Design is dead, they say. Hardly, I say.

At its roots, design was never about a tool or a prettifying pit stop in the implementation process. It has always been an intelligence layer that permeates the business, aligning, shaping, and multiplying it. Over the decades, the companies that recognized this and developed deep partnerships with design have seen unparalleled momentum and success that their competitors couldn't match. I don't see that changing, even as the process, the time to ship, and the tools all shift underneath us. I've spent 17 years running that thesis inside healthcare, fintech, and government software, and I currently run it as a design and research function serving 150+ hospitals.

It is a strange sort of death, then, that keeps drawing record checks toward design.

The loudest of those checks came in 2025. The same year as the “design is dead” headlines, OpenAI made the largest acquisition in its history, roughly $6.5 billion, all in equity, to bring Jony Ive and his studio in to lead its hardware design. Sam Altman called him “the greatest designer in the world” and framed the work as living at the intersection of technology, design, and an understanding of people. The most consequential AI company on the planet, holding the very technology that was supposed to make designers redundant, spent its biggest check ever to put a designer at the center of what it builds next. Markets vote with capital. That was the vote, and it was not close.

The claim does have a real source. In March 2025, Fast Company sat in on a closed-door meeting of design executives convened to ask whether design was finished. The numbers they traded were grim. In a survey of hundreds of large companies, 39% had cut the top one or two levels of their design organization. In 84%, design now reported into another function, often marketing, and often not even to that function's head. Designers were laid off at more than twice the rate of their peers. Figma shipped tools that turn a prompt into working UI, and “vibe coding” went from novelty to normal.

Read that honestly and you don't see design dying. You see design being demoted, and you see a craft shedding the part of itself that was always the most replaceable. Some of that demotion the field invited. When we let design become the gloss at the end of the line, we taught everyone above us that it was overhead. Overhead gets cut. The companies that filed design under marketing and now wonder why their product feels like everyone else's are not victims of a dead discipline. They are looking at the consequence of moving the intelligence layer out of the room where decisions get made.

The historical record is unambiguous on what design has actually been.

Raymond Loewy, the man usually called the father of industrial design, took a $50,000 bet in 1940 from the president of American Tobacco that he couldn't improve the Lucky Strike pack. He turned the background from green to white, which cut printing costs and read as cleaner to the women entering the market. He put the target logo on both sides so the brand stayed visible however the pack landed on a table. Every move was commercial, and sales rose. He did the same for more than 200 companies, from the Coca-Cola fountain dispenser to the Shell and Exxon logos to the livery on Air Force One. A national magazine wrote in 1950 that Loewy had “affected the daily life of more Americans than any other man of his time.” That is a designer functioning as business strategy, 80 years before anyone thought to call it a layer.

Dieter Rams gets remembered for what he made. I'm more interested in what made it possible. When the Braun brothers took over the company in 1951, they made a deliberate bet, drawn from their own market analysis, that design would be the thing Braun competed on. Rams ran that bet. His design function reached into engineering and marketing, and it shaped what Braun chose to make, not only how those products looked. When his team built the SK4 in 1956, they threw out the bulky “music furniture” every rival was selling and defined a cleaner category Braun could own. The discipline he summed up as “Less, but better” became a brand, and then a philosophy that outlived every quarter it shipped in. Jony Ive built much of Apple's design language on that foundation and never pretended otherwise, which carries the line straight to the company most people reach for when they argue design matters.

At Apple, design is not the thing you consult near the end. It sits close to the center of what gets made. Airbnb made the same arrangement explicit. Brian Chesky holds a degree in industrial design and, by Figma CEO Dylan Field's count, is the only designer-CEO on the Fortune 500. At Figma's 2023 conference, in a session titled “Leading Through Uncertainty: A Design-Led Company,” he described folding the classic product management function into a combined product and marketing model borrowed from Apple, with designers as equals in the room where direction gets set. A year later, Paul Graham's “founder mode” essay generalized the same instinct: keep the people with the clearest picture of the product close to the decisions instead of relaying them down through layers. At that altitude, design sets the direction the business commits to.

A pit stop is overhead. An intelligence layer changes what the business can see and decide.

Now to the part the headlines have to explain.

The harder version of “design is dead” is that AI will write it out of the loop. The opposite is happening, for a reason specific to AI. When the model is one anyone can rent, the build stops being the differentiator and the experience becomes the whole contest. So watch where the companies moving fastest are actually spending.

Ramp shipped about 270 new features in the first half of 2025 and reports a Net Promoter Score on par with Apple's. Its product leaders treat velocity as a design decision and say it plainly: you ship your org structure. When Lightspeed led one of Ramp's 2025 rounds, the thesis it published was that Ramp is building “the intelligence layer for finance.” There it is again, in their domain, in the words of the people writing the check.

Notion got there through craft. Ivan Zhao, a designer, runs it, rebuilt the product four times before it found traction, and kept the team small so taste stayed close to the work. He treats Christopher Alexander's 1964 Notes on the Synthesis of Form as the operating manual: design is mapping the real problem and finding the best fit inside the materials and constraints you actually have. 100 million users later, the supposed tradeoff between craft and scale never showed up.

