Software stocks lost roughly a trillion dollars in market value at the start of 2026, and the label that stuck, “SaaSpocalypse,” pushed everyone toward the same question of whether specific software companies will still exist. That’s not the right question, or at least it’s an incomplete one. The better question is whether a company can protect its terminal value: its ability to remain competitive with emerging AI companies and embrace the same mindset and pace of innovation. The trillion-dollar figure looks like the whole category is dying. What’s happening is actually dispersion. Some software companies have clawed back double-digit gains since the selloff. Others are still down 60% or more from their highs. The category-wide story hides the more useful one underneath it.

The Real Divide

Morgan Stanley’s “Moat & Journey” framework explains this well and it’s a three-way split, not two. There are companies with no real moat: thin products sitting on top of someone else’s workflow, that’s easy to replicate now that AI has collapsed the cost of building a passable clone. There are companies with a moat that have gone complacent, meaning real defensibility today but no reinvestment into what comes next. That erodes terminal value more slowly than having no moat at all, but it still erodes it. And there are companies with a moat that are using AI to extend it, deepening the workflow, the data, and the trust they’ve already built with customers. Only that third group is actually protecting its terminal value. 

The Practical Test

That three-way split is also why build vs. buy keeps resurfacing as a live question inside finance organizations, and it’s the same math I laid out in a previous issue: buy the foundation, build the advantage. The SaaS market is now proving that logic at scale, with company valuations instead of implementation budgets.

It’s the same reason finance organizations buy their systems of record instead of building them. Tax law, GAAP and audit standards all change. A homegrown system built to answer today’s regulatory questions is obsolete the moment those questions change, and in finance that’s constantly. Buying means a vendor absorbs the cost of keeping pace with that change across thousands of customers instead of one. AI is putting the same pressure on every other build decision now, because the target you’re building toward doesn’t hold still long enough for a build to ever really finish.

I heard a version of this from a banker who made exactly this call early in his career. His team built a software solution instead of buying one. It worked, until the engineers who built it left the company. No one who remained understood the system well enough to update it, and the cost of that decision eventually ran into the millions. AI coding tools make that story more common, not less. They make the initial build faster but they don’t make the code underneath it any more documented or transferable.

The Finance Bar

Finance has less room for error than most functions making this call. A 90% accurate answer isn’t good enough when the output feeds external reporting, revenue recognition and billing are examples that need to be 100% accurate. There are real consequences for being even slightly off that can damage both the company and leaders’ reputations. A homegrown system also has to hold up under audit and, if it ever comes to it, in a court of law. 

Before greenlighting a build now, we ask these questions: 

1. Is this core to what we sell, or infrastructure that supports it? 
2. Will this still be differentiated in eighteen months, or will a vendor ship the same capability as a feature and force us to rebuild around theirs? 
3. If the engineer who built it with an AI coding tool leaves the company, does anyone else actually understand how it works?
4. Will this hold up under audit and regulatory scrutiny?

The Discipline That Protects Terminal Value

Getting this right comes down to a single, disciplined habit: pricing the full lifecycle cost before the team falls in love with a prototype. A quick win might justify a custom build, but a long-term need requires a clear plan for ownership and maintenance to ensure it doesn’t become tomorrow’s tech debt. The same test applies when you’re the one being evaluated as a vendor. The moat is no longer the interface. It’s how deep the product goes, how much a customer can trust the system, and how well it fits into how the business actually runs. Products can be copied faster than ever, but monetization models, workflows, and trust built up over years still can’t be.

Not all SaaS is created equal, and AI didn’t create that gap. It just made the market start pricing it, based on who is protecting terminal value rather than who has AI. That’s the same test finance should be applying to every build decision it signs off on.

Continue The Conversation

From Product Launch to Recognized Revenue: A Zuora AI Journey Through Quote-to-Cash 

If terminal value increasingly comes from how well a company can operationalize complexity, implementation speed becomes part of the story. Moving from months to weeks means faster time to value, earlier visibility into downstream decisions, and more capacity for the next strategic priority. In this on-demand session, learn how AI can turn the contracts, pricing rules, and workflows a business already has into a working model of its quote-to-cash process before implementation begins. 

Hyland case study 

For Hyland, the value story is less about speed alone and more about what better finance infrastructure makes possible over time. With Zuora Revenue, the team cut its close from 12 days to 3, automated SSP allocations and contract carve-outs, and gave revenue accounting more capacity to review deals earlier and support the business more strategically.

