David Holz

Holz grew up in Florida, ran a design business in high school, and studied math and physics; he has said he went to college for physics and math and later pursued a PhD, described in different interviews as being in fluid mechanics or applied mathematics.⁠[1][2][1]

In 2008, Holz took a leave of absence from his PhD program and co-founded the hardware peripheral company Leap Motion with Michael Buckwald, who later became CEO. The company developed hardware that tracks hand motion for use as device input and became known as a pioneer in VR hand-tracking controls, with Holz as CTO. In another interview, Holz recalled moving to San Francisco and starting Leap Motion around 2011.⁠[1][1][3][1][3][3][2]

The year after co-founding Leap Motion, Holz spent a year as a student researcher at Germany’s Max Planck Institute, followed by two years as a graduate student researcher at NASA’s Langley Research Center working on lidar, Mars missions, and atmospheric science. He recalled asking himself why he was working on so many things and deciding to work on one thing he cared about, so he focused on Leap Motion.⁠[1][1]

On May 30, 2019, the British company Ultrahaptics partially confirmed that it was acquiring Leap Motion for about $30 million, roughly a tenth of the company’s valuation a few years earlier; Apple had twice pursued Leap Motion without completing a deal. Holz said the two companies had worked together for almost six years and that their technologies were complementary rather than competitive; he and the engineering staff joined Ultrahaptics, while CEO Buckwald did not.⁠[3][4][4][3][3][3]

Holz ran Leap Motion for about twelve years, and it had about 100 employees when he left; in 2021 he left to found Midjourney. He said he wanted an environment different from a big venture-backed company, a home for the next ten years to work on cool projects that matter. He recalled noticing how much AI’s ability to understand language was advancing and said what struck him most was CLIP-guided diffusion, developed by Katherine Crawson.⁠[1][1][2][2][1][1]

In August 2022 interviews, Holz said Midjourney had about 10 people, was self-funded with no investors, and was not primarily financially motivated; he described it as a self-funded research lab that was already profitable, with a business model in which users pay the cost of using the service plus a percentage on top. He called the technology an “engine for the imagination” and said that while the technology’s progress was easy to foresee, its human ramifications were hard to imagine and required many experiments.⁠[1][1][1][2][1]

Asked about the copyright dispute over artists’ work used as training data, Holz said artists in the community were universally positive about the tool and thought it would make them more productive, and that he held office hours in which he answered questions from about 1,000 people on voice chat for four hours.⁠[2][2]

In a September 2022 interview with Forbes, Holz acknowledged that Midjourney’s training data was “just a big scrape of the Internet” drawn from published open datasets, calling it “something that 100% of people do” and adding, “We weren’t picky.” He also said there was no real way to get a hundred million images and know where they came from, that it would be good if images carried embedded copyright-owner metadata, but that no such registry exists and a picture found online cannot be automatically traced to its owner.⁠[5][5][5]

After the interview circulated online in December 2022, many artists and photographers criticized Holz for knowingly using datasets containing millions of existing works without consent and showing no remorse; one photographer noted that many works already carry embedded copyright metadata and contact information. In the same interview, Holz defended Midjourney’s goal as making humans more imaginative rather than making imaginative machines, saying the machine has no intention and that the company’s intention has nothing to do with art.⁠[5][5][5][5]

In March 2023, after fake images made with Midjourney, such as Donald Trump being arrested and Pope Francis in a puffer coat, spread widely, Midjourney suspended free trials on March 28. Holz denied that deepfakes were the reason, saying users signing up for multiple free trial accounts were bringing down the service for paying users; he said the company was still figuring out the right moderation policies, taking feedback from experts and the community, with new systems coming soon. The company later tried reopening trials with new abuse safeguards but turned them off again when those proved insufficient.⁠[6][6][6][6]

Midjourney allowed users to generate images of world leaders such as U.S. President Joe Biden and Russian President Vladimir Putin but not of Chinese President Xi Jinping. Holz explained on Discord that the company just wanted to minimize drama, that “political satire in china is pretty not-okay,” and that “the ability for people in China to use this tech is more important than your ability to generate satire.”⁠[7][7]

On March 31, 2023, Sarah McLaughlin, a scholar at the Foundation for Individual Rights and Expression (FIRE), criticized the rule, arguing that it applied not only to users in China but to users everywhere, including the United States, effectively spreading an authoritarian state’s censorship rules worldwide through a tech company; she argued that Holz’s stated aim of preserving Chinese users’ ability to use the technology overlooked that Chinese citizens may themselves want to satirize their own government.⁠[7][7][7]

On June 11, 2025, Disney and Universal sued the Holz-run Midjourney for copyright infringement in federal court in Los Angeles, calling it a “bottomless pit of plagiarism.” The complaint quoted Holz as admitting that, to collect training data, Midjourney “pulls off all the data it can, all the text it can, all the images it can.”⁠[8][8]

On June 17, 2026, at an event in San Francisco, Holz unveiled Midjourney’s first hardware product, the full-body ultrasound scanner Midjourney Scanner, along with a new Midjourney Medical division. He claimed no such device had ever been built, said its imaging was in many ways superior to MRI, and said the goal was a fleet of 50,000 scanners, declining to give a price. He also acknowledged that the device was “not even using any AI in this yet, just really cool hardware and software.”⁠[9][10][9][10][9]

The scanner was developed with the ultrasound device maker Butterfly Network, with which Midjourney signed a licensing agreement in November 2025 securing exclusive rights to its ultrasound-on-chip technology; Midjourney plans to run research trials, seek FDA approval for diagnostic capabilities, and hopes to deploy 50,000 scanners worldwide by 2031.⁠[11][11][11]

Ana Maria Constantin, writing for The Next Web, questioned Holz’s claims: Midjourney had never built a physical product, the scanner had no regulatory clearance and currently produced only body composition maps, and any diagnostic use would require approval from the U.S. Food and Drug Administration, which the company said it had not yet obtained and intended to pursue over time; she added that whole-body screening of healthy people is contested in medicine, chiefly because false positives and incidental findings lead to unnecessary follow-up procedures.⁠[9][9][9][9]

X: Public account: https://x.com/DavidSHolz (opens in a new window)⁠[12]

X (second account): Public account: https://x.com/davidvibesonly (opens in a new window)⁠[12][13]

Midjourney: Official website: https://www.midjourney.com/ (opens in a new window).⁠[12]