DEV Community

Cover image for Even Figma isn't sure about its own design tokens
Stéphane LaFlèche
Stéphane LaFlèche

Posted on

Even Figma isn't sure about its own design tokens

#webdev #figma #frontend #ai

The whole industry seems to have agreed on a standard for design tokens. The shift it sets up is still on its way.

Design tokens are not new. The term was coined in 2014, at Salesforce, by Jina Anne and Jon Levine.1 By 2017, Amazon had open-sourced Style Dictionary and the idea had spread well past Salesforce. We have been shipping design tokens for over a decade.

What we never did, in all that time, was agree on a format. Every tool and every team rolled its own shape. There was never one neutral way to write a token down, its value and its meaning, so that any other tool could read it.

Have you heard of DTCG? I hadn't, until recently. It is the Design Tokens Community Group, a W3C effort to finally settle that format.2 The repo is quiet, but that is because the spec reached its first stable version in late 2025, not because anyone walked away. The quiet is a thing being finished, not abandoned.

The list of who is backing it is not quiet at all.

Adobe. Google. Microsoft. Meta. Amazon. Shopify. Salesforce. Sony. Pinterest. The New York Times. Disney. Framer. Penpot. Figma. Plus a dozen more.2

That is not a side project. That is most of the industry quietly agreeing on something. One of those names, Figma, is the reason for the title of this piece. We will get to it, because the irony is the whole point.

Here is my bet, and I will say up front that it is a bet. I think a storm is coming for design tooling. You do not have to believe me about the storm, because the bet does not depend on it. If you are wiring your tokens straight into one vendor's format, you are exposed. Anchor them to the open standard instead and you are not. The downside is lopsided. If I am wrong, you have lost almost nothing. If I am even half right, everyone hard-coded to a single tool is facing a rewrite.

The format is young and already fragmenting. That is the point.

The obvious objection is that the standard is too new to bet on, and already splintering. It is splintering. Google's DESIGN.md takes DTCG as inspiration and then forks the schema into its own shape, prose rationale wrapped around DTCG-style tokens.3 Other tools keep their own dialects. This is normal. It is the browser wars again, the old web pattern where everyone agrees on a direction and argues about the details for a few years.

What matters is where they argue. The forks agree on the base: the primitive tokens, colours, spacing, radii, and the reference syntax. They diverge at the edge, on composite tokens like typography, shadows, and gradients, which are the youngest part of the spec and where tool support still trails.4 So the disagreement sits in one known place, and it is closing rather than spreading. Style Dictionary, Tokens Studio, and Terrazzo are all converging on the same base, and there is already a joint Style Dictionary and Terrazzo proposal for a shared format.5

That is the case for anchoring to the standard rather than to any one flavour. DTCG is vendor-neutral and governed in the open under the W3C; a company's fork is governed by that company. The forks are not replacing the standard; they ride on it. The closer analyses call DESIGN.md a wrapper around DTCG tokens, not a competitor.3 You build on the settled base and adapt to whatever flavour wins at the edge.

One honest caveat. That base is rock-solid for primitive tokens today, and more aspirational for composites. If your tokens are colours and spacing, you can lean on the standard now. If they are type ramps and elevation shadows, you are betting slightly ahead of the tooling. Worth knowing before you wire it up.

Which brings us to Figma

Figma is on that buy-in list. Figma also cannot tell you what its own variables mean.

Open a Figma variable and you get a bare number. A FLOAT of 10. Is that 10px, a z-index of 10, or 10ms? The file does not say. You guess from context. So does any tool reading it. So does any LLM you point at it. Getting those values out in clean DTCG form still means reaching for a community plugin today; Figma's native export is rolling out, but it does not cover the composite tokens yet.6 The most valuable design tool in the world has not settled its own token format. That is the irony the title promised.

It gets sharper. Figma went public in July 2025 as one of the year's hottest IPOs and popped to around $142. A year later it trades near $17, more than 85% below that peak.7 The cause is contested. Frothy valuation, widening losses, AI pressure, take your pick; I will not pretend to know it. The point is narrower. A year ago, betting your token pipeline on Figma's permanence looked safe. It looks different now.

Here is what makes the bet easy, though: you do not need Figma to fall. Figma is on the DTCG list, and it is already rolling out native DTCG export.6 The rollout is incremental, but the direction is set. Anchor to the standard and you win whether Figma thrives or fades. That is the whole shape of the bet.

It is not just Figma

If Figma feels like a special case, look at Adobe. Adobe is the giant that tried to buy Figma for twenty billion dollars, got blocked by European and UK regulators, and paid a record one-billion-dollar fee to walk away.8 Adobe is also on the DTCG list. In 2026, Adobe is making more money than ever and watching its stock fall about 30% anyway.9

That is the cleaner version of the whole argument. Revenue is up and the AI products are growing, yet the market is marking the company down by a third on a single fear: nobody knows what AI will do to creative software. The fundamentals are fine. The uncertainty is what is being priced. When the most entrenched tool company in the industry gets repriced on "we cannot see what is coming," that uncertainty is not a Figma problem. It is the weather.