Then there is Anthropic, whose entire brand campaign is two words: keep thinking. Its head of design, Joel Lewenstein, one of the most-cited voices in AI design, is doubling his team. Not cutting it. Doubling it. And he is doing it at the one company that could most plausibly claim it no longer needs designers at all. The work he describes is not screen decoration. It is designing Claude's character so the thing pushes back like a sparring partner instead of flattering you, because a model that earns trust is a design problem before it is a model problem. One of the most valuable companies on earth has decided the difference between a useful AI and an unsettling one lives in the design layer, and is staffing accordingly.

The talent market votes the same way the capital does. In July 2026, Cursor hired Jenny Wen, the design lead behind Claude and its Cowork agent, as its head of design, weeks after the announced $60 billion acquisition of Cursor's parent company. The logic in the coverage was plain: as the advantage from raw model performance narrows, the moat moves to the product experience, so the company bought design leadership the way it would buy any other strategic asset. A discipline in decline does not get poached at the top of the market.

Abridge works in my field, clinical AI, and the run it has put together is almost unfair. A $2.75 billion valuation in February 2025. $5.3 billion 4 months later. Another raise in April 2026. Deployed across more than 150 health systems, from Kaiser to the Mayo Clinic to Memorial Sloan Kettering, and rated the category leader by KLAS two years running. Clinical AI is one of the most constrained design problems there is, hemmed in by regulation and by the handful of seconds a clinician has to trust the thing in front of her. Abridge treated those constraints as the material and built within them, which is how it became one of the best-designed clinical AI products the field has yet produced. Here is the part worth sitting with. The transcription model underneath ambient clinical documentation is not the moat anymore. Half the market has one. What separates the leader is the design of how the note earns a clinician's trust inside a real workflow and then disappears into the background so a doctor can stay in the room with a patient. Abridge's own clinician-satisfaction numbers are the tell. Those are design outcomes wearing a healthcare badge. The capital chasing this company is chasing the experience layer, because that is the part competitors can't copy by training a bigger model.

So here is the question the obituaries keep dodging. If design is dead, why is the smartest, most impatient capital in the world spending record sums to buy more of it? Dead things don't get a Series E. You don't double a team you are eulogizing. Capital is loud, and it is not always right. What it does not do is keep funding a discipline it has decided is finished.

There is a sharper objection than the money, and it deserves a straight answer. The Nielsen Norman Group, the field's most-cited authority, now says the interface is becoming less of a differentiator, that as models do more of the rendering, polishing a screen buys you less than it used to. I think they're largely right. But look at the conclusion they draw from it: design deeper. Move into trust and into judgment about what to build at all, because that is where the differentiation went. The claim only threatens design if you had quietly defined design as the execution stage. If design is the polish at the end, the machine is already coming for it. If design is the intelligence layer that decides what gets built and why, the same research just cleared off the work design never should have been holding.

For the clearest case of design deciding what gets built, look at who is winning the categories everyone had written off as settled. Issue tracking was one of them. Jira had held most of the market since 2002, and engineers used it because they had to. Then Linear arrived, run by Karri Saarinen, a designer who ran design at Airbnb and Coinbase before founding it. He built the company on one conviction: that craft is the strategy itself. Linear is fast and unapologetically opinionated, every call about what it will and won't do made from a designer's read on how the work should feel. Five years in, it is valued at $1.25 billion, with OpenAI, Ramp, and Vercel among its customers and engineering teams leaving the incumbent on craft alone. Business banking was just as settled, and Mercury ran the same play there. Its first hire was a designer, on the bet that banking is a product design problem and not only a financial one. It now manages billions in customer deposits at a $5.2 billion valuation. In each case, design decided what the product would be.

Now hold all of that against the companies still running the opposite play, the ones where a design team is asked to build a business case for its own existence, to prove in a slide deck that taste moved a number. They are losing a race they don't know they're in, against competitors who just handed the same work a seat at the top table. Asking design to justify itself in the AI era is like asking the foundation to justify the house. The argument only looks foolish once the building starts to lean. None of this asks every company to agree. It only asks the market to keep rewarding the ones that already do.

The pattern across every company in this essay is easy to check against your own. In each one, design sat close enough to the center to shape the trajectory of the company, not just the output. So ask where design sat the last time the company placed a real bet, and when design last changed the roadmap, or stopped a bet from being made at all. The winners here would answer instantly.

I work in clinical AI and trust-critical systems, where this stops being a debate about org charts. Picture the caseworker on morning rounds with 30 seconds per patient to judge who is close to discharge and who is not. Or the pre-anesthesia care coordination nurse with a few minutes to clear an 80-something patient for a biopsy the next morning, the one that will tell her team whether the patient has cancer. Those are the moments where the design layer earns its place or doesn't. It decides whether the system points a clinician at the right patient at the right time, in a form she can read and trust without compromising her license, in the seconds she actually has. The tools to build that have changed and the time to ship it has collapsed. The intelligence layer has not moved an inch. It sits where it has always sat, between what a business knows and what a person experiences, deciding whether the two ever meet.

Design isn't dead. It's being asked, yet again, to prove it was always essential to the business. The companies writing the largest checks have already done the math. It was. It is.

Written by Gauri Sharma, June 2026. v1, a working theory and therefore provisional. Currently advising at the intersection of clinical AI, healthcare, and trust-critical service design.