By Lina Irawan 
The design-software company recently cut its expected revenue growth rate by a substantial third, bringing it down to 20% from a previously projected 30%. This rare downgrade was a direct consequence of the unexpectedly high operational cost associated with delivering its burgeoning suite of AI features. The financial implications prompted a strategic deceleration in the rollout of these advanced capabilities. Canva CEO and co-founder Melanie Perkins elaborated on this critical juncture, telling Fortune that user demand for the new AI features "significantly exceeded" the company’s initial expectations. While this overwhelming user engagement unequivocally validated the market’s appetite for AI-powered design tools, it simultaneously exposed a critical economic vulnerability.
"This validated the demand, but also showed us we needed to reduce the cost of completing an AI task to support a broad rollout," Perkins explained in an email. Her statement underscores a pivotal moment for Canva: recognizing the immense potential of AI while simultaneously confronting the unpalatable unit economics of its immediate implementation. Rather than pushing forward with a broad, potentially unprofitable, product launch, the company opted for a more prudent approach. "Rather than broadly rolling out a product before the underlying economics were ready, we decided to slow the rollout while we rebuilt the architecture, reduced unit costs and strengthened the business model." This decision, though impacting short-term growth projections, reflects a commitment to long-term profitability and sustainable innovation, a hallmark of Canva’s operational philosophy. It signals a strategic pause to re-engineer its backend, optimize its AI models, and ensure that the cost of serving each AI-generated request aligns with its robust business model, thereby protecting its enviable margins.
The gravity of this cost problem is amplified by its timing, landing at a pivotal moment in Canva’s strategic evolution. AI is not merely an add-on feature for Canva; it is central to its ambitious effort to transition from a leading design tool into a broader, more comprehensive workplace-software platform. Perkins had previously articulated her concerns about the fragmented nature of the AI market, suggesting a cautious but determined approach to integration. Since then, Canva has aggressively expanded its AI-powered offerings, introducing sophisticated tools such as Canva Code and significantly upgrading its core AI capabilities with Canva AI 2.0. These innovations are designed to push the platform beyond its traditional design strongholds, enabling it to cater to a wider array of enterprise workflows, from document creation to data visualization and collaborative coding. This strategic pivot means that the cost efficiencies of AI are not just about enhancing existing features but are foundational to unlocking entirely new revenue streams and market segments. The success of this ambitious expansion hinges directly on its ability to deliver AI features at a scalable and profitable cost point.
This dilemma faced by Canva is not an isolated incident; it illustrates a broader, increasingly prevalent challenge spreading across the entire software industry. Companies find themselves in an unenviable position: they simply cannot afford to sit out the generative AI boom, given its transformative potential and rapid user adoption. Yet, fully embracing it, particularly in its current nascent and often resource-intensive form, carries the significant risk of undermining the very lucrative economics of the businesses they have meticulously built and are striving to protect. The traditional SaaS model, characterized by high gross margins and near-zero marginal costs for delivering additional software instances, is being fundamentally reshaped by the computational demands of AI.
Derek Hernandez, Pitchbook’s senior research analyst covering the critical intersection of SaaS and AI, succinctly articulated this paradigm shift. "AI is making SaaS no longer a zero marginal cost solution, which has really been what I would call a lot of software’s secret sauce up until now," he told Fortune. For decades, the beauty of SaaS lay in its ability to replicate software at virtually no additional cost once the initial development was complete. Each new subscriber meant higher revenue with minimal incremental expense. However, generative AI introduces a variable cost component tied to usage – every AI query, every generated image, every text output consumes computational resources, incurring a tangible cost. "People want a much more capable product and solution, which through today’s technology means cost of usage is becoming a really global challenge for all of these companies." This shift means that the scalability model that defined SaaS success is now under intense pressure, forcing companies to re-evaluate their pricing strategies, infrastructure investments, and overall unit economics.