The standard does not make the hard part easy

None of this makes anchoring to DTCG free. Getting clean tokens out of a messy tool is the FLOAT 10 problem again: something has to decide whether that 10 is a length or a duration, and the standard will not decide it for you. There is no free lunch, which is the same point I made last time: the tooling gets you a real step closer; the judgment stays a human's to own. What the standard gives you is a stable target to aim that judgment at, one format instead of a converter per tool.

You might object that the glue is throwaway: with Figma's native export still incomplete, you build a Figma-to-DTCG adapter now and bin it the day Figma ships full support. True, but that adapter is the only disposable piece. Everything downstream stays aimed at the standard, so when the adapter goes, nothing else moves. Couple to the tool instead and the same change rewrites the whole chain. The loose part today is exactly that Figma-to-token step, so make it the deliberately disposable one. When native support lands, the rest of your app is already built on the standard, and you just start pulling clean tokens straight from Figma.

What anchoring actually looks like

Put DTCG in the middle, and build that layer to expect change. The format itself will keep shifting, and depending on the company you work for, your house standard might be a fork of it rather than DTCG proper. A middle layer built to absorb those variants stays resilient today, and it leaves you positioned to adapt the moment the industry consolidates on DTCG. Keep your source of truth there and generate everything downstream from it: your Figma variables, your CSS custom properties, your iOS and Android tokens. The design tool stops owning your tokens and becomes one more consumer of them. That same DTCG source is the input to the typed foundation I argued for in an earlier piece: the standard upstream, a typed layer downstream.

Better tokens are not the whole answer. They are one piece of that same larger problem: using AI to turn designs into systems, and who owns the decisions in the middle. A shared format settles where your tokens live. The system they feed is still yours to design.

The Design Tokens Community Group is doing quiet, unglamorous, genuinely important work, and most developers have never heard of it. If you build tooling in this space, it is worth your attention, and worth building toward. I would like to hear from anyone already betting this way, and from anyone who thinks the bet is wrong. That is the conversation I am trying to start.

  1. "Design token" was coined at Salesforce by Jina Anne (with Jon Levine) in 2014, on the Lightning Design System; broad adoption followed around 2017 with Amazon's Style Dictionary. See the design-token histories ("The incomplete history of design tokens", Design Systems Collective; Smashing Magazine's interview with Jina Anne). ↩

  2. The Design Tokens Community Group is a W3C community group; its Design Tokens Format Module reached first-stable (v2025.10) in October 2025, with 20+ editor organizations and 24+ participating companies (Adobe, Google, Microsoft, Meta, Amazon, Shopify, Salesforce, Figma, Framer, Penpot, and more). See w3.org/community/design-tokens and designtokens.org. Verify the participant list and date at publish. ↩

  3. Google open-sourced DESIGN.md in April 2026; it bases its tokens on DTCG but diverges ("inspired by, not strict compliance"). Independent analysis frames DESIGN.md as a wrapper that contains DTCG tokens plus prose rationale, complementing the standard rather than replacing it. ↩

  4. DTCG 2025.10 defines composite token types (typography, gradients, shadows), but real-world tool support for composites lags the primitives, and composite-token export is an active area of discussion. The bet is strongest for primitive tokens today and more aspirational for composites. Verify which composite types are normative-stable at publish (Design Tokens Format Module 2025.10; Figma forum "DTCG Composite Token Export Support"). ↩

  5. Style Dictionary v4 has first-class DTCG support; Tokens Studio and Terrazzo are reference implementations, and there is a joint Style Dictionary and Terrazzo RFC on a shared token-listing format (styledictionary.com/info/dtcg; GitHub style-dictionary discussion #1479). ↩

  6. Figma announced native DTCG variable export at Schema 2025. Confirmed rolling out as of June 2026: you can right-click a variable collection and Export to JSON in DTCG-aligned form, then run it through Style Dictionary. The rollout is still progressive and spec compliance is incomplete (the description field is dropped, and composite tokens still trail), so community plugins (Token Press and others) provide the fuller export today. Re-verify exact status at publish. ↩

  7. Figma (NYSE: FIG) IPO'd in July 2025 at $33, closed its first day near $115 and peaked around $142 (closing high ~$122 on 1 Aug 2025). By late June 2026 it traded near $17, more than 85% below that peak. Figures move daily; re-verify at publish (stockanalysis.com/stocks/fig; macrotrends.net/stocks/charts/FIG). ↩

  8. Adobe agreed to acquire Figma for ~$20B in September 2022; the deal was terminated in December 2023 after European Commission and UK CMA antitrust pushback, with Adobe paying a $1B reverse termination fee (CNBC; Axios; Adobe SEC 8-K). ↩

  9. Adobe (NASDAQ: ADBE) was down roughly 30% in 2026 (sources vary, about 23% to 37% by date and baseline, and ~46% from its all-time peak), at a multi-year low despite record revenue and a growing Firefly business, on fears that generative AI commoditizes creative software. One analysis called it "narrative risk being priced as terminal risk, which the fundamentals do not yet confirm." Verify the current figure at publish. ↩

Top comments (0)