Sour FIG: Figma’s future is uncertain despite IPO hype

SAN FRANCISCO — It was a hot June morning in 2023 when Jordan Singer took the stage at Config, Figma’s annual conference, playfully dubbed “Coachella for designers.” Just months after his AI design startup, Diagram, was acquired by Figma, Singer offered a glimpse of the future of AI in Figma.

The crowd, mostly designers and product managers, greeted the vision with enthusiasm. The company was riding high—Adobe had agreed to acquire it for $20 billion, and Figma had become the de facto canvas for software design.

Two years later, that future remains unwritten.

The blockbuster $20 billion acquisition was scuttled. Singer and other colleagues from the Diagram acquisition departed Figma, prompting questions about execution and internal alignment. Figma’s first AI initiative, “Make Designs,” was apologetically pulled from public use after backlash that its outputs were derivative—nearly indistinguishable from Apple’s own apps. At Config 2025, Figma raced to present a version of the company that was positioned to impress Wall St, in anticipation of its IPO announcement in a month. Having built up its position as a category-defining tool, Figma now faces a reckoning: can it lead in the AI era—or be overtaken by it?

The S-1’s Dual Narrative

Figma’s draft S-1 filing lays out a blockbuster growth story: $749 million in revenue for fiscal 2024, up 48 percent year-over-year, and $228.2 million in Q1 2025, up 46 percent versus a year earlier. Annual recurring revenue stood at $912 million, and Q1 2025 net income hit $44.9 million—its first profitable quarter this cycle. The platform boasts 13 million monthly active users—two-thirds non-designers—underscoring its cross-functional reach. Net Dollar Retention clocks in at 132 percent as of March 31, 2025, and 76 percent of customers now use two or more Figma products. Yet buried in the risk disclosures is a blunt warning:

“In the short term, we expect that our AI investments … will negatively impact our gross margins and operating margins … [and] the extent of such impacts … are currently unknown.”

This footnote sits uneasily beside claims of near-90 percent gross margins and a balance sheet cheekily buoyed by a $1 billion breakup fee from Adobe. It points to a structural tension: investors may cheer the growth, but the upcoming cost of AI could undercut the very metrics justifying Figma’s lofty valuation aspirations.

A Bull Case on Shaky Ground

On paper, Figma checks the classic SaaS boxes (despite leaving the valuation box in its S1 filing unchecked). Browser-native collaboration gave rise to powerful network effects: product-led expansion, land-and-expand motions, and best-in-class retention.

Seventy-eight percent of Fortune 2000 companies use Figma, and users spending over $10,000 ARR jumped 39 percent year-over-year. Free cash flow in Q1 2025 nearly doubled to $94.6 million, driven by the tail-off of one-time stock-based compensation charges. At this pace, management argues Figma can crack $1 billion in revenue in calendar 2025 “without heroic assumptions.”

Figma has quietly woven a monopoly into the fabric of modern workflows. Even companies already entrenched in Microsoft 365 or Google Workspace increasingly turn to Figma—for design-led collaboration those suites can’t replicate. That cross-disciplinary value is Figma’s secret sauce: not through outright dominance, but by becoming indispensable across teams. Yet for all its traction, the valuation rests on shaky assumptions. The financials look strong—for now—but sustaining this momentum while absorbing the cost of AI may test the limits of Figma’s moat.

Even if Figma’s financial engine hums flawlessly, investors will demand proof that AI can deliver incremental revenue rather than merely inflate costs. And keep in mind: except for a handful of giants like Microsoft, OpenAI, and Nvidia, everyone else is being bludgeoned by AI costs. It’s what gives Figma a quasi-monopoly status. 

Figma Follows Apple in Being Slow to AI

Figma’s entry into generative AI has been markedly uneven—and slower than many expected.

1. Timing miscalculation
The hype around the failed $20 billion Adobe acquisition collided with the explosion of generative AI. But instead of doubling down, Figma paused—possibly underestimating how quickly rivals would weaponize natural-language prompts to leapfrog static design.

2. The “Make Designs” debacle
Launched in July 2024, “Make Designs” promised AI-generated first-draft UIs. But within hours, the tool drew backlash for reproducing near pixel-for-pixel copies of Apple’s apps. Figma swiftly pulled it and issued a retrospective, blaming its rigid “bespoke” design system for the low variability. CEO Dylan Field publicly accepted responsibility for the rushed QA process.

3. Silence and slow follow-through
For nearly a year afterward, Figma stayed quiet. No meaningful AI update emerged until Config ’25, where the company unveiled “Figma Make”—a prototype-to-site tool still in beta. AI received just 9 minutes of a 90-minute keynote, suggesting it was still an afterthought. Early users reported bugs and critiqued a hand wavy demo, casting doubt on enterprise readiness.

