that means faster flows, reduced inefficiencies and higher retention rates, keeping your customers happy - and most importantly - loyal.
Cloud file storage company Dropbox was one of the first SaaS companies to adopt a hybrid infrastructure strategy. And while motivated by multiple factors - from profit margins to the impending launch of the company’s Magic Pocket solution - the driving impetus was a need for performance optimization and tailored solutions.
It’s why in 2015 (at the height of the cloud boom) the company chose to move a large chunk of its data off AWS S3 and onto custom bare metal infrastructure. In the aftermath of migration, former Head of Production at Dropbox, Dan Williams, told TechCrunch: “For us it was about quality and control management.”
Dropbox’s migration speaks heavily to the benefits of hybrid architectures and avoiding an ‘all or nothing’ approach to infrastructure. Former Lead Developer at Dropbox, Preslav Le, told Data Center Dynamics back in 2021: “We built the ability to move data back and forth between the two locations. Over the years, we decided that it’s worth retaining that capability.”
Cost and performance aren't the only benefits of a hybrid SaaS model. Redundancy is another important, but all too often overlooked, consideration.
The global AWS outage on October 20, 2025, served as a timely reminder that over-reliance on the big three raises systemic risk.
“This outage once again highlights the dependency we have on relatively fragile infrastructures,” commented Jake Moore, Global Cybersecurity Advisor for ESET. "Incidents like this make clear the need for a more open, competitive and interoperable cloud market, one where no single provider can bring so much of our digital world to a standstill,” said Nicky Stewart, Senior Advisor at the Open Cloud Coalition.
“This outage once again highlights the dependency we have on relatively fragile infrastructures.”
For SaaS platforms, the outage was felt closely. Canva reported ‘significantly increased error rates,” due to “a major issue with our underlying cloud provider”, for example. And countless more, including the likes of Monday.com, Vimeo, Xero and Mailchimp, were reported to have experienced similar issues.
The real problem here is that these interruptions have the potential to cause severe reputational damage and poses the risk of customers looking to go elsewhere as a result.
The outage affected all industries that rely on AWS, but it also brought to light some systemic issues in relation to how SaaS companies have been operating.
First, that there has been a chronic over-reliance on a single point of failure - with many SaaS platforms defaulting to AWS’ services in the US-East-1 region specifically for its maturity and low latency. And second, that without adequate transparency into a provider’s infrastructure (or control over failover pathways), being ‘always on’ is not guaranteed. One outage can, and will, stall an entire workflow if redundancy has not been prioritized.
It also reiterates the need to plan for things going wrong. EVP Product and Engineering at servers.com, Ilya Sokolov, explains:
“Outages like the recent AWS incident are a reminder that no provider is immune to failure - and every business should have a backup plan. That doesn’t mean maintaining a one-to-one copy of your infrastructure elsewhere. What matters is having a clear, tested path to switch workloads when it makes sense for your business - whether that’s hours or minutes.
“The best-prepared teams treat this as an active discipline: they regularly perform ‘switch’ or failover exercises or continuously run a portion of their workloads in an alternate environment.
“That’s how you build real resilience - not by hoping your provider never fails, but by designing for when it does.”
“Outages like the recent AWS incident are a reminder that no provider is immune to failure - and every business should have a backup plan.”
This is where adopting a hybrid SaaS infrastructure strategy can make a real difference. Distributing workloads across diverse environments means that operations can be maintained even if one part of the infrastructure fails. Similarly, by combining bare metal hardware for steady state workloads with public cloud services for periods of peak demand, alongside a tried and tested business continuity plan, SaaS platforms can ensure that applications can and do remain always available.
Cloud-only infrastructure can leave SaaS teams boxed in by jurisdictional rules they can’t fully control. Even when opting for a ‘regional’ cloud, tooling and services may replicate globally, triggering cross-border transfer obligations and audit headaches.
Multi-tenant environments amplify the risk of personal data being moved into different jurisdictions and this can result in significant legal hoops to jump through. Even with the necessary provisions in place, there’s still no way to eliminate hidden data movement, human error, interpretation risk or vendor sub-processor drift (when a cloud service shifts where your data is processed when adding to or changing sub-processors).
A hybrid setup, mixing public cloud with bare metal servers strategically placed in your required geographies, is a way to keep control over sensitive data while still benefitting from public cloud elasticity when its needed. This could mean placing customer records, backups, and audit logs on local machines while running burst tasks like web traffic handling or reporting in a hyperscale cloud environment. Albeit at the cost of a bit more operational overhead, this greatly reduces legal and compliance risk while maintaining fast delivery.
Data protection is a challenge that businesses across industries are grappling with right now and one that is set to become only more imperative as laws tighten; failing to keep up will come with legal and financial ramifications.
For example, in the UK, data sovereignty has become a strategic priority, and updated GDPR guidelines are making overseas hosting more of a challenge for UK-based organizations. As a result, businesses are being forced to consider the potential legal and operational consequences that could come with hosting data either outside of the UK or with a non-UK provider. To protect data sovereignty and remain compliant, organizations must be prepared to conduct comprehensive auditing, which may require moving workloads to UK certified data centers.
“Organizations need to ensure their data is protected independently, immutably, and in alignment with evolving sovereignty requirements,” comments Niels van Ingen, Senior Vice President of Business Development and Strategy at Keepit. “In today’s environment, control over your data location and architecture isn’t just an IT preference - it’s a business imperative.”
Cloud repatriation trends are still going strong. According to G2, 83% of enterprise CIOs planned to repatriate at least some workloads in 2025. Coupled with very public repatriation narratives from SaaS industry leaders, it’s clear that cloud-only mentalities are a thing of the past.
And while some SaaS organizations will follow in the footsteps of 37signals and repatriate their entire estates, for the majority it’s about keeping options open. Monopolizing both on what the hyperscalers have to offer and combining it with the performance and control of bare metal is key to create a ‘right-sized’ hybrid SaaS architecture.
Frances is proficient in taking complex information and turning it into engaging, digestible content that readers can enjoy. Whether it's a detailed report or a point-of-view piece, she loves using language to inform, entertain and provide value to readers.