Despite the initial shock, Canva has been aggressively tackling the cost challenge. Perkins revealed in her email that the company has achieved a remarkable reduction in the cost per AI task, slashing it by nearly 90% since the launch of Canva AI 2.0 in April. This agentic upgrade to the Canva platform introduced more sophisticated AI capabilities, driving user engagement to unprecedented levels. However, even with this significant cost optimization, the sheer volume of user activity presents an ongoing challenge: Canva AI users are creating three times as many designs as in the previous version of Canva AI. This exponential increase in usage, while a testament to the product’s value, means that the company must continue to relentlessly focus on improving its underlying economics to support such robust demand without sacrificing profitability. The scale of usage can quickly overwhelm even substantial per-task cost reductions if the foundational architecture isn’t optimized for hyper-efficiency.
Canva’s public-market parallel, Figma, a collaborative design platform, has also openly disclosed its own version of AI trade-offs, providing further evidence of this industry-wide trend. Figma reported a notable decline in its free-cash-flow margin, which fell from a robust 27% in the first quarter to a more modest 14% in the second quarter. Furthermore, the company’s forecast for third-quarter revenue growth stood at 36%, a clear deceleration from its 48% growth rate in the June quarter. While both companies are giants in the design software space, their business models differ in nuances. Yet, the impact of AI on their financial performance – manifested as slower growth for Canva and margin compression for Figma – points to a universal pressure point.
Hernandez emphasized that Canva and Figma represent the "biggest signals" that AI is fundamentally breaking SaaS’s traditional model. The rising "inference expenses"—the recurring cost incurred each time an AI model processes a request or generates an output—are now undeniably showing up on financial statements. He clarified the concept using a vivid analogy: "If you have a basic analogy of a car, everything it takes to build a Ford F150 would be training, and then gas, mechanic costs, and anything else would be inference, because that’s the point of using the product." The initial investment in "training" an AI model, akin to manufacturing a car, is a one-time, albeit substantial, cost. However, the "inference" costs are the ongoing operational expenses—the "gas" and "maintenance" of using the AI model repeatedly. These recurring costs accumulate rapidly, especially with high user demand, directly impacting profitability. Hernandez noted the synchronous nature of these revelations, stating, "Canva and Figma both hit the same wall about five days apart, but they cited it in different places," underscoring the pervasive and immediate nature of this AI-driven economic challenge.
The AI cost reset carries particular weight for Canva as it evaluates a potential Initial Public Offering (IPO). Fortune reported last year that an employee share sale had valued Canva at a staggering $42 billion, with experts anticipating a public listing as early as 2026. However, Hernandez now suggests that Canva might be targeting a slightly later timeline, possibly next year. The decision to "basically tap the brakes" on the broad AI rollout is a strategic move, undoubtedly made with an eye toward public investors. In the current market climate, investors are increasingly scrutinizing profitability and sustainable growth, moving away from the "growth at any cost" mentality that characterized earlier tech booms.
"I’m sure they’re trying to protect their profitability, especially if they want to go to public investors," Hernandez affirmed. A company poised for an IPO needs to present a clear path to profitability and demonstrate predictable, scalable unit economics. Uncontrolled AI costs could significantly erode margins, making the company less attractive to discerning public market investors who prioritize financial health and stability. By taking the time now to re-architect and optimize its AI infrastructure, Canva is essentially de-risking its future public offering, ensuring that its powerful AI features can be delivered not just effectively, but also profitably, aligning with the expectations of sophisticated investors. This careful calibration of innovation with financial prudence will be crucial for Canva to maintain its premium valuation and successfully transition to the public markets, solidifying its position as a long-term leader in the evolving software landscape.
Looking ahead, the challenges faced by Canva and Figma signal a broader transformation across the entire software ecosystem. Venture capitalists and private equity firms are already adjusting their valuation models for AI-centric startups, placing a greater emphasis on demonstrable unit economics and profitability pathways rather than just user growth. Companies across various sectors are exploring diverse strategies to mitigate these costs, including developing more efficient, smaller AI models, investing in specialized hardware for inference (like custom AI chips), or adopting hybrid cloud strategies to optimize compute expenses. Some may opt to pass on a portion of these costs to end-users through tiered pricing models or premium feature subscriptions, while others might focus AI development on features that deliver such immense value that higher costs are justified. The era of "free" AI, at least from an operational perspective for SaaS providers, is rapidly drawing to a close. The companies that successfully navigate this shift – by innovating on both the technological and economic fronts – will be the ones that ultimately thrive in the AI-powered future, cementing their competitive advantage in an increasingly complex and computationally intensive market.
