Canva is undergoing major restructuring, affecting 22% of its employees.
On November 3, Melanie Perkins, CEO of Canva, released an internal employee memo announcing a major restructuring of the company: 1,271 employees, accounting for about 22% of its global workforce, will be affected. Some positions will be adjusted or relocated to centralized office locations, and some employees will leave the company. Perkins stated that this is the "toughest moment" in Canva's 13-year history.
Why would a design software company with hundreds of millions of monthly active users, billions of dollars in annual recurring revenue, and profitable for consecutive years lay off more than one-fifth of its employees at this point? If you only read the memo, the answer seems to be organizational issues: too many hierarchies, overly fragmented professional divisions, too slow decision-making, and product teams being too far from the collaboration hub. But organizational flaws are only superficial. The deeper pressure comes from AI, which has disrupted the once-clear financial calculations of the SaaS industry.
An Unusual Round of Layoffs
Canva is not a company facing an existential crisis. Its user scale, cash reserves and profitability track record are all healthy in the SaaS industry. That is precisely why the 22% restructuring is even more of a signal: it is not that the company can no longer survive, but that the company has judged that its original organizational model and cost structure cannot support its next stage of development.
Perkins outlined three key changes. First, merge the product, growth and platform departments into a unified product organization to reduce cross-team handovers, and give teams clearer accountability for goals and greater autonomy. Second, reduce professional segmentation, streamline hierarchies, integrate overly narrow positions into broader roles, and avoid multiple teams being repetitively responsible for the same task. Third, centralize closely connected teams in the same office, making Sydney the primary workplace for more product teams, while continuing to invest in global teams. Starting from January next year, most hybrid employees are expected to come to the office 2 to 3 days per week.
These measures all point to the same goal: to shorten the path from product decision-making to engineering implementation. The competitive pace of the AI era does not allow a company to spend a lot of time on coordination, handovers and layer-by-layer approval. The "small, highly empowered teams" mentioned by Perkins are not a management slogan, but a necessary cost that the organization has to pay. Over the past 13 years, Canva has turned the dream in a business plan into a platform with hundreds of millions of users; however, the success of the first stage cannot automatically guarantee the success of the second stage. The larger the company, the easier it is for responsibilities to be fragmented, the more back-and-forth communication there is between product, growth and platform teams, and the less time is actually spent on product development. What the restructuring aims to solve is exactly this "scale-induced lag".
The Bill for Free Users
Roll the timeline back three months, and the outline of another set of accounts becomes clearer. In August 2026, Canva lowered its full-year revenue growth forecast from 30% to 20%. One of the reasons is that the inference cost of AI functions far exceeded previous estimates. Cliff Obrecht, co-founder of Canva, once publicly stated that before AI, the cost of serving a free user was extremely low; after AI was launched, this cost became "many, many cents".
Canva's long-prized freemium model was built on near-zero marginal cost: adding one more free user barely costs an extra cent, but that user could potentially convert to a paid subscriber in the future. AI has overturned this equation. Every time a free user calls for image generation, every intelligent template fill, generates a real computing power bill. Free users are no longer just assets waiting for monetization, but have also become a continuous cost bill sent to the server. For a company with the mission of "democratizing design", this transition is particularly tricky: free users are not only the community, but also the data source for product iteration, and the reservoir for paid conversion; but in the face of AI inference costs, the size of the community also means the size of costs.
The capital market will not wait for a company to slowly digest this bill. In August 2026, Canva's two major Australian investors Blackbird and Airtree lowered the valuation of their stakes by 17%; the valuation of Canva's internal employee shares dropped from 38.9 billion USD to 31 billion USD. T. Rowe Price later also wrote down its stake by 20%, implying an implied valuation of about 30.4 billion USD. There is even an implied pricing of about 30 billion USD in the secondary market, a nearly 30% discount from the peak. The widely expected IPO in the second half of 2026 has been pushed back to 2027.
Canva has not taken no action. Through self-developed models and the acquisition of Leonardo.AI, it has reduced AI service costs by about 90%, and the operating cost of its video model has dropped to one-seventeenth of that of similar cutting-edge models. However, the improvement of cost efficiency cannot replace the improvement of organizational efficiency. The 22% restructuring is the extension of this multiple-choice question on the human resource structure: instead of forcibly pushing growth when unit economics does not add up, it is better to take the initiative to slow down and rebuild the underlying architecture at the same time. Slowing down does not mean giving up growth, but re-establishing growth on calculable and sustainable unit economics.
The Entrance Is Shifting
Cost is only half of the story. The other half comes from the displacement of the competitive landscape. Canva once occupied a very comfortable position: it is much lighter than Adobe's Photoshop and Illustrator, and much more "editable" than pure AI generation tools, perfectly meeting the demand of hundreds of millions of non-professional creators who "want to make a usable image". It does not require users to learn professional software, nor does it require users to generate everything from scratch; templates, drag-and-drop, brand kits and collaboration processes form a low-threshold but high-stickiness workflow.
