AI is changing how private equity values software companies

High software valuations are hiding a more difficult investment market

Software companies are still attracting high acquisition prices in 2026, even as industry growth slows and private equity (PE) investment returns weaken. The apparent contradiction reflects which companies are reaching the market. PE sponsors are focusing their exit efforts on their strongest assets, while weaker businesses remain in their portfolios awaiting better conditions. As a result, completed transactions reflect the valuations of a particularly attractive segment of the software market.

Broader financial indicators show the pressure facing the industry. Annual software revenue growth has slowed from around 20% to approximately 10%, while net revenue retention (NRR) has declined by roughly 9 percentage points since 2021. NRR measures how much recurring revenue a company retains and expands from existing customers, accounting for expansion, contraction, and customer losses. Together, these changes show that software companies are struggling to sustain the growth that previously supported premium valuations.

Investment returns are beginning to reflect those conditions. Data from SPI by StepStone, published in April 2026, shows that technology buyouts completed after the Covid-19 pandemic are generating returns below pre-Covid averages. Investments from the 2020–2022 vintages face particular challenges because many were acquired at high valuations before the slowdown became apparent. Those entry prices leave less room for weaker growth during ownership.

The StepStone analysis covers fully realized investments and partially realized deals that have returned at least 25% of their proceeds, across all transaction sizes. Many investments still have value left to realize, so their eventual returns could improve. Even so, slower revenue expansion makes it harder for sponsors to achieve the returns generated during the preceding decade, especially when acquisition prices assumed stronger growth.

These pressures are changing dealmaking. Transaction activity has weakened, holding periods are extending, and aging technology portfolios are becoming a more pressing concern for general partners (GPs), the firms that manage PE funds. Strong companies can still command high prices, but sponsors holding less attractive assets face difficult decisions about when and how to exit.

This divide complicates acquisition underwriting, the process of estimating an investment’s future performance and acceptable purchase price. Historical growth rates, margins, and customer expansion patterns offer less certainty about what a software business can earn during its next ownership period.

Artificial intelligence adds another layer of uncertainty. AI can strengthen a software company’s products and earning power, but it can also raise operating costs, change customer purchasing behavior, and expose established products to new competitors. Investors therefore need to understand how AI changes the business behind the financial metrics they have traditionally relied on.

AI is changing the economics behind software valuations

For more than a decade, software-as-a-service (SaaS) investing benefited from a relatively consistent business model. Customers paid recurring subscriptions, providers expanded those relationships over time, and the cost of serving additional users was generally low. These characteristics made subscription growth, retention, and margins useful indicators of future cash generation.

Revenue expansion followed familiar patterns. Customers purchased additional user seats as their workforces grew, upgraded to premium functionality, or bought complementary products from the same provider. Each purchase increased recurring revenue, often with little additional delivery expense. Successful SaaS businesses could therefore grow revenue faster than their operating costs.

Established providers also benefited from their position inside customer organizations. Their applications supported important workflows, employees became familiar with them, and customer data accumulated within their systems. Replacing those applications could disrupt operations and require significant effort, which increased switching costs and supported long-term commercial relationships.

AI is changing the assumptions behind these advantages. An AI-native software company, built around artificial intelligence from the beginning, can solve an established customer problem using a different technical and commercial model. Rather than reproducing every function of an existing application, it may automate the work employees previously performed through that application.

This creates two risks for established SaaS providers. AI competitors may offer a more effective way to achieve the customer’s objective, while automation may reduce the number of conventional software seats customers require. Both developments weaken the historical relationship between workforce growth, software usage, and subscription expansion.

AI also creates opportunities to charge for different forms of value. A product that performs work autonomously may generate revenue based on consumption or completed outcomes rather than the number of employees using it. These models can connect pricing more closely to the work delivered, although revenue may become harder to forecast when customer activity fluctuates.

