HarbourVest managing director Scott Voss says investors must rethink valuation frameworks
As AI continues to disrupt and reshape how secondaries price risk, durability and growth across software portfolios, investors are being forced to rethink long-standing assumptions about valuation, underwriting and portfolio construction.
In an interview with InvestmentNews, Scott Voss, managing director at $140 billion AUM Boston based investment firm HarbourVest, described a secondary market that is already showing signs of strain — particularly where broadly diversified software exposure is concerned.
“Given recency, this is purely anecdotal,” Voss said. “It seems that deal volume in software secondaries is currently in flux, with many GPs considering opportunities but facing considerable caution from the buyside. Most single asset secondaries under evaluation are outside the software sector, and the few multi-asset CVs with software exposure are largely in a holding pattern. Overall, it seems that while select, more discrete software opportunities might get done, the environment is not favorable for broadly diversified software portfolios at the moment.”
That caution reflects what Voss sees as a more fundamental divergence emerging across software assets — one driven not by macro cycles but by structural disruption linked to AI.
“The divergence is structural, not cyclical,” he said. “AI is breaking two core SaaS assumptions: seat-based pricing no longer scales when AI allows one user (or agent) to replace many. Fixed-cost leverage is eroding as AI introduces variable inference costs.”
He added that the impact is already visible in valuation trends.
“Software whose core value is ‘thinking’ — analysis, drafting, summarization — is being commoditized by foundation models and is seeing sharp multiple compression,” Voss said. “By contrast, software embedded in transactions, regulated workflows, or systems of record often benefits from AI, which increases volume, compliance complexity or switching costs rather than replacing the platform itself.”
Public markets clues
Public markets have provided an early signal of how this dynamic could unfold in private secondaries.
“This story is playing out in the public markets. The drawdown did not discriminate. The rebound has,” he said. “Further, this seems to hold true today, but no guarantees in the future. How the markets react to the next LLM release will be telling.”
For secondary buyers, the implications extend well beyond sector allocation and into the core mechanics of valuation. Metrics that once served as shorthand for quality are becoming less reliable indicators of long-term resilience.
“I would expand this further. ARR, GRR, TAM, NRR and Rule of 40 are no longer good enough,” Voss said. “Durability of ARR matters more than growth rates. GRR might matter more because it tells you whether the core business is actually holding.”
Investors are also introducing new layers of discounting to reflect evolving business model risks.
“Investors now discount for revenue replacement risk — not just churn — margin volatility and erosion from usage-based or other new pricing models, and shorter product cycles,” he said. “As a result, we have seen and will continue to see even ‘quality’ software companies with multiple compression.”
In the most extreme cases, this could reshape the very rationale for secondary acquisitions.
“The most vulnerable assets may get bought for steep discounts where the acquisition is purely about buying customers versus buying the business,” Voss said.
Durability in focus
Central to navigating this environment is what Voss repeatedly describes as durability — a concept that is increasingly defined by structural necessity rather than product sophistication.
“Durability now comes from structural necessity, not feature depth or scale,” he said. “Right now, resilient platforms typically sit in the path of money, compliance or regulated execution; are the system of record, not just a decision aid; generate proprietary data as a byproduct of workflow; face high switching or recredentialing costs; and benefit from network effects across their industry vertical.”
“In all of these cases, AI expands the value of the platform rather than displacing it,” Voss added.
This durability lens is also shaping how investors distinguish between mission-critical software and tools that may be more vulnerable to substitution.
“First, this question has been on the table for all investors since ChatGPT in November of 2022. The future may have come faster than some expected,” Voss said. “The key diligence question is pretty straightforward: is the software executing a required outcome — or merely producing an answer?”
He draws a clear distinction.
“Mission-critical platforms move money, enforce regulation or record truth. Replacement risk is operationally prohibitive,” he said. “Vulnerable tools are those where AI can deliver the output without owning the system — research, BI dashboards, drafting tools. I believe this distinction consistently explains pricing dispersion across software secondaries today.”
Structural defensibility
For financial advisors and allocators assessing secondary managers, Voss argues that scrutiny is shifting away from thematic narratives toward evidence of structural defensibility.
“It is the same set of questions we are talking about right now,” he said. “How has the portfolio been re-underwritten asset by asset for AI substitution risk? Which holdings benefit from AI adoption versus face pricing or relevance pressure? How are valuation marks and continuation vehicle pricing reflecting AI risk? Does the GP have operational AI capability, or only thematic exposure?”
“LP scrutiny is shifting from ‘AI narrative’ to evidence of structural defensibility,” he added. “In some cases, LPs are even using AI to help identify risks and assist with underwriting. It is another data point in the inform.”
Looking ahead, Voss expects further disruption — but also a clearer separation between winners and losers.
“Near term: more dislocation, forced selling and wide pricing dispersion as portfolios are repriced asset by asset,” he said. “Medium term: a much more selective market, where capital concentrates in durable platforms and AI-native winners.”
He also sees scope for active ownership to play a larger role in reshaping business models.
“There is an opportunity for value-added capital to help some companies transform their business models to an AI-durable one,” Voss said. “Unless the secondary market builds this capability, it is the value-added part of private equity that will need to step in.”
Ultimately, he believes success in this new environment will depend less on broad sector exposure and more on the quality of underwriting discipline.
“For disciplined secondary buyers, this environment favors deep underwriting over broad exposure, with AI acting as the catalyst rather than the cycle itself,” Voss concluded.