But the real threat does not come from Adobe. In April 2026, OpenAI launched GPT Image 2.0, advancing image generation to a stronger level of text processing, editing and layout control; almost at the same time, Anthropic launched Claude Design, directly targeting the "first draft" of prototypes, slides and single-page visual materials. The most valuable moment in design work is the moment when an idea just takes shape and the first version of the sketch is drawn. In the past, that moment belonged to people who opened Photoshop, Illustrator or Canva; now AI companies want that moment to happen in the chat box.
If the "generation" step is taken over by general-purpose AI systems, platforms like Canva run the risk of being reduced to "a place to clean up and distribute AI drafts" — still useful, but no longer defining the starting point of the workflow. Canva's response is two-pronged: on the one hand, it integrates the output of Claude Design into Canva as editable branded designs, trying to draw a division of labor between "generation" and "governance"; on the other hand, it expands its AI capabilities at a record speed, launching Canva AI 2.0 in April 2026, upgrading Magic Studio from a set of buttons to a conversational design partner, and successively acquiring marketing automation and AI workspace companies to extend to the integrated "creativity + marketing" scenarios.
The direction of these moves is correct: if it cannot exclusively own the "first draft", it will deepen its irreplaceability in editing, brand control, template system and distribution links. What enterprise customers need is not just a single image, but visual assets that meet brand specifications, can be collaboratively modified, can track versions, and can finally be distributed to multiple channels. Canva's opportunity lies in turning the AI-generated "draft" into a governable "finished product". But this strategy requires the organization to operate at a faster speed with less coordination loss. The "small, highly empowered teams" mentioned by Perkins are precisely the necessary condition for this strategy at the organizational level, not a rhetorical device. Merging product, growth and platform, flattening hierarchies, and concentrating closely collaborating teams in Sydney essentially shorten the path from product decision-making to engineering implementation, and reduce one of the most expensive costs in the AI era: organizational friction.
The Organization Must Run Faster
There is another easily overlooked dimension in Canva's restructuring: geography. The memo makes it clear that Sydney will become the primary workplace for more product teams, as well as the workplace for employees in other professional fields that require close collaboration and information sharing. For employees who are asked to consider relocation, the company says it will provide full support. Cross-border collaboration brings talents and benefits, but it also increases the difficulty of gathering the right people together to solve complex problems.
Under the triple pressures of slowing growth, high AI costs and shifting competitive entrances, Canva has chosen to put "face-to-face collaboration" back at the center of organizational design. This is not a simple return-to-office policy, but a reconfiguration of decision-making speed: when product, growth and platform are merged, when hierarchies are compressed, and when more key teams are concentrated in the same location, what the company hopes to reduce is not only communication costs, but also the lag caused by scattered responsibilities. For a company operating across 37 countries, centralized office will inevitably bring difficult personal choices; but for Canva, it obviously believes that the coordination cost brought by dispersion has become high enough that it must be addressed.
Perkins emphasized in the memo that the decision was made to first clarify the mission and goals, then design the organization, skills and office locations, and finally assess the impact on positions; this does not negate the talent and contribution of the departing employees. The wording is restrained, but the logic is clear: strategy comes before positions, and the organization serves the strategy. The restructuring is not a negation of the past 13 years, but an advance layout for the next decade. The problem is that the cost of this layout is borne by specific individuals.
Compensation and Relocation
For affected employees, the support provided by Canva includes: at least six months of salary and statutory benefits; waiving the one-year equity vesting restriction for employees who joined within the past year; continuing to provide 12 months of medical coverage in countries that provide medical services; 12 months of Canva Care employee assistance program covering family members; one-on-one coaching, interview preparation and one year of LinkedIn Premium; the talent recruitment team assisting with resume revision and interview practice; sharing the voluntarily joined alumni talent directory with investors, partners and industry networks; departing employees can keep their Canva MacBook; launching the Canva Collective Fund to provide 10,000 USD grants for employees who start their own businesses, with registration before December 14, and the company will not hold ownership or equity in the projects.
These compensations are not low among similar restructurings: six months of salary gives a buffer period, the cancellation of equity vesting restrictions avoids the worst-case scenario of "being laid off right after joining and all options being invalidated", and career support and the alumni directory try to reconnect departing employees to the industry network. Retaining MacBook and setting up the collective fund also show a certain degree of decency. But they cannot change a fact: the 22% workforce adjustment means that Canva is exchanging short-term pain for long-term organizational speed. Compensation can mitigate the impact, but cannot eliminate it; relocation support can reduce friction, but cannot make choices for employees.
This article is from the WeChat Official Account "Tech Business", authored by Tech Business, and authorized for release by 36Kr.