Saviynt has raised $700 million in a round led by KKR at a valuation of approximately $3 billion, a raise the company describes as establishing identity security as the foundation for the AI era. For teams that run identity governance and administration, the headline number is the least interesting part. What matters is the funding thesis underneath it: that identity governance has moved from a reporting obligation to core security infrastructure, and that the buyers funding it now expect it to govern things that are not people.

That shift reframes a problem identity teams have lived with for years. IGA programmes were routinely justified by audit findings, then scoped down when the audit was satisfied. Access reviews happened annually because the policy said so. Role mining stalled because the business would not fund the workshops. Meanwhile the estate kept growing: more SaaS tenants, more contractor accounts, more service accounts, and now agents and automation operating with credentials that look human enough to pass a joiner-mover-leaver process but behave nothing like a person.

IGA is being repriced as security spend

The practical consequence of a raise at this scale is that identity governance stops being defended as a compliance cost and starts being defended as risk reduction. That matters inside organisations where the IGA budget sits with the CISO rather than internal audit. It also changes the language of the business case: not how many findings were closed, but how much standing access exists, how quickly it is removed, and whether the governance system can produce evidence that a machine action was authorised.

Consolidation puts fewer, larger decisions in front of buyers

Capital at this level usually funds two things: platform breadth and go-to-market reach. Both push the market toward fewer vendors with wider coverage across governance, privileged access and application onboarding. For buyers, the effect is a shorter shortlist and a heavier bet per decision. Choosing a platform now means committing to a roadmap for the next three years, in a category where the requirements are still being written by regulators and by the arrival of autonomous agents.

The agent governance requirement is the real test

The hardest design question in the category is where runtime control ends and governance begins. An agent that spawns a short-lived credential at three in the morning needs a policy decision at that moment, and it also needs to appear in an access review six months later next to the human who owns the workflow it serves. Vendors that treat those as separate products create the same reconciliation failure that privileged access management and IGA had for a decade, where the standing entitlement lived in one system and the privileged session in another.

Evaluators should look for a single identity record that carries both the decision and the evidence, an ownership model that assigns every agent to a human sponsor, and a review cadence that scales to identities that churn hourly rather than annually.

Financial scale changes due diligence

A vendor with a $3 billion valuation and a large private equity participant is a different counterparty than a growth-stage IGA startup. Multi-year agreements should reflect that: roadmap commitments in writing, clarity on what happens to custom connectors during a reorganisation, data portability terms, and a realistic view of how much professional services budget the deployment will consume. The licence is rarely the expensive part of an IGA programme; the integration and role modelling work is.