Costs change alongside pricing. Conventional SaaS products often have low additional delivery expenses after development, allowing margins to improve as usage grows. AI products can require substantial computing resources whenever customers use them, including payments to third-party AI model providers. Greater adoption can therefore increase operating expenses alongside revenue.

A successful AI product may attract customers and grow quickly while producing lower margins than an established subscription offering. Its financial performance depends on what customers pay relative to the computing resources and other services needed to deliver the product.

These changes are particularly challenging for incumbents whose development teams were built around steady improvements to existing applications. Adding an AI feature may be relatively straightforward compared with redesigning the broader workflow customers use the software to complete. Yet a competitor that automates more of that workflow could change what customers are willing to purchase.

Incremental improvements can still be valuable when they address important customer needs and generate attractive returns. The investment decision depends on whether those improvements can preserve the company’s competitive position as alternatives develop. In some businesses, targeted enhancements may be sufficient; in others, remaining competitive may require a more fundamental product redesign.

AI investment is advancing faster than financial measurement

These product changes have encouraged GPs and portfolio management teams to commit substantial resources to AI development. Companies are introducing features, building products, and deploying AI internally to improve efficiency. The activity shows that management is responding to technological change, but reveals little about the resulting financial performance.

Consider a software company that launches several AI assistants, reports strong initial adoption, and expands its development team. Those developments may indicate customer interest and management commitment. Investors still need to establish whether customers are paying more, whether retention is improving, and how much the functionality costs to operate.

Internal AI projects raise similar questions. Deploying AI across an organization can reduce operating expenses or improve employee productivity. Management must connect those deployments to measurable operational changes before treating the expected benefits as improvements in business performance.

The gap between activity and results can create valuation disagreements. Sellers may emphasize innovation, customer interest, and future market opportunities, while buyers need to estimate the revenue and profitability those investments can reasonably generate. The disagreement becomes especially important when acquisition prices already assume continued growth.

AI’s contribution to financial performance is therefore becoming a central underwriting question. Investors need to determine whether new products increase customer spending, protect existing revenue, or improve operating efficiency. They also need to understand the resources required to sustain those benefits as adoption increases.

Why traditional SaaS metrics need additional detail

Traditional SaaS metrics remain essential for evaluating software businesses. Annual recurring revenue (ARR) measures contracted recurring subscription income, NRR captures customer retention and expansion, and gross margin shows how much revenue remains after the direct costs of delivering products or services. Together, they help investors assess growth, customer relationships, and operating efficiency.

The limitation is that aggregate figures can combine products with very different financial characteristics. A company may have a healthy subscription business alongside AI offerings that generate variable revenue and require expensive computing resources. Investors need more detail to understand how those activities affect overall performance.

Three changes make that additional measurement particularly important.

Revenue is becoming less dependent on software seats

Under a conventional subscription model, revenue often increases when customers add employees who need access to an application. The provider sells additional licenses, and ARR captures the resulting increase in contracted recurring revenue. This creates a relatively direct connection between customer headcount and software spending.

AI-enabled automation can weaken that connection. When software performs tasks that previously required employees, a customer may receive greater value while needing fewer user licenses. The provider must then find a commercial model that reflects the work its technology performs.

Usage-based pricing offers one approach by charging customers according to their consumption of AI functionality. Outcome-based pricing goes further by connecting payment to completed work or achieved results. Both approaches can capture value beyond conventional seat expansion, but they introduce different forecasting requirements.

Consumption can fluctuate between periods, and completed outcomes may vary independently of employee headcount. ARR remains useful for assessing contracted subscription revenue, while additional measures are needed to understand these newer revenue streams.

Delivery expenses can increase with consumption

Pricing changes matter because AI products can have different cost structures from conventional SaaS applications. Marginal cost is the additional expense incurred when a company serves one more customer or processes one more transaction. Established SaaS applications typically benefit from low marginal costs after development.

AI applications behave differently because each query may consume computing capacity or generate a charge from an external model provider. More customer activity can therefore produce additional expenses even after the product has been built.

