RALEIGH, N.C. – November 5th, 2026 – Traditional SaaS businesses have long targeted 70% to 80% gross margins, a benchmark that shaped everything from pricing strategy to valuation multiples. K-38 Consulting says that benchmark no longer holds cleanly for companies layering AI features into their products. Industry data from ICONIQ’s 2026 State of AI report puts average AI product gross margins at approximately 52%, up from 41% in 2024 — a meaningful improvement year over year, but still well below the margin profile investors and founders have used to evaluate SaaS businesses for the past decade.

“A software company can add an AI assistant to an existing product and watch gross margin on that feature drop from 80% to somewhere closer to 65% almost overnight,” said Dallas Alford IV, CPA, Founder of K-38 Consulting. “That’s not a rounding error — it’s a structural shift in the cost base, and a lot of finance teams are still building their financial models as if the old cost structure still applies.”

Why AI Costs Behave Differently Than Traditional SaaS Infrastructure

K-38 Consulting says the core issue is that AI inference costs behave fundamentally differently than the infrastructure costs that shaped traditional SaaS economics. Traditional hosting costs grew sub-linearly as a company scaled — more customers generally meant better infrastructure utilization and declining unit costs. AI inference costs scale linearly, or worse, with usage: every user interaction that triggers a model call carries a real, variable cost that doesn’t shrink simply because the company has grown.

That dynamic means AI cost-to-serve can vary dramatically even between customers on identical pricing plans, depending on how heavily each customer actually uses the AI features — a level of cost variability traditional SaaS finance teams generally never had to model at the individual customer level.

“The old finance playbook assumed cost of goods sold was relatively stable and predictable across a customer base,” Alford said. “AI breaks that assumption. Two customers paying the identical subscription price can have order-of-magnitude different costs to serve, based purely on how they’re using the AI features.”

Where the Margin Pressure Is Concentrated

K-38 Consulting says the extent of margin compression varies significantly depending on how deeply a company relies on AI as part of its core product:

AI-augmented products — traditional SaaS products with AI features layered on top of an existing core offering — tend to hold closer to traditional margins, often around 80%, since AI represents a smaller share of overall cost of goods sold.

AI-enabled products — where AI plays a more central role in the product’s value proposition — typically see gross margins in the 60% to 79% range, reflecting AI’s larger share of the overall cost structure.

AI-native products — built around AI as the core product itself, rather than an add-on — commonly run in the 50% to 65% range, with some fast-scaling AI-first companies reporting margins as low as 25% in early stages before cost optimization takes hold.

Notably, industry-wide benchmark data shows the median public and late-stage private SaaS company hasn’t yet seen this compression show up clearly in aggregate reporting — meaning the pressure is real and significant in AI-native and AI-forward companies specifically, without yet being visible in broader industry averages that blend many types of companies together.

“This is exactly the kind of shift that gets missed if a finance team is only looking at company-wide, blended gross margin,” Alford said. “A business can have a healthy blended number that’s masking real, significant margin compression happening specifically in its AI product line.”

How Companies Are Responding

K-38 Consulting says the SaaS companies managing this transition most effectively are treating AI cost structure as a distinct financial reporting category, rather than folding it into general cost of goods sold:

Isolating AI costs as a separate P&L layer. Structuring financial reporting to separate traditional COGS from AI inference and infrastructure costs allows a company to see AI-specific gross margin clearly, rather than having it obscured inside a blended company-wide number.

Adopting hybrid or usage-based pricing. Pricing structures that combine a stable subscription base with a usage-based AI layer have grown from roughly 25% to 37% adoption among surveyed B2B software companies in a single year — a direct response to aligning pricing with the variable nature of AI costs.

Optimizing model architecture to control cost. Techniques including tiered model routing — using smaller, cheaper models for simple queries and reserving expensive frontier models for genuinely complex tasks — and prompt caching, which can cut effective per-query costs significantly, are increasingly treated as financial levers rather than purely technical decisions.

“The companies protecting their margins aren’t necessarily using less AI,” Alford said. “They’re being more deliberate about how they price it and how efficiently they’re running it — and that requires finance and engineering working from the same cost data, which isn’t how most SaaS finance functions have historically operated.”

What K-38 Consulting Recommends

Based on the patterns it sees among SaaS clients adopting AI features, K-38 Consulting recommends companies:

• Separate AI-related costs from traditional COGS in financial reporting, rather than allowing AI infrastructure and inference costs to blend into a single, less informative gross margin figure.

• Model gross margin at the product or feature level, not just company-wide, since blended margins can mask significant compression happening in AI-specific product lines.

• Reassess pricing strategy alongside cost structure, particularly evaluating whether a hybrid or usage-based pricing model better reflects the variable nature of AI costs than a flat subscription alone.

• Track cost-to-serve variability at the customer level, given how significantly AI usage patterns can differ between customers on otherwise identical plans.

• Coordinate finance and engineering on cost optimization decisions, since architecture choices like model routing and prompt caching have direct, measurable financial impact.

How K-38 Consulting Supports SaaS Companies

K-38 Consulting’s SaaS CFO services help software companies build the financial reporting structure needed to track AI’s real impact on gross margin, separating traditional and AI-driven costs so leadership can see the true economics of their product line. This work is part of the firm’s broader outsourced CFO services, helping SaaS founders make pricing and product decisions based on accurate, current cost data rather than a financial model built for a pre-AI cost structure.

“Gross margin has always been one of the most important levers in a SaaS business,” Alford said. “AI hasn’t eliminated that importance — it’s just made the calculation more complicated, and companies that don’t update their financial reporting to match are flying without the visibility they need to protect their margins.”

About K-38 Consulting

K-38 Consulting provides fractional and outsourced CFO services, controller services, and tax strategy — including R&D tax credit and cost segregation services — to startups and midsize businesses across the country. The firm serves clients in SaaS, biotech, healthcare, law, ecommerce, CPG, construction, and real estate, delivering the financial leadership, forecasting tools, and strategic guidance typically available only to companies with a full in-house finance team. K-38 Consulting is headquartered in Raleigh, North Carolina, with clients nationwide.

Media Contact: K-38 Consulting 3809 La Costa Way, Raleigh, NC 27610 (910) 262-4412 (https://k38consulting.com)

Media Contact
Company Name: K38 Consulting, LLC
Contact Person: Dallas Alford
Email: Send Email
Phone: 9102624412
Address:3809 La Costa Way
City: Raleigh
State: NC
Country: United States
Website: https://www.k38consulting.com/