Figma and Canva are navigating the difficult transition to AI-driven products, highlighting the substantial costs and monetization challenges that accompany the technology's promise.
In a special report focused on the 2026 second-quarter earnings season for US stocks, software companies are racing to showcase progress with their AI offerings. They are trying to ease investor fears that the very AI applications they are developing might be undermining their core business models. Two industry reports this week reveal that creating a popular AI product is only half the battle.
Canva, a design software platform for students and small business owners, warned investors this week that its full-year revenue growth would slow to 20%. The reason: the company has had to slow the rollout of AI features that were expected to drive paid subscription growth. Demand has outpaced expectations, and the cost of delivering these services has risen sharply.
The following Thursday, Figma released its earnings outlook, forecasting revenue growth of 36% in the quarter ending September, down from 48% in the June quarter. The publicly traded design software firm said it is still figuring out how to monetize new AI tools that remain in testing and trial phases.
Both companies are acutely aware that deploying AI comes with high costs. Canva's Chief Operating Officer, Cliff Obrecht, told colleagues that before AI, the cost of serving millions of free users was "extremely low." With the introduction of AI, "costs have risen significantly, and the unit economics have completely changed, making it a top priority to reduce AI-related expenses."
Figma, which primarily sells subscriptions to enterprise clients, acknowledged that these costs could eat into its gross margins. Chief Financial Officer Praveer Melwani said on an investor call, "At this stage, we are not charging customers separately for the beta AI products. The model inference costs are entirely borne by the company, with no corresponding usage revenue to offset the expenses."
The common solution both companies are exploring is developing their own large language models to reduce costs. Canva is heavily promoting its proprietary AI model, claiming it generates images and videos faster and at a much lower cost than the leading models from major labs. However, when its AI features first launched, the proprietary model was not yet ready.
Figma management stated that the company is also increasing its investment in self-developed AI models, which are now being used in combination with cutting-edge commercial models to power its new Figma AI agent. Yet, training large models is both time-consuming and expensive, and investors are showing little patience. Following the announcement, Figma's shares fell by approximately 15% on Thursday.
Research: Only a Few Field Deployment Engineers Can Generate Returns from AI Projects
A recent survey from executive search firm Christian & Timbers reveals that out of approximately 17,000 technology professionals in the US with the title of Field Deployment Engineer (FDE), only a small fraction are helping businesses and government agencies achieve a return on investment from AI agents. FDEs, who combine the roles of software development engineer and business consultant, have become a hot job category in 2026.
The model was pioneered by Palantir (PLTR). This year, major firms like Meta, Google, Amazon, Salesforce, and even new companies spun off from Anthropic and OpenAI have been aggressively hiring FDEs to help clients deploy AI products. However, the Christian & Timbers study found that, to date, only about 12% of FDEs have used customized AI agents to help multiple companies achieve tens of millions of dollars in cost savings or revenue growth.
The research team interviewed over 250 executive hiring managers and 300 FDEs from 180 organizations (mostly large enterprises), while also integrating data from LinkedIn and Indeed. This data reinforces the finding that many companies are still in the exploratory phase and have not yet found a clear path to monetizing their AI investments. Whether a company can build a profitable AI cycle will directly determine if it continues to purchase services from AI tech firms on a large scale, and the AI companies themselves urgently need to prove that their technology investments yield returns.
Christian & Timbers CEO Jeff Christian stated that the ability to recruit top-tier FDEs will determine the success of a company's AI product. "The core issue is how to use people to implement projects and achieve process automation... Right now, deployment is the key," he said. Christian revealed that approximately 80% of the industry's recognized top FDE talent is employed by Palantir, with the rest distributed among Anthropic, OpenAI's deployment subsidiary, Google, Amazon, and Microsoft. These senior FDEs typically specialize in a single field, such as semiconductors or financial services. The vast majority of other FDEs are early-career engineers, with annual salaries ranging from $175,000 to $300,000.
Palantir CEO Alexander Karp has criticized the industry's rush to copy his company's FDE model. At a client event in late July, Karp said, "Younger friends might not know that the concept of the field deployment engineer was once mocked by the entire Silicon Valley. Now, all the companies that were criticizing it have started copying it." Data from Christian & Timbers shows a sharp surge in demand for FDEs. At the start of the year, only 10% of surveyed companies planned to hire for this role; by the end of the June quarter, roughly 70% of companies had either completed or initiated hiring plans.
Comments