This affects gross margin, which measures revenue remaining after direct delivery costs. An AI feature bundled into an existing subscription may increase expenses without generating corresponding revenue. A separately priced product may cover those costs, but profitability depends on consumption, pricing, and the resources required for each interaction.

Investors consequently need to understand how margins change as adoption expands. An AI product that appears profitable at limited usage may perform differently when customers depend on it throughout their daily operations. The financial assessment must connect customer activity to both revenue and delivery expense.

Automation changes the competitive value of software functionality

Changes in revenue and costs become more important when AI also affects what customers need from established applications. Incumbents retain advantages through customer relationships, workflow integration, accumulated data, and employee familiarity. Those advantages can make replacement difficult and support continued demand.

Yet AI-native competitors may challenge functionality that customers previously relied on established software to provide. The level of exposure depends on the work involved and the durability of the incumbent’s advantages. A useful distinction is whether an AI product assists employees or executes larger portions of their work.

A copilot assists an employee with individual tasks. It might generate content, analyze information, or prepare a recommendation while the employee remains responsible for completing the broader process. This type of product can increase the value of an existing application and support premium subscription pricing.

An agentic system uses AI agents to perform a sequence of tasks and execute larger portions of a workflow with greater autonomy. Depending on the application, it may complete an entire process that previously required several human interactions with software. That broader responsibility can support pricing based on completed work while potentially reducing demand for conventional applications.

The two models raise different financial questions. Copilots may increase spending within existing subscriptions, whereas agentic products can generate new revenue streams and change how customers purchase software. Both may require substantial computing resources, so adoption and willingness to pay must be assessed alongside delivery costs.

Neither approach automatically produces superior returns. A profitable assistant can be more valuable than an agentic product with rapid adoption and expensive execution. Investors need to evaluate what each product earns, what it costs, and how its growth affects the established business.

These differences can disappear inside aggregate software metrics. Strong subscription retention might coexist with expensive AI development, while growth in a new AI product could partly replace revenue from existing offerings. Separating those activities gives investors a clearer view of which products are strengthening the business.

Measuring AI value requires separating revenue, costs, and scalability

The limitations of aggregate reporting lead to three financial questions:

1. Is AI generating incremental revenue?
2. How is AI changing the company’s cost structure?
3. Can AI products grow without consuming a disproportionate share of the additional revenue?

Answering these questions requires management to identify AI revenue and expenses separately from established software operations. Investors can then examine individual products, assess their contribution to the wider business, and compare performance as adoption increases.

This distinction also improves capital allocation. Management can identify initiatives generating new customer spending, products that mainly support existing relationships, and activities consuming resources without sufficient commercial returns. The information helps determine which investments deserve further development.

Three AI revenue categories require separate measurement

AI revenue should be divided into three categories because the products serve different customer needs and can produce different financial outcomes.

Traditional AI and machine learning includes capabilities that many software companies have offered for years. Predictive models use historical data to estimate future outcomes, while risk-scoring systems assess the likelihood or severity of particular events. These functions may already be embedded in established products and reflected in their existing revenue streams.

Their commercial position can differ substantially from that of newly launched generative AI offerings. Established machine learning capabilities may contribute meaningful customer value and support retention, particularly when they rely on proprietary customer data. Their growth, margins, and competitive exposure should therefore be assessed independently.

Separating these capabilities also clarifies the performance of newer investments. A company reporting growth in total AI-related revenue may be benefiting from products developed years earlier. Identifying revenue from established functionality allows investors to evaluate recent development initiatives on their own merits.

The distinction also matters for competition. An established predictive model supported by proprietary customer data may face different risks from a recently introduced assistant that relies heavily on widely available AI models. Each requires an assessment of its customer value and defensibility.

AI add-ons cover functionality introduced within existing software products, including copilots and assistant features. These offerings generally help customers complete tasks inside established applications. Their commercial purpose may be to support premium pricing, encourage broader product adoption, or strengthen customer relationships.