4. Talent flight and leadership distancing
Jordan Singer, a key architect of Figma’s AI strategy, exited in September 2024. A number of Diagram engineers followed. Meanwhile, an internal memo revealed that data training only began in August 2024, with public benchmarks lagging behind both dev-first platforms like Cursor and no-code rivals like Lovable.

Strategic takeaway:
Figma’s AI arc has moved from early ambition to cautious recalibration—long on brand, short on infrastructure. As competitors sprint ahead with prompt-to-product pipelines, Figma enters its IPO moment needing to prove it can scale AI with the same rigor and velocity that made it the go-to tool for interface design.

A Lovable Equation Bolted On

Lovable, launched in November 2024, is one of the breakout stars of the AI-native wave. Its premise is bold in its simplicity: describe what you want in natural language, and Lovable generates the entire stack—front end, back end, database, and deployment.

The early traction has been staggering. Within three months, it hit $17 million ARR, with 30,000 paying customers and 1.2 million apps generated. By May 2025, ARR climbed to $50 million, and by early July, reached $75 million, making it one of Europe’s fastest-growing startups. Lovable now claims 500,000 users, with 25,000 new apps built daily.

In June 2025, the company raised $150 million in a round led by Accel, at a valuation between $1.8–2 billion—a ~26× ARR multiple. For comparison, Figma’s IPO valuation aims for a similar ~27× multiple at a far larger $20 billion scale.

But what truly sets Lovable apart isn’t just growth—it’s accessibility. While tools like Cursor or Replit focus on developers, Lovable is built for non-technical users: founders, PMs, marketers. A prompt like “I want a two-sided marketplace with scheduling and Stripe payments” returns a deployable product in minutes. No handoffs. No interface design. No code.

This model doesn’t just compete with Figma—it bypasses its entire workflow. Where Figma optimized collaboration across design and engineering, Lovable collapses that workflow into a single prompt.

The traction suggests real product-market fit. Its $25/month price point is sticky and accessible, and its valuation—less than a year in—signals that Lovable is not just another startup, but a credible contender in the prompt-native software economy.

For Figma, this isn’t just competitive noise. It’s a deeper threat—a redefinition of how software gets made. Lovable doesn’t replicate design tools. It replaces the need for them. In doing so, it raises the stakes for Figma to accelerate its AI execution—or risk being left behind in a world where design is no longer a phase, but a prompt.

A Pointed Cursor: The Developer Frontier Closes In

While tools like Lovable and Bolt target non-coders, Figma’s most formidable competition is emerging from AI-native developer platforms: Cursor (Anysphere), Replit, Windsurf, and emerging interfaces like OpenAI Code Interpreter and Apple’s “vibe coding” alliance with Anthropic.

These platforms don’t improve the design-to-code handoff. They eliminate it.

Why This Wave Threatens Figma

1. Design abstraction is vanishing: These tools erase the traditional pipeline between design and engineering. What once started in Figma now begins—and ends—in Cursor or Replit. Figma's collaboration moat weakens as "handoffs" vanish.

2. Speed favors orchestration: Cursor and Replit are shipping faster than Figma’s AI team, whose headline tools remain in beta. Cursor alone commands a 45x ARR multiple and is highly capable of faster AI delivery.

3. Barriers to entry are collapsing: Initially pro-grade, these platforms are rapidly simplifying their UX—threatening to absorb Figma’s early-stage use cases.

These developer-first platforms aren’t trying to assist with design. They’re making parts of it obsolete. Building products has always been about outcomes and impact—not pixel perfection. Today, anyone can use Tailwind, Radix, or shadcn components and spin up beautiful, functional interfaces directly in Cursor. It may not satisfy the bespoke aesthetic that traditional pixel-pushing designers obsess over, but it delivers on velocity, clarity, and deployment.

What’s collapsing isn’t just the design-to-dev handoff—it’s the entire software development lifecycle. For decades, teams followed a linear path: design in Figma, spec in Jira, build in VS Code. But now, platforms like Cursor and Lovable let PMs, founders, or engineers simply describe what they want, and receive production-ready output. Natural language has become the new UX. The shift isn’t cosmetic—it redefines roles, dissolves stages, and compresses entire functions into orchestrated prompts.

If software creation becomes a prompt-first exercise, Figma must evolve from being a collaboration canvas to an orchestration engine. Anything less, and it risks being sidelined—not by a better design tool, but by a world that no longer needs one.

