Well, the internet did what the internet does best this week: it collectively lost its mind over a number in an S-1 filing. Figma disclosed they signed a ~$550 million contract with AWS, someone used arithmetic (the secret weapon of Cloud Finance) to determine that this was roughly $300,000 per day on AWS, and suddenly everyone with a social media account became a cloud economics expert.
“FIGMA IS DOOMED!” they cried. “FIRE THE CTO!” they demanded. “They’re gonna blow through their commitment in record time if they spin up another Managed NAT Gateway,” I observed.
Here’s the thing: I negotiate AWS bills for a living. My company, The Duckbill Group, has seen more AWS invoices than most people have seen Marvel movies. And I’m here to tell you that Figma’s spending is about as scandalous as “Mr. Roger’s Neighborhood.”
Let’s Talk Numbers (The Real Ones)
Figma’s S-1 filing revealed a $545 million five-year AWS commitment. That’s about $109 million annually, or roughly 12% of their $821 million in rolling revenue. For those keeping score at home, that’s supporting 13 million monthly active users and 450,000 paying customers while maintaining a market-leading 91% gross margin.
But sure, let’s panic about the AWS bill.
While we’re at it, let’s also make sure we get the math wrong in two key ways.
First, let’s ignore that most AWS contracts increase year-over-year (if they don’t, the company either isn’t growing, or doesn’t expect to grow much further, which is absolutely not Figma), so the idea that they’re currently spending $110m under this contract year isn’t reasonable. It’s far likelier it starts at something like $80m a year, ramping up to ~$150m a year in Year 5.
Second, what everyone’s missing is that these spend targets are post discount. Remember that Figma didn’t sign a half-billion dollar non-cancelable contract for funsies. At this scale, with no specific knowledge of their deal, they likely have something on the order of a 30% effective discount on their cloud spend. So when you scroll through AWS’s byzantine pricing pages doing mental math, you can think of Figma spending not $550m, but $785m at retail prices.
The best part? Rather than pointing out realities like this, someone on Twitter instead claimed Figma only makes “$220m a year in revenue” (actual: $749M) while spending “$110m a year on AWS.” Even Twitter’s community notes had to step in, which is like being fact-checked by a YouTube comment section.
This Is What Normal Looks Like
Here’s what the pearl-clutchers don’t fully grasp: 12% of revenue on cloud infrastructure for a compute-intensive, real-time collaborative platform is completely reasonable. We see this all the time with our clients.
Industry benchmarks show:
- Compute-lite SaaS companies: ~5% of revenue
- Compute-heavy platforms: 10–15% of revenue
- AI/ML-intensive companies: Often exceeding 15% (but do note that it’s so early days right now it’s hard to speak definitively here)
Figma is rendering complex designs in real-time for millions of users with sub-100ms latency. They’re not running a blog on a t3.micro instance. This is heavyweight technical infrastructure, and it costs what it costs.
The “AWS Dependency” Non-Story
My favorite part of this whole saga is the breathless reporting about Figma’s “risky dependency” on AWS. The S-1 contained standard boilerplate language about vendor dependencies that appears in virtually every cloud-dependent company’s SEC filings. You know, the kind of language lawyers make you include about how your business could be affected if your primary infrastructure provider suddenly decides to take up goat farming instead, or your CEO is Elon Musk.
Breaking news: SaaS company uses cloud provider, could be affected by outages. In related news, restaurants report troubling dependency on food suppliers.
Let’s Add Some Context, Shall We?
Want to know what real AWS spending looks like at scale?
- Snowflake has a $2.5 billion commitment from 2024–2028
- Apple was spending an estimated $360 million annually back in 2019
- Snap got over their skis by making $3 billion in cloud commitments, spending over 40% of revenue before having to renegotiate
But yes, let’s all clutch our pearls about Figma’s completely reasonable infrastructure costs.
The Armchair Architects Have Logged On
HackerNews commenters claimed they could cut Figma’s costs by “at least 30%, often more than half.” Sure, Steven; that seems credible. I’m certain your experience running a Minecraft server uniquely qualifies you to architect infrastructure for 95% of Fortune 500 companies.
These are the same people who think they could rebuild Facebook over a weekend if only they didn’t have their day job at the DMV.
Here’s a hint: you might have a misconfiguration costing you 20% of your AWS bill when you’re dropping $80K a year, but by the time you’re annually spending nine figures, somebody has asked where the hell all the money is going.
Here’s What Actually Matters
Figma has documented optimization efforts including transitioning from Ruby to C++ pipelines, and migrating workloads that don’t need it away from GPU-based instances. They’re implementing dynamic cluster scaling. They’re doing the work.
More importantly, they’re growing at 46% year-over-year with a 91% gross margin. If you’re losing sleep over their AWS bill while they’re printing money like this, you might want to reconsider your priorities. Once again the innovation <–> optimization continuum rears its head.
If you enjoyed this analysis, subscribe to Last Week in AWS for more cloud commentary, absurdity, and actual insights that won’t make your finance team cry.
The Bottom Line
After years of helping companies optimize their AWS spending, I can tell you definitively: Figma’s infrastructure costs are boringly normal. The real story here isn’t their AWS bill—it’s that they were transparent enough to disclose it when most companies hide behind vague “infrastructure costs” line items.
The hyperbolic reactions to this disclosure reveal something more concerning than any AWS bill: a fundamental misunderstanding of what it actually costs to run enterprise-scale infrastructure in 2025.
So the next time someone tries to tell you that a company spending 12% of revenue on the infrastructure that literally runs their entire business is “doomed,” maybe ask them how much they think it costs to serve real-time collaborative experiences to 13 million users across the globe.
Spoiler alert: It’s not cheap. And that’s perfectly fine.