For example, a software provider might introduce an assistant that helps employees analyze information or prepare work that previously required more manual effort. Customers may find the feature valuable, but additional revenue depends on whether that usefulness leads to new spending. Adoption and monetization must therefore be measured separately.

An assistant included in an existing subscription may attract substantial usage while generating little direct revenue. It could still improve retention or strengthen the provider’s competitive position, which may justify the investment. Management needs to identify those benefits and compare them with the cost of delivering the functionality.

Pricing becomes particularly important when the provider charges a fixed premium while computing costs vary with usage. Customers who use the assistant heavily may generate substantially different margins from those who use it occasionally. Investors should examine paid adoption, incremental revenue, customer consumption, and associated delivery expenses together.

Agentic products represent the third category. These systems automate workflows or execute tasks with limited human intervention, allowing providers to address a larger portion of the customer’s objective. A product may coordinate several steps to deliver a completed outcome.

That broader scope creates opportunities for new commercial models. Providers may charge according to the volume of work completed or the outcomes delivered, creating revenue that can grow independently of employee headcount. The model also introduces operational complexity because successful execution may require multiple AI actions and substantial computing resources.

Financial performance depends on what customers pay, how reliably the system performs, and what resources each workflow consumes. Investors must also consider whether agentic products generate additional demand or replace spending on established applications.

For an incumbent, this creates a revenue cannibalization question. Customers may transfer spending from conventional subscriptions to the new agentic offering, making the AI product’s growth appear stronger than the company’s overall expansion. Alternatively, automation may attract new spending that the provider could not previously capture.

Both outcomes can be commercially reasonable, but they imply different growth and valuation assumptions. Management needs to measure the effect on total revenue and profitability, including changes to the established product portfolio.

With these categories separated, investors can compare revenue growth, margins, adoption, and competitive exposure within each group. A rapidly growing agentic product may justify significant investment, while a profitable AI add-on or established machine learning capability may offer a more predictable return. The decision should reflect each product’s financial performance and durability.

AI-specific costs must be connected to AI revenue

Separating revenue is the starting point, but costs determine how much value the products create. AI expenses may be spread across infrastructure budgets, third-party service contracts, and employee compensation. Without a consistent allocation method, management can underestimate product costs or assign them to the established software business.

The assessment should capture four categories.

These costs should be assigned to the products consuming the resources. Management also needs to distinguish development investment from the recurring expense of delivering functionality to customers. The distinction matters because those activities affect financial performance differently.

Development spending covers the resources required to create and improve AI functionality. It includes AI-related research and development (R&D), along with the fully loaded costs of employees performing that work. Fully loaded employee costs include compensation and associated employment expenses, giving management a more complete measure of development investment.

The calculation should cover employees dedicated entirely to AI initiatives and those dividing their time between AI and other responsibilities. When an employee works partly on AI development, an appropriate share of their cost should be allocated to the relevant initiative. Otherwise, companies may underestimate investment when development happens across existing teams.

Accurate allocation matters because engineering capacity has alternative uses. Resources assigned to AI initiatives cannot simultaneously support every improvement planned for established products. Management needs to understand those trade-offs before expanding, redirecting, or discontinuing development work.

Inference costs measure the expense of running trained AI models to produce results. Inference occurs when a model processes an input and generates an output, such as an answer to a query or an action within an automated workflow. Each interaction can consume computing resources and potentially generate charges from an external model provider.

Those expenses can become significant when usage is high or unpredictable. As customers adopt AI functionality more widely, the system performs more work and may require greater computing capacity. These costs continue after initial product development is complete.

Hosting and infrastructure expenses also need to be attributed to the AI products using those resources. Without this allocation, delivery costs can remain mixed with those of established software products. That makes the profitability of both activities harder to interpret.

Third-party model expenses require similar attention. A company that relies on externally supplied models may face charges that vary with customer consumption. Understanding those costs helps management evaluate pricing, product design, and the financial effects of increased usage.