To address the shift from predictable SaaS revenue to dynamic AI models, finance teams must move from static annual planning to continuous, usage-based forecasting. Start by auditing your current data pipeline to ensure you can track real-time infrastructure costs alongside customer usage. Next, establish a tight feedback loop between engineering, IT, and FP&A to map token consumption or compute power directly to unit economics, allowing you to adjust pricing and investment levels before margins are impacted.

Every AI request has an associated cost, and because customer usage varies significantly from one account to another and infrastructure expenses fluctuate based on model selection, the relationship between revenue and cost is becoming much more dynamic. This complexity makes it harder to forecast profitability with the same level of confidence, raising an important question: Does finance have the insight it needs to understand the financial drivers behind AI?

To gain the foresight needed to forecast margins, evaluate pricing, and guide investment, finance teams must proactively shift their planning focus. First, audit how AI-specific drivers differ from traditional SaaS metrics – usage, infrastructure costs, and overages. Second, identify data gaps: Can you track margins by customer? Costs per request? Usage vs overages? Third, restructure your planning models to account for these unique variables. 

The sections below show how—and where they differ from traditional SaaS.

Shift from fixed metrics to variable AI-driven drivers

While the core objective of planning—protecting margins and guiding investment—remains the same, the drivers have fundamentally shifted. Traditional planning relied on stable metrics like headcount and fixed subscription tiers. In the AI era, these are replaced by variable infrastructure expenses, model selection costs, and highly individual customer usage patterns. Finance must now reconcile the fixed costs of development with the fluid, request-based costs of fulfillment.

That requires finance to understand new drivers of business performance: the cost to fulfill AI requests, which products generate the strongest margins, how customers consume infrastructure resources, and how usage patterns, such as unused credits and overages, influence profitability. These metrics form the foundation for forecasting performance and identifying where the business creates value and where it should invest next.

Take committed AI credits as an example. Not every customer will consume credits the same way. Some will leave purchased capacity unused, while others will consistently exceed their commitments. Those usage patterns influence margins, pricing decisions, and future investment. Understanding them helps FP&A forecast performance more accurately and identify opportunities to refine packaging, improve adoption, or adjust pricing over time.

These metrics help FP&A forecast performance, evaluate investments, and plan with greater confidence as AI adoption grows.

Partner with product and engineering to design your data model

Many of the insights FP&A needs can’t be created after an AI product launches. By then, decisions about usage data, cost attribution, and reporting have already been made. If finance isn’t involved early, planning becomes much more difficult.

FP&A should work alongside product, engineering, and operations as new AI offerings are designed to ensure the business can measure the metrics that matter most. That includes understanding how usage will be tracked, how costs will be attributed, and how performance will be evaluated over time.

That also means thinking ahead about the metrics finance will need once AI products are in market. How predictable is unused capacity across different customer segments? What is the margin impact if the business chooses not to charge for certain overages? Do usage patterns change as customers mature? These are planning questions, and answering them requires the right data from day one.

When those building blocks are in place from the beginning, FP&A can forecast with greater confidence, evaluate pricing decisions more effectively, and help the business adapt as customer behavior evolves.

Use granular usage insights to optimize profitability

Better insight also changes the conversations FP&A can have with the business. Instead of simply reporting that AI adoption increased, finance can help answer more strategic questions. Are we investing in the right AI capabilities? Which customer segments generate the healthiest margins? Should we adjust pricing or packaging? Are infrastructure costs scaling as expected?

Those are the decisions that shape long-term growth. When FP&A understands the financial drivers behind the business, it can move beyond reporting results and become a stronger strategic partner.

The future of FP&A

As AI continues to reshape how companies build, price, and monetize products, FP&A has an opportunity to help shape how those businesses grow. Finance leaders who proactively understand their data pipelines, collaborate with product and engineering, and build AI-specific planning models will guide better investment decisions, evaluate new business models, optimize pricing, and adapt faster as markets shift.

That is where FP&A can create lasting strategic value.

Keep Exploring

Revenue Complexity Assessment

More bundles, global expansion, and AI product pricing are creating a level of revenue complexity that many teams are struggling to manage. Take Zuora’s assessment to identify the top drivers of complexity in your business and get tailored recommendations on how to start addressing them.