For the past 15 years, the consensus for how SaaS monetization worked was straightforward and relatively static: optimize for human users, center the value conversation around seat-based subscriptions, and revisit pricing tiers annually (at most). Today, as AI rewrites the underlying architecture of software, it’s forcing an industry-wide reckoning with that inherited playbook.

“AI is kind of ripping through everything, but one of the unintuitive things about AI is that it's incredibly expensive. And so what it means is that businesses that used to not need to think too much about operating costs and not need to think too much about how their infrastructure impacts their actual product or their revenue bottom line. They're now being forced to.”
— Scott Woody, Metronome founder and CEO 
_{‍Watch it here}

To understand where the broader market is headed, companies have to look at today’s infrastructure layer. In a recent webinar, Scott sat down with Ville Lehto, VP of Strategy at Aiven, to discuss the structural forces that are dismantling legacy monetization models. What came out of their conversation is a new playbook for commercialization that’s defined by radical flexibility of delivery models, architectural margin shifts, and a new class of buyer: the autonomous agent.

Force 1: Shifting delivery models (managed SaaS to BYOC)

In the software era, fully managed SaaS was the gold standard. Vendors carried the operational infrastructure, marked up the underlying compute, and had predictable, high-margin revenue. Today, that model is running into a wall of intense customer scrutiny around total cost of ownership. Buyers are increasingly demanding that vendors ship software directly into their own tenant via bring your own cloud (BYOC) models. This shifts the infrastructure cost burden back to the customer, letting them draw from their own cloud enterprise discounts and bypass double-charging on network fees.

As Ville explained, this completely alters the commercial relationship:

“In the SaaS era, those were the happy days of 80% margins. Nobody really needed to think about what’s happening downstream with your data infrastructure when a human is the user. But now infrastructure is putting everything under much more scrutiny. Hence, we’re also seeing many more customers asking for bring-your-own-cloud deployments where the economics are typically more fruitful for a large use case.”
— Ville Lehto, Aiven VP of Strategy
_{‍Watch it here}

When deploying in a customer’s own cloud, vendors no longer have the leverage to arbitrarily mark up raw compute or network usage. Instead, pricing models now must anchor on the absolute economic value of the software, instead of the alternative of the customer building it from scratch. “Buyers are really, really looking into the trade-off between hiring a team to do it by themselves or having a vendor that they can trust to run this infrastructure for them,” said Ville.

Force 2: Architectural shifts and the storage moat

The second structural force rewriting the approaches to monetization is more technical: the separation of compute and storage. Across data and cloud architectures, object storage has become the default disk layer, compressing the pure cost of hosting data by up to a factor of 10.

“These used to range from 10 to 20¢ per gigabyte, but now we're talking about public storage pricing,” Ville pointed out. “So it's a tenfold difference in terms of unit economics when you look at the pure storage as these architecture shifts are happening.”

This compression means that passing through heavily marked up storage costs to customers isn’t a viable commercial strategy anymore. Instead, winning vendors are pricing storage at near-zero markup as a defensive play, just to win through data gravity. If your platform is cheap enough to house a business’s entire data estate, you capture the entry point. The real monetization opportunity happens recursively downstream, where you earn the right to charge for transformations, metadata enrichments, context-matching, and inference queries. As Ville put it, “the value is on top of the data.”

Force 3: When the buyer is a machine (the agentic profile)

Human-centered pricing design optimizes for predictability and low cognitive load. Humans like simple, per-month numbers. The trouble with this is that these human-first conventions completely break down when software usage is driven by AI agents.

Agentic workloads are spiky, nonlinear, and volatile. An agent doesn’t navigate software linearly; it spins up parallel queries, executes them in seconds or minutes, and then drops back down to zero. If a vendor bills based on traditional hourly increments, they become cost-prohibitive or structurally incompatible with agentic platforms. Metering engines need to evolve to track active active-seconds and granular units like raw CPU time and RAM time.

Even more than that, an AI agent doesn’t care about and isn’t really even affected by cosmetic pricing tricks or tiered simplicity. “Every agent is the most technical buyer you could ever have,” Scott said, “and it can handle a level of data complexity that even the best humans could never even come close to doing.”

It’s likely that agent buyers will optimize purely against value functions, machine-readable rate cards, low-latency spend telemetry, and hard programmatic budget caps. Given what we’re seeing now, it shouldn’t come as a surprise when traditional human-facing pricing pages quickly become a secondary asset compared to machine-readable pricing APIs and real-time budget guardrails.

Looking ahead: Building pricing as a capability

If the macro environment is shifting this quickly, treating pricing like a periodic, cross-functional project managed by committee is ossifying into an existential liability. Almost regardless of domain, agility itself seems to be the clearest, next competitive moat. The companies winning the AI transition are moving on the offensive, using highly fluid pricing and packaging changes to aggressively win market share.

“The only constant thing is that the market only appreciates innovation,” says Ville. “If you cannot get new stuff out despite the fact that the market around you is changing very fast, then you will lose. What we’re pushing on heavily internally is that we need to keep launching new products and new pricing models and iterate on those very aggressively, and we are doing so almost every single month right now.”

Operationalizing this level of continuous pricing iteration requires two internal requirements:

1. Abstracted primitives
   To tame the operational nightmare of pricing matrix explosion (e.g., managing permutations across multiple clouds, hundreds of regions, and dozens of products), companies need to decouple billing from raw instance costs. Aiven achieves this by abstracting compute into standard units and applying simple, global regional multipliers off a base region, like us-east-1. This strips out data complexity and lets sales and product teams deploy new pricing configurations in just hours.
   
   
2. Granular cost telemetry
   Product managers can no longer build software in a vacuum, oblivious to infrastructure margins. Because LLM pass-throughs and spiky agent workloads carry immediate financial effects, product leaders should probably be treated as mini GMs who own the bottom line and the top line, equally. Unlocking this means investing in effective telemetry to tag and trace every infrastructure cost down to the customer and service-ID level. As Scott stated, "If you never make them accountable for cost, then they just never have to flex that muscle."

In the value era, your billing infrastructure is a live, real-time product runtime system. The quality of your financial plumbing will ultimately dictate how fast your business is allowed to grow.

📺 Watch the full webinar on The New Monetization Playbook for Data Infrastructure.