Together, these allocations establish how much product revenue remains after direct delivery expenses. They also reveal how much additional development investment is required to maintain and improve the offering.

The distinction becomes especially useful when comparing products at different stages. A newly launched AI product may require substantial upfront development spending while customer revenue remains limited. A mature offering may need less additional development investment but continue to generate significant inference expenses.

These situations call for different operating decisions. Management needs to determine which costs arise from creating the product and which will continue as customers use it. Investors can then assess current profitability alongside the investment required for future growth.

Scalability depends on revenue growth relative to consumption

Once revenue and expenses are visible, investors can assess scalability: whether a product can grow while maintaining or improving its financial performance. Customer adoption provides an early indication of demand, but the relationship between usage and cost determines how profitable that demand becomes.

Consider an AI assistant sold as a premium feature. If customers pay a fixed monthly fee, revenue may remain stable even when usage increases significantly. Inference expenses, however, can continue rising with consumption, causing gross margin to deteriorate as the assistant becomes more popular.

An agentic product priced according to completed work presents a different situation. Revenue may increase with the number of workflows executed, creating a closer connection between consumption and payment. The provider still needs to ensure that the cost of completing each workflow leaves an attractive margin.

Both cases require management to examine revenue and delivery expenses together. The assessment should answer several questions:

• Does additional AI usage generate corresponding revenue growth?
• Does revenue increase faster than the cost of serving customers?
• Are gross margins improving or deteriorating as consumption expands?
• Does customer adoption create new spending or replace existing software revenue?
• How much ongoing development investment is required to maintain and improve the product?
• Can the company demonstrate a positive contribution to overall financial performance?

The answers should be developed separately for materially different AI offerings. Aggregate revenue growth can conceal unfavorable product economics, particularly when execution costs rise as quickly as sales. Product-level measurement shows whether additional demand is improving profitability.

For example, a company might report strong expansion in agentic products while margins deteriorate because each completed workflow requires expensive computing resources. Management would then need to examine pricing, model consumption, and product design to improve the relationship between revenue and delivery cost.

An early-stage product can show the opposite pattern. Development spending may initially exceed customer revenue because the company is building functionality before adoption becomes widespread. Investors need to assess whether subsequent revenue growth and delivery costs support an attractive business as the product matures.

These patterns explain why a single margin expectation may be unsuitable for every AI category. Product maturity, pricing structure, and customer usage all influence financial results. Comparisons become more meaningful when those factors are visible.

The same principle applies when AI replaces existing software revenue. Suppose customers shift spending from conventional subscriptions to a new AI offering. Management may report impressive AI growth even though some of that revenue has moved between product categories.

The company could still benefit if the new offering improves retention, strengthens its competitive position, or delivers better margins. Establishing that benefit requires measuring the effect on total revenue and profitability.

Incremental revenue is therefore a critical measure. Investors need to distinguish genuinely new customer spending from revenue transferred between products. Combining that information with delivery costs reveals whether the shift improves the company’s overall financial performance.

Better measurement changes capital allocation

Product-level reporting becomes most valuable when management uses it to make investment decisions. Customer enthusiasm, product launches, and expectations about future AI demand can identify opportunities, but they provide an incomplete basis for allocating capital.

Revenue, cost, and adoption data allow management to compare initiatives by their financial results. An AI assistant with strong usage but limited monetization may require a pricing change. A workflow automation product with attractive sales growth but high inference expenses may need a different delivery approach or commercial model.

An established machine learning product with durable demand and attractive margins may also justify continued investment, even if newer technologies attract more attention. These comparisons help management direct resources toward products with credible prospects for improving the business.

The information also supports decisions about how far to transform existing products. A company considering a major workflow redesign needs to assess expected customer demand, development requirements, delivery expenses, and the possible effect on conventional subscription revenue.

That assessment may support a broader product transformation. In other circumstances, targeted improvements to an established offering may provide a better use of resources. The choice depends on the expected financial return and the competitive position each approach can sustain.

