A carve-out is the mirror image of an acquisition: instead of combining two companies' technology, a divested business has to be separated from its parent and stood up as something that can run entirely on its own. It's routinely the longest and most expensive workstream in the whole transaction, and it's the one that sets how long the transition services agreement has to last.
What Actually Has to Separate
Applications, infrastructure, identity systems, data, and the vendor contracts behind all of them. In almost every carve-out, the divested business has been sharing some or all of these with its parent, a shared ERP instance, a shared identity directory, a shared network, and none of that can simply be turned off on day one. It has to be replicated, migrated, or rebuilt before the new standalone company can actually stand alone.
The TSA Clock
The transition services agreement is the bridge: the parent keeps providing shared services, IT included, for a fixed window while the new owner builds its own. That window is a clock, and it's usually shorter than the separation actually takes if it isn't planned carefully. Three common patterns show up in practice, and they trade off speed, end state, and cost differently.
| Pattern | What it involves | Timeline | Best fit when |
|---|---|---|---|
| Lift and shift | Stand up a right-sized standalone environment; migrate applications and data with minimal change to how they work | Fastest of the three; the default for most PE buyers | The business needs to be running standalone quickly and the existing systems are reasonably sound |
| Greenfield build | Build the target environment new, often cloud-native, treating separation as forced modernization | Longest; needs a TSA long enough to cover the full build | The legacy estate was already the problem, and standing up a clone would just carry it forward |
| Direct migration to acquirer | Skip standing up anything standalone; migrate straight into a strategic acquirer's existing systems | Varies; effectively becomes an integration project instead of a separation | A strategic buyer, not a standalone PE hold, is absorbing the business directly |
The honest framing for a board evaluating which pattern fits: greenfield is really a transformation program with a divestiture deadline attached, not a simple separation. Choosing it should be a deliberate call about the end state the business needs, not a default.
What Actually Drives the Cost and Timeline
Two factors matter more than deal size: entanglement and shared-system depth. A business that's operated as a fairly distinct unit within its parent, with its own applications and its own contracts, separates far faster than one that's shared a single ERP instance, a single identity directory, or a single network with the rest of the organization for a decade. Identity and ERP separations are consistently the longest workstreams in a carve-out, often extending well past the initial TSA window, because unwinding who has access to what, and which financial data belongs to which entity, touches nearly every other system in the business.
A useful early exercise is mapping every shared system by how entangled it actually is, not by how important it feels. A shared email domain is high-visibility but usually low-effort to separate. A shared general ledger, by contrast, is often invisible to the deal team until the separation program starts and is consistently one of the largest sources of timeline slippage. Getting this map built in the first weeks after signing, before the TSA clock starts running in earnest, is what turns a carve-out from a series of surprises into a planned sequence.
What This Looks Like in Practice
One documented example from Resourcive's case study library: a global information services company, PE-owned, more than $6 billion in revenue, engaged Resourcive for portfolio-wide telecom and network work that included post-TSA carve-out support, standing up standalone connectivity and voice infrastructure as part of a broader technology separation. The pattern held across categories: figure out what has to be rebuilt versus what can simply be renegotiated in place, and sequence the work against the TSA clock rather than against an idealized timeline. In that engagement, the network and voice workstreams moved on a lift-and-shift basis, replicating existing service on new, standalone contracts, precisely because the underlying systems didn't need to change, only the ownership and billing behind them did.
Frequently Asked Questions
How long does an IT carve-out usually take? Full separation across the harder categories, like a shared ERP or identity system, commonly takes one to three years, well beyond how long it takes legal to close the deal itself.
What happens if the TSA expires before separation is complete? It gets extended, usually at a cost, or the business runs into real operational risk. This is why planning the separation timeline early, ideally before signing, matters more in a carve-out than in almost any other type of technology transition.
Which systems usually take the longest to separate? Identity and access management, and ERP or financial systems, consistently take longer than network, telecom, or endpoint categories, because so much else in the business depends on them.
Can carve-out separation start before the deal closes? Planning can, and probably should. Actual execution, standing up new systems and migrating data, typically can't begin in earnest until close, since it depends on decisions and access that aren't final until then.
Where to Go Next
For the combining side of a transaction rather than the separating side, see M&A Technology Integration. For what needs a decision in the earliest weeks after the deal closes, see The First 100 Days.
