Canva on Tuesday rolled out Canva Code 2.0, a significant upgrade to its AI-powered coding tool that lets users create interactive websites, apps, and experiences from plain-language prompts and then edit the results visually. The update is available to all of Canva's more than 265 million monthly users across every pricing tier, including free accounts.
Strategy: focusing on design and editability rather than raw code
Canva is taking a different route from many vibe-coding rivals by betting that the main obstacle is not producing working code but making that output look and feel like the user's own. Competitors such as Lovable, Replit, and Bolt.new have focused primarily on generating functional code; Canva emphasizes editability and visual polish.
Danny Wu, Canva's Head of AI Products, told VentureBeat ahead of the launch that the product is deliberately aimed at non-technical users. "Canva Code isn't a tool we're building for developers," he said, describing the goal of bringing lightweight AI coding into the Canva platform while increasing interactivity, customization, and flexibility for websites and interactive presentations.
New features and performance improvements
Canva Code 2.0 introduces several capabilities intended to bridge the gap between code generation and publishing a polished interactive experience:
- Drag-and-drop editing of generated elements directly on the canvas.
- Ability to create Canva Code projects inside other design projects, embedding interactivity in whiteboards, presentations, and pages.
- Import of raw HTML files from other AI coding tools and conversion into editable Canva designs.
- Over 50 new templates tailored to interactive designs.
- Integration with the broader Canva editor so coded outputs behave like other design elements.
Canva reports material performance gains: average code generation time has been reduced by 75 percent, and the median time from initial prompt to a published site has fallen by 30 percent. The company also says integrating Canva Code into the main editor increased active Code users by 25 percent.
A standout difference is the editing flow: instead of re-prompting or editing raw code to change visuals, users can click into generated elements to change text, drag images from Canva's library of more than 120 million templates and assets, update colors and fonts via a familiar toolbar, or refine a specific element using conversational AI. Outputs are fully interactive and automatically adapt to different screen sizes, with a built-in mobile preview. One current limitation Wu noted is that moving elements around still requires re-prompting.
Market context and competition
The vibe-coding and AI app builder market has grown rapidly. Luminix AI's May 2026 market research estimated the segment at about $4.7 billion in 2026, with a projection to reach $12.3 billion by 2027 at roughly 38 percent compound annual growth. The research also estimated that AI-generated code accounts for roughly 41 percent of all code written globally.
Major players in the space include Lovable, which reportedly hit about $400 million in annual recurring revenue by early 2026; Replit, which has tripled its valuation to $9 billion and targets $1 billion in run-rate revenue by the end of 2026; and Bolt.new, which scaled rapidly in annual recurring revenue after launch. Against this backdrop, Canva brings a different asset: a quarter-billion-user design ecosystem where brands and teams already store identities and collaborate.
Canva positions Canva Code not as a direct head-to-head with developer-focused platforms but as a complementary tool that fills a gap—allowing interactive elements to be part of broader visual communications like sales decks or classroom hubs.
HTML import as a strategic move
One strategically notable feature is the HTML import capability, which allows users to bring code generated by any AI tool—such as ChatGPT, Claude, Lovable, or Bolt—into Canva as a fully editable design. That positions Canva as a potential finishing layer where AI-generated code receives its visual polish regardless of the original generator.
Wu framed the capability as consistent with Canva's broader goal of making design easy and compatible: Canva has long supported importing PDFs and PowerPoint files and translating them into editable formats. He emphasized that the company aims to be "most pluggable," enabling users to incorporate external outputs into Canva workflows without forcing them to abandon other tools.
Use cases and limits
Wu was candid about the product's boundaries: Canva Code is well-suited for front-end apps, especially those that use data, accept submissions, and need interactivity at small to medium scale. He said it is probably not suitable for websites requiring complex backends or for services that must handle hundreds of thousands of daily visitors.
This distinction separates Canva Code from platforms pushing toward full-stack application development with databases, authentication, and production-grade hosting. For teachers, small business owners, and marketing teams—the bulk of Canva's user base—the ability to create professional-looking interactive pages and mini-apps without hiring developers is the primary value.
Models, acquisitions and AI infrastructure
Wu confirmed Canva uses a mix of proprietary and third-party AI models, including those from OpenAI and Anthropic, but declined to detail the exact mix, saying it varies by use case.
Canva's AI capabilities have been bolstered by acquisitions: in March 2024 the company acquired Affinity, and four months later bought Leonardo.ai, which had over 19 million registered users and more than a billion images generated. Canva says its AI products have been used more than 32 billion times to date.
Traction: six million published sites, retention unknown
Canva highlights that users have created and published more than six million websites using Canva Code since the feature's initial release a year ago. Wu clarified this covers sites published publicly, behind passwords, or via private links. He acknowledged the company does not yet provide clear retention metrics—how many of those sites remain live, receive traffic, or have been updated since publishing.
Early real-world uses suggest genuine utility: educators building classroom hubs, small businesses creating mini-apps for fundraising and interactive roadmaps, and groups producing educational games with pedagogical guides.
Pricing, data governance and enterprise concerns
Canva Code 2.0 is available across all Canva plans, including the free tier. Usage is credit-based: higher plans provide more AI credits and greater usage allowances. This approach differs from some competitors that reserve top features for paid subscribers.
On data governance, Wu said users retain full control over how their prompts and data are used: settings allow opting out of having prompts used for AI training. For businesses, team admins can set organization-level preferences to prevent inputs, content, and outputs from being used to train models.
Outlook: narrowing the gap between ideas and execution
Wu described Canva Code as part of Canva's mission to reduce the distance between imagination and what everyday users can build without technical overhead like deployments or APIs. Progress in AI models has made complex visual outputs more reliable than a year ago, increasing the chance that one-shot generative workflows produce acceptable results.
Whether Canva Code becomes a standalone product category or remains an integrated feature within Canva's broader AI toolbox will depend on how the company balances its front-end focus with the full-stack ambitions of competitors. Historically, Canva has succeeded by prioritizing accessibility and design outcomes over technical parity with niche professional tools. With Canva Code 2.0, the company is again betting that for most users the hardest part is not building, but making the result look like theirs.