Once management selects a direction, continued measurement helps determine whether to expand the product, change its commercial model, or redirect engineering capacity. Tracking how revenue and expenses respond to adoption also improves forecasts for margins and cash generation.

For GPs, these measures provide a more useful basis for monitoring portfolio companies and evaluating additional investment. They also establish a financial history that prospective buyers can examine during an exit process.

That history matters because buyers need to distinguish earnings already generated by AI from future opportunities that still require execution. Consistent reporting helps management explain where value has been created and what further investment may be necessary.

Product adoption, competition, customer behavior, and technology costs can still evolve unexpectedly. Better measurement allows investors to identify which AI initiatives are producing commercial results and where the investment case depends on future performance.

AI evidence must shape acquisitions, ownership, and exits

AI-specific financial reporting affects the entire PE investment process. The same underlying information can influence which businesses investors acquire, how they develop those businesses during ownership, and how they support valuations when preparing an exit.

The decisions differ at each stage. Acquisition diligence focuses on future risks and opportunities, ownership requires choices about products and resources, and exit preparation must demonstrate the results achieved. Maintaining consistent measurement across those stages allows investors to connect initial assumptions with subsequent performance.

Acquisition: underwrite AI exposure and financial performance

Acquisition diligence increasingly requires investors to understand the customer workflows supported by a target’s software. The starting point is what customers are trying to accomplish and how the product helps them complete that work. Investors can then assess which activities AI might improve, automate, or eliminate.

This workflow analysis helps distinguish sources of competitive advantage. A product advantage may depend on specific functionality, while a workflow advantage comes from integration into the customer’s operating processes. A data advantage depends on information that helps the provider deliver valuable results.

Each advantage can face a different level of exposure to AI competition. Functionality that is easy for an AI-native competitor to reproduce may offer less protection than deeply integrated processes or valuable proprietary data. Investors need to understand which advantages customers rely on and how difficult they would be to replace.

The assessment should examine whether AI-native providers could reproduce important functionality, whether customers would continue to need the incumbent’s application, and how existing workflow integration or customer data might protect its position.

Diligence must also examine the target’s own AI deployment. Internally, investors should determine whether management is using AI to improve operating efficiency and whether those initiatives have generated measurable benefits. On the product side, they should assess AI adoption, related revenue, delivery costs, and effects on existing subscriptions.

These investigations change the emphasis of management discussions. The traditional starting question is “do they have an AI strategy?” The more consequential question is “can they show me the proof points?”

A credible strategy explains what management intends to accomplish. Financial and operational results show how far the company has progressed and help investors estimate the remaining execution risk. Those results should inform growth assumptions, margin forecasts, and the purchase price an investor is prepared to pay.

Ownership: redesign workflows where the economics justify it

Once an acquisition is completed, AI creates opportunities to improve internal operations and change what portfolio companies deliver to customers. Internal deployment can support efficiency initiatives, while product transformation requires management to reconsider how customers achieve their broader objectives.

Established SaaS companies are often organized around improving existing applications. Development teams add functionality, refine interfaces, and introduce premium offerings while maintaining the core business. This approach can produce valuable improvements, but it may be insufficient when AI can execute a larger portion of the customer’s workflow.

A broader redesign begins with the customer’s intended outcome. Management should examine the complete process required to achieve it, identify which steps are handled by employees and existing software, and determine where AI could assume greater responsibility.

The difference between assistance and automation matters here. A copilot may help an employee complete individual tasks more quickly, while an agentic system may execute a sequence of tasks and deliver a final result with less human intervention.

The operational effects can be substantial. Some applications may allow employees to redirect their time toward more productive work. In other circumstances, automation may reduce the need for particular roles.

For the software provider, the investment opportunity depends on whether customers value those changes enough to pay for them. The product must also deliver the promised work reliably and at a cost that supports an attractive margin.

Management then needs to decide how far to change the existing product. Some companies may benefit from focused AI capabilities added to established software. Others may need to reconsider their original product assumptions and develop offerings organized around complete workflows.