The Canva Problem: Scale Without the Designers

While Figma redefined the design workflow, Canva conquered everything else. What began as a simple drag-and-drop tool is now a sprawling creative suite for marketers, HR teams, educators, small businesses, and nonprofits. And it’s operating at a scale that Figma hasn’t come close to matching.

Canva now serves 220 million monthly users, up from 200 million just months ago, and counts 21 million paying subscribers. Over 35 billion designs have been created on the platform—1 billion in a single month alone, which translates to more than 400 per second. It’s entrenched across 90% of the Fortune 500, and nearly ubiquitous in the education and SMB sectors.

That reach is translating into real financial momentum. Canva’s ARR crossed $3 billion in 2025, up from $2.4 billion the year before, and it now commands a $40 billion valuation. Its latest product rollout, Visual Suite 2.0, introduced AI-driven tools for document automation, spreadsheets, and microapps. With features like Canva Code, users can now generate working microapps from simple text prompts—encroaching directly on the roadmap of Figma’s “Make.”

This is more than just feature overlap. Canva’s growth presents three major threats to Figma. First, Canva has already captured the non-designer market that Figma is just beginning to court. Second, Canva is embedding generative AI across formats with far greater speed. And third, Canva’s performance sets a harsh benchmark—10× the user base, 3–4× the revenue, and a valuation that makes Figma’s $20B IPO target look less like a premium and more like a pressure test.

Figma may remain beloved by professional designers, but Canva has become the default for everyone else. As Figma moves beyond UI design and into productivity tooling, it runs headfirst into a company that’s already there—with global distribution, mass-market appeal, and a head start in AI-native product expansion. Canva isn’t Figma’s next move. It’s the ceiling Figma now has to break.

The Business Model Strains: AI Is Expensive—And Figma Knows It

Figma’s uphill battle with AI isn’t just about execution gaps or product misfires—it’s about economics. The company’s own S-1 makes that plain. Building and operating AI-native tools for design—especially ones that span vector graphics, layout engines, and code—is deeply capital-intensive, and the financial strain is starting to show.

“In the short term, we expect that our AI investments…will negatively impact our gross margins and operating margins…[and] the extent of such impacts…are currently unknown.”

That’s not cautious hedging. That’s a warning.

“Our cost of revenue increased by $2.6 million…primarily driven by higher technical infrastructure and hosting costs related to AI…”

That figure, just for Q1 2025, offers a snapshot of the compute burden—one that's growing with every new feature added.

“We have entered into a long-term agreement with AWS for cloud infrastructure with a commitment of $545 million over the next five years.”

That’s nearly $300,000 per day in baseline cloud spend—regardless of revenue. And as Figma expands into real-time inference, multimodal tooling, and collaborative sessions requiring low-latency AI, those costs will escalate.

These pressures come at a time when investors are seeking AI-driven margin expansion—not erosion. In a landscape where:

• Stable Diffusion cost $600K to train,

• Midjourney burns millions in monthly inference,

• Google Veo 3 can cost $30+ to generate a few seconds of video, and

• Frontier models like GPT-4 reportedly cost $30–40M to build,

Figma’s ambitions will require not just great models, but ruthless orchestration.

Its entire business is built on the promise of 88–91% gross margins. But design-focused, latency-sensitive AI isn't cheap to train or serve. Unless Figma finds a way to tightly control cost or directly monetize its AI features, its own infrastructure could begin to eat away at the very metrics investors count on to justify even a $12 billion valuation.

A Contrarian Reckoning

Figma’s IPO arrives in a climate of caution—an IPO drought where public markets crave “AI at scale” stories but remain skeptical of margin erosion. The S-1 sings of growth and enterprise entrenchment, yet whispers of AI missteps, talent flight, aggressive multiples, and escalating infrastructure costs. With Lovable collapsing design pipelines, Cursor vaporizing handoffs, and Canva enveloping non-designer workflows, Figma stands at a crossroads:

1. Pivot or perish: Evolve from a canvas-first tool to a prompt-native orchestration engine. Deeply integrate AI—not bolt it on.

2. Monetize or marginalize: Build AI features that generate net-new revenue, not just user engagement.

3. Optimize or hemorrhage: Ruthlessly manage infrastructure spend—whether through orchestration, vendor diversification, or custom inference layers.

Absent decisive action, Figma risks being unbundled from the very software lifecycle it helped define. The future it previewed at Config—where designers could conjure products with AI—remains tantalizingly out of reach. Its rivals are no longer just design tools. They are systems of production. If Figma can't evolve from canvas to conductor, from interface to infrastructure, it won’t just miss its moment.

It will have seen the future—and still been overtaken by it.

And that may leave Figma’s future… well, a sour fig.