Two leading companies in the design software space are using real financial data to reveal the hidden costs of an AI transformation.
Canva warned investors this week that its annual revenue growth rate will slow to 20%, after the company voluntarily halted the planned rollout of a key AI feature designed to drive paid subscription growth. The reason: demand far exceeded expectations, but service costs also far exceeded expectations. Figma reported on Thursday that it expects third-quarter revenue growth to slow from 48% in the June quarter to 36%, and admitted that several of its AI tools remain in beta testing without a clear path to commercialization. Following the news, Figma's stock fell by approximately 15% in a single day.
The experiences of both companies point to the same structural issue: the popularity of AI products does not automatically translate into healthy unit economics. Under the narrative framework where investors have placed high hopes on AI application companies, this is a real-world stress test from the front lines.
Canva: Too much demand becomes a burden, prompting a voluntary slowdown
Canva's dilemma is somewhat dramatic—not because no one is using its product, but precisely because too many people are using it.
According to reports, Canva Chief Operating Officer Cliff Obrecht told colleagues that before introducing AI, the cost of serving a large number of free users was "very low." However, after AI features were launched, "those costs have risen substantially, and the unit economics have fundamentally changed, making it much more important than before to reduce AI costs."
Based on this assessment, Canva chose to voluntarily slow down the pace of its planned AI feature rollout, rather than continuing to trade user growth for high costs. This decision directly lowered the company's expectations for its full-year revenue growth rate, from previously higher market expectations to 20%.
To fundamentally address the cost issue, Canva is aggressively pushing forward with developing its own proprietary AI models. The company says its internal models are faster and cheaper for image and video generation than those from top-tier AI labs. However, the problem is that these models did not make it in time for the current AI feature rollout, forcing the company to rely on more expensive external models at a critical juncture.
Figma: Beta products offered for free, with inference costs fully absorbed
Unlike Canva's voluntary slowdown, Figma faces a different form of cost pressure: paying for products that have not yet been commercialized.
Figma Chief Financial Officer Praveer Melwani directly outlined this logic during an investor conference call:
"We are currently not charging customers for the use of products in beta testing. The inference costs are borne by us, with no consumption revenue to offset them."
This means Figma is subsidizing user access to its AI features with its own funds, while waiting for these tools to move from beta testing to formal commercialization. The company expects this cost structure to pressure its gross margins.
Figma is also increasing its investment in proprietary AI models and has begun combining internal models with frontier models to support its newly launched Figma AI agent. However, training proprietary models itself requires time and capital investment, and the capital market's patience is clearly limited—after the earnings guidance was released, Figma's stock fell by approximately 15% in a single day.
Shared dilemma: The cost structure of AI transformation is not yet properly aligned
Reports indicate that the cases of Canva and Figma reveal a common, stage-specific contradiction currently facing AI application companies: product-level appeal has been validated, but the path to converting it into a sustainable profit model remains unclear.
Both companies view proprietary models as the key to breaking the deadlock—by reducing reliance on expensive external models to fundamentally improve unit economics. However, there is an unavoidable time lag between the construction cycle of proprietary models and the market's immediate demand for performance growth.
Analysts believe these two earnings reports provide an important reference point: the cost of AI transformation is not only reflected in R&D investment but is also more deeply embedded in the inference expenses behind every user call. Until the commercialization model matures, this portion of costs will continue to erode gross profit margins.
Software companies' AI narratives are undergoing their first real round of financial examination.
Comments