Summary

• Keepit says ARR has passed $100 million after growing more than thirteenfold since One Peak first invested in 2020.
• Its vendor-independent cloud separates backup infrastructure from the hyperscalers hosting many customers’ production SaaS data.
• AI Truth Cloud extends that architecture towards AI provenance and recovery, although Keepit has not disclosed how much revenue the newer strategy generates.

Danish data protection company Keepit says annual recurring revenue has passed $100 million, giving the Copenhagen business a substantial subscription base as enterprise backup expands into questions around cloud concentration, sovereignty, and the integrity of data altered by artificial-intelligence systems.

The privately held company said on 10 September that it had grown more than thirteenfold since growth investor One Peak first backed it in 2020, equivalent to a compound annual growth rate above 55%. Keepit employs around 700 people and operates independent data-centre infrastructure across several regions.

The milestone is company-reported, and Keepit has not published audited financial statements alongside the announcement or disclosed profitability. It has also not broken out revenue associated with its newer AI Truth Cloud strategy, so the $100 million figure should be read primarily as evidence of scale in the established SaaS data protection business.

That business is nevertheless becoming more strategically relevant as organisations concentrate email, documents, customer records, and workflows inside a relatively small number of cloud platforms.

Backup independence becomes part of the architecture

Keepit’s central proposition is that protected copies should not depend on the same provider that hosts the original application. Its own cloud infrastructure stores backups away from hyperscalers such as Microsoft, Google, and Amazon, reducing the chance that one access problem or infrastructure dependency affects both production and recovery.

The concept predates the current AI cycle. Ransomware, account compromise, accidental deletion, outages, and retention requirements have long provided reasons to keep immutable copies outside primary systems.

Yet SaaS concentration has increased the operational importance of that separation. An organisation can move large portions of daily work into externally hosted applications without necessarily acquiring equivalent control over how historical data is retained or restored.

European sovereignty concerns add another dimension, particularly for regulated organisations examining where information resides and which legal jurisdiction applies to infrastructure. Keepit’s status as a Danish provider running its own platform gives it a commercial argument distinct from backup products layered onto the same hyperscaler environment as the source data.

AI changes what recovery means

Techopia covered Keepit’s AI Truth Cloud when it launched in August, so the strategy itself is not new. The more useful question raised by the revenue milestone is whether an established backup business can extend its existing architecture into the problem of verifying and reversing changes made through AI workflows.

Traditional recovery has a relatively obvious trigger when somebody deletes a mailbox or ransomware encrypts information. AI agents create subtler possibilities because software can retrieve records, modify data, write outputs back into systems, and then expose those altered records to another automated process.

An incorrect change can therefore propagate without appearing immediately catastrophic. Recovering from it requires historical versions and enough provenance to determine which state was trustworthy before automation changed it.

Keepit’s AI strategy is built around using immutable historical data as a source of truth, with planned connectors and isolated environments intended to let AI systems access protected information without relying solely on live production datasets.

Whether customers will pay separately for those capabilities remains unclear. The company has not provided adoption figures for AI Truth Cloud or identified the share of new bookings attributable to AI-related use cases.

A mature market is acquiring another workload

That leaves Keepit in a different commercial position from startups trying to create an entirely new AI-security category. It already sells backup and recovery to organisations that understand why they need historical data, which gives it an installed base through which to introduce AI-specific governance and recovery functions.

The challenge is proving that these functions are sufficiently different from ordinary backup to justify new spending rather than becoming expected features of an existing subscription.

Competition is also substantial. Large data protection companies, specialist SaaS backup providers, cloud vendors, and application developers can all improve their recovery features, while customers may prefer fewer infrastructure suppliers if native capabilities prove sufficient.

Independent infrastructure therefore has to earn its additional operational cost. Separation becomes valuable when it materially improves resilience, legal control, recovery confidence, or the ability to reconstruct data after a system has altered it.

The $100 million ARR figure suggests enough organisations already see value in that underlying proposition to support a sizeable business. It does not yet prove that AI creates a second growth curve, although AI does give backup vendors another reason to discuss data integrity as an operational requirement rather than an insurance policy used only after a disaster.

Keepit’s next evidence will come from customer behaviour: how many organisations use immutable histories to govern AI workloads, how often they need rollback or provenance functions, and whether those capabilities expand contracts. Until then, the milestone says more about the scale of independent SaaS protection than the maturity of a new AI market, while giving the company an unusually large base from which to test whether the two eventually converge.