A major redesign can require new technical expertise, different development practices, and organizational changes that allow teams to execute quickly. These requirements are particularly challenging for incumbents that must continue supporting established products and customer commitments.

Development capacity is limited, so pursuing a transformation creates trade-offs with improvements to the existing business. Management needs to compare the expected revenue opportunity with development investment, delivery costs, execution risks, and competitive consequences.

Financial measurement should guide that decision. Product-level revenue, cost, and adoption data can show where AI is gaining traction and whether the results justify additional resources. Once a direction is selected, management must build the necessary capabilities and scale the offering.

Continued measurement then determines how the roadmap evolves. Strong customer demand and improving margins may support expansion, while weak monetization or rising delivery expenses may justify changing direction. This connects product strategy to the financial results achieved during ownership.

Exit: demonstrate value before entering the sale process

Exit preparation brings AI measurement directly into valuation discussions. Software sellers may have invested heavily in AI development and expect buyers to recognize the future value of those efforts. Buyers need to assess how much of that expected value is already visible in financial performance and how much depends on future execution.

This difference in expectations contributes to valuation disagreements between buyers and sellers. The disagreement becomes harder to resolve when management has extensive records of AI development activity but limited information connecting that work to revenue and profitability.

A portfolio company that begins measuring AI economics early in ownership is better positioned to address the issue. Management can demonstrate incremental AI revenue, identify changes in operating costs, and explain how margins behave as customer usage increases. It can also show how financial results have influenced product development and resource allocation.

That operating history gives prospective buyers a clearer basis for evaluating the durability of the business. They can examine whether customer adoption has translated into revenue growth, how much each product costs to deliver, and which competitive advantages may protect future earnings.

The same evidence helps buyers identify remaining risks. Established earnings can be evaluated separately from growth opportunities that require substantial development investment or depend on uncertain customer adoption.

Stronger measurement can support a more credible valuation discussion, although it cannot guarantee a higher transaction multiple. Buyers will still assess competitive threats, customer behavior, technology expenses, and the investment required to sustain growth.

For sponsors, exit preparation therefore begins during ownership. Financial reporting should establish a consistent record of AI-related revenue, costs, adoption, and profitability as products develop.

That record becomes especially important when a portfolio company moves from selling conventional software access toward delivering automated work. Its historical subscription metrics may describe the established business accurately, while product-level AI results show how the company’s future earning power is developing.

By the time the sale process begins, management should be able to connect its product decisions to observed financial outcomes. Buyers can then underwrite the established business and its AI opportunities using evidence accumulated throughout the ownership period.

In conclusion

Software investing is entering a period in which historical growth and profitability are becoming less reliable indicators of future performance. Slower revenue expansion, pressure on private equity returns, and the emergence of AI-native competitors are changing how investors assess software businesses. Strong valuations remain possible, but sustaining them increasingly depends on demonstrating durable competitive advantages and credible earnings growth.

AI adds both opportunity and complexity. It can create new revenue streams, improve operating efficiency, and expand the work software performs for customers. It can also increase delivery costs, weaken traditional subscription models, and expose established products to new competition. Understanding these effects requires more than tracking product launches, customer adoption, or aggregate SaaS metrics.

For executives and investment leaders, the priority is to connect AI strategy to measurable financial outcomes. Separating AI revenue, development spending, and recurring delivery costs provides a clearer view of which initiatives are creating value. It also helps management determine where to invest, which products to redesign, and how to balance innovation with the profitability of the established business.

These decisions will shape acquisition underwriting, portfolio performance, and exit valuations. Companies that establish consistent AI financial reporting during ownership will be better equipped to allocate capital, demonstrate progress, and support their investment case when entering a sale process.

As AI changes how software is developed, delivered, and purchased, the central investment question is shifting from how quickly a company can grow to whether its business model can sustain profitable growth. Executives who can answer that question with financial evidence will have a stronger foundation for making investment decisions and defending valuations in a changing software market.