Why CFOs Are Ditching Old Models and Rewriting the Playbook for AI
The smartest finance leaders aren’t just managing numbers. They are building intelligent systems to forecast, price, and scale with precision in an AI-first world.
If you think AI is just a technical initiative, you’re missing the point.
AI is redefining how companies run. And no one is feeling this shift more than the office of the CFO.
In Andreessen Horowitz’s recent article, “How CFOs are navigating growth, pricing, and forecasting in an AI world,” finance leaders from Databricks, ElevenLabs, Ambient.ai, Together AI, and Concourse dropped serious knowledge on how AI is changing the DNA of financial operations.
Let’s break down the most actionable takeaways, and how you can adapt fast, even if you’re not running a billion-dollar enterprise.
1. Pricing is now about outcomes, not effort
Old model: Charge for licenses or monthly access
New model: Charge based on value delivered
AI-native companies are flipping pricing on its head:
Databricks only books revenue when customers derive value. If usage drops, so does their revenue.
ElevenLabs gives discounts to drive volume while locking in larger commitments. Strategic, scalable, smart.
Startups like Concourse? They changed their pricing more than seven times in the 40 days post-launch. It’s something they will keep iterating on and improving.
💡 What does this mean?
In an AI-powered world, pricing isn’t a one-time decision…it’s an iterative process.
You can move just as fast. Tools like Stripe, Outseta, or Lemon Squeezy let you run usage-based and outcome-based pricing experiments. Pricing is no longer set-and-forget. Right now, it’s test, learn, and repeat.
2. ARR is broken (but you can fix it)
Traditional ARR (Annual Recurring Revenue) was built for a SaaS world where pricing was flat, predictable, and tied to time-based contracts.
But AI doesn’t work that way.
In AI-driven businesses, revenue is tied to actual usage, not just subscriptions. And that usage? It’s messy, nonlinear, and highly variable.
Take ElevenLabs, for example. Their enterprise customers routinely exceed their usage quotas. If they only tracked ARR, they’d be undervaluing their growth, underreporting performance, and underplanning capacity.
So what did they do?
They created a hybrid KPI: ARR + Annualized Usage.
This lets them account for committed revenue and the surges in real-time consumption that drive upside.
“Without this, we’d be underselling what we actually earn.” – Maciej Mylik, Finance at ElevenLabs
In AI, your biggest revenue driver may not be the annual contract. It’s how fast customers scale usage once they see value. So if you’re only tracking ARR, you’re flying blind on your upside potential and your infrastructure needs.
💡 The insight?
ARR doesn’t reflect how AI companies actually earn, grow, or scale. You need new metrics that blend predictability with flexibility.
You don’t need a massive finance team to do this.
Start by modeling usage-based growth with tools like Pigment or Anaplan. What matters is that your KPIs reflect how your product is actually used.
3. Gross margins are under attack
Every API call to a foundation model (think OpenAI or Anthropic) = new cost. In AI, every token, API call, and GPU hour has a cost. And margins can erode fast if you’re not watching closely.
At Together AI, finance tracks something they call “regrettable idle GPU time”because every unused GPU impacts margin.
This highlights a major shift: unlike SaaS, where user growth is mostly free, AI infrastructure is expensive and dynamic.
CFOs now need to think like operators. It’s not just about controlling spend. It’s about optimizing usage in real time. That means tighter collaboration with engineering, smarter pricing tied to actual cost, and ongoing deal re-evaluation to protect margins as models evolve.
If you’re building anything AI-powered, assume your cost structure isn’t fixed. So your pricing model can’t be either.
💡 DIY Tip:
Use Datadog, Finout, or CloudZero to set up real-time cost observability without custom builds.
4. You must invest in R&D even if the ROI isn’t immediate
Yes, AI commoditizes quickly. But, no. You cannot afford to wait.
In the AI era, not every investment maps cleanly to near-term revenue. And that’s exactly the point. The smartest CFOs are doubling down on strategic R&D to create long-term differentiation and defensibility.
At Databricks, CFO Dave Conte points to Unity Catalog as a clear example. It started as an internal R&D bet, but over time, it became a critical adoption driver, increasing customer stickiness and boosting revenue through expanded usage.
ElevenLabs is thinking just as strategically. They’re not content with building great voice models. They’re layering in workflows, APIs, and data-rich features that make their platform indispensable. As text-to-speech commoditizes, they’re already a step ahead, creating switching costs and embedding themselves deeper into customer operations.
💡 The Takeaway?
In AI, short-term efficiency alone won’t protect you. The CFOs winning in this space are helping their companies invest where others hesitate. And it’s those investments that will define who leads as the market matures.
5. Forecasting with Excel? That’s a liability now.
You can’t survive in a nonlinear world with linear tools.
The tools for forecasting in AI simply don’t exist yet (at least not off the shelf).
That’s why leaders like Databricks are turning inward, using their own AI platform to build custom forecasting systems that predict usage by customer, product, and workload. Why? Because one month a customer surges, the next they optimize. New use cases pop up overnight.
The only constant is change — and trying to force traditional planning models onto that reality just doesn’t work.
Most finance teams are flying blind. Even ElevenLabs admits no one has fully cracked revenue forecasting for AI.
But here’s the opportunity:
For CFOs: The internal play is clear. If you can build tools that adapt to constant change, anticipate usage patterns, and set dynamic sales quotas, you’ll create a level of precision your competitors simply don’t have. This is your edge. The AI-native CFO is designing the finance stack from the inside out.
For founders and builders: This is your signal. The tooling CFOs need doesn’t exist yet, not in the way AI-native businesses demand it. That’s a massive whitespace for product innovation. If you can build solutions that help finance leaders model usage-based revenue, track margin volatility, or forecast AI-driven growth in real time, you’re not just solving a pain point, you’re unlocking enterprise budgets.
This isn’t just a challenge. It’s a category-defining opportunity.
💡 No Code Tip:
Plug your CRM or billing data into tools like Falkon, Polymer, or Obviously.ai to auto-generate predictive dashboards without a single line of code.
Final Take
AI isn’t just changing how we build products. It’s transforming how we run the entire business.
Today’s most forward-thinking CFOs aren’t reacting to AI, they’re rewriting the finance playbook around it.
From outcome-based pricing to real-time cost tracking
From redefining ARR to investing in R&D with conviction
And building forecasting tools in-house
In this new AI economy, you need clarity, speed, and the courage to test, measure, and evolve. The real risk isn’t making the wrong decision. It’s waiting too long to decide.
Change is the only constant. And the CFOs willing to embrace that will own the next era of growth.
#AIforFinance #NoCode #CFOPlaybook #FutureOfWork
Which one of these five shifts is your team already feeling?
Drop it in the comments ⬇️
Let’s build smarter.
– Angela
♻️ Repost to help someone in your network unlock growth through no code and automation.
Follow Angela Stewart for more.
Subscribe here 👇
