Figma CEO Dylan Field on the software reckoning
Watch on YouTube ↗  |  February 17, 2026 at 18:57 UTC  |  26:12  |  CNBC
Speakers
Dylan Field — CEO, Figma
Georgia — Host

Summary

  • Figma is integrating Anthropic's "Claude Code" directly into its canvas, betting that the future workflow involves fluid movement between code and design rather than linear handoffs.
  • There is a "software reckoning" occurring where public market investors fear AI agents will reduce "seats" (human employees), thereby crushing the traditional SaaS per-seat business model.
  • Dylan Field argues that while individual coding tasks are being automated, the volume of software being built will go exponential, requiring more design/oversight tools, not fewer.
  • A specific anecdote from the host highlights a shift in user preference: She abandoned Google's Gemini and OpenAI's ChatGPT in favor of Anthropic's Claude Code for coding tasks, citing superior context retention and "taste."
  • Traditional media (Disney, Paramount) is reacting defensively (Cease & Desist orders) against new AI video models (ByteDance), signaling the rapid erosion of IP moats in video creation.
Trade Ideas
Ticker Direction Speaker Thesis Time
AVOID Georgia "Bite Danc's Sea Dance 2.0 dropped about a week ago. I mean it's so good that Disney and Paramount have sent cease and desist." When legacy media giants resort to legal threats (C&Ds) rather than competition, it signals that the disruptive technology (AI video generation) has breached their quality moat. The ease of creating "Disney-quality" content via AI is a deflationary pressure on their IP value and production advantages. Avoid legacy media stocks relying on production barriers to entry; the cost of content creation is collapsing. Legal regulation could successfully stifle AI video distribution in the short term. 21:53
WATCH Georgia "I switched from chatbt and Gemini to Claude Code and I have not looked back. I have not opened ChachiBT in weeks. The model is just that much better..." In the high-value vertical of "AI Coding/Development," Anthropic (Claude) is winning market share from the incumbents (Google/Gemini and Microsoft/OpenAI). If developers—the power users of AI—are churning away from Gemini/ChatGPT, these platforms lose critical data feedback loops and subscription revenue in the developer segment. Watch for slowing growth in Gemini/ChatGPT developer adoption; Anthropic is emerging as the superior "coding brain." Google or OpenAI could release a model update (e.g., GPT-5) that leapfrogs Claude immediately.
AVOID Dylan Field
CEO, Figma
"We do not uh have uh any offerings around Chinese models right now... we need to really be auditing um every model we look at." Enterprise software platforms (like Figma) act as gatekeepers. Due to security, auditability, and trust concerns, Western enterprise platforms will exclude Chinese AI models from their integrations. This limits the commercial reach of Chinese AI in the Western B2B market, regardless of model quality. Avoid Chinese AI plays expecting Western enterprise adoption; the "Trust Moat" favors US/European models (Anthropic, OpenAI, etc.). Chinese models become so superior in performance that companies bypass safety protocols to use them.
LONG Dylan Field
CEO, Figma
"If you're trying to build software, there will be exponentially more software in the world... It just happens to be going asmtoic now and vertical." The market fear is that AI reduces the number of human developers (selling fewer SaaS seats). The counter-thesis is that AI lowers the cost of creation, leading to an explosion in the *quantity* of software built. This increases the total addressable market for tools that manage, design, and deploy this massive influx of software. Long essential infrastructure software that benefits from *volume* of applications, regardless of whether humans or agents are writing the code. The "Per-Seat" pricing model collapses faster than companies can pivot to "Usage-Based" pricing.