The Morning Download: The AI Flywheel Speeds Up

Good morning. The power and durability of AI, so evident right now, has defied easy explanation. Is it a bubble? A boom? Neither concept captures the remarkable period that began with the launch of ChatGPT in late 2022 and continues to this day.

The idea of the flywheel does a better job of capturing what’s going on in AI and in tech more broadly. The AI market consists of many elements, and the success of one is providing the energy and resource to feed the next, eventually coming full circle in a self-sustaining pattern. Nvidia CEO Jensen Huang has been drawing on this concept for years as a way to describe the company’s business model. In March, Isabelle Bousquette and I went to Nvidia’s GTC event in San Jose, where Jensen pointed to a flywheel with the company’s CUDA software system as its core.

“The installed base of CUDA is the reason why the flywheel is accelerating. The installed base is what attracts developers who then create new algorithms that achieve a breakthrough … Those breakthroughs lead to entirely new markets, which build new ecosystems around them, with other companies that join, which creates a larger installed base,” Huang said during his keynote speech.

That flywheel effect describes more than Nvidia and its ecosystem of partners and clients. It describes the broader ecosystem of AI and tech itself, including the rapid development of quantum computing. We’re seeing evidence of that flywheel right now in everything from Nvidia’s quarterly earnings, to Anthropic’s impending profitability and the approaching IPOs of Anthropic, OpenAI and SpaceX. CoreWeave issued an investment-grade financing in March, and in recent weeks beaten-down stocks such as ServiceNow and Adobe have come off lows of recent months, all signs that the flywheel is drawing in more elements of tech from infrastructure to software.

Now, it remains to be seen if that flywheel includes enterprise tech companies, the end-users of that technology and the drivers of the revenue that supports all of the capital spending and investment behind it.

A flywheel doesn’t necessarily spin forever. It can slow down, stop or go into reverse. Right now, however, it is accelerating, as we see below. Given the level of scientific and technological innovation driving the tech flywheel, there’s good reason to think that it will continue to gain momentum.

U.S. to award quantum computing firms $2 billion. The Trump administration is awarding $2 billion in grants to nine quantum-computing companies, including a $1 billion allocation to IBM, the Journal reports. Other companies slated to be awarded funds include publicly traded firms D-Wave Quantum, Rigetti Computing and Infleqtion.

The government will receive a minority equity stake in each quantum company. Funding comes from the 2022 Chips and Science Act passed during the Biden administration and now overseen by Commerce Secretary Howard Lutnick. 

Workday’s returning CEO has a plan to survive the AI era. Workday co-founder Aneel Bhusri, who returned to lead the business software company in February, tells the WSJ Leadership Institute's Belle Lin that he is overseeing what he calls the company's “re-founding” moment—one in which the stakes are incredibly high.

Pleasanton, Calif.-based Workday’s stock has been battered by investor fears that AI will make conventional business software obsolete.

Bhusri founded the human-resources and finance software company alongside Dave Duffield in 2005, and served as CEO through multiple tenures, most recently stepping down in early 2024 to become executive chair. He came back to the helm with a plan to guide the organization through an era in which software’s future is in question.

For Workday, getting back to those roots has meant building an AI task force, consolidating teams around its AI agent products and services, and eliminating the idea of “shared ownership of new products” which led to internal confusion, according to Bhusri.

On Thursday, Workday will announce agents for corporate travel and a new segment: IT service management. The new agents, which build on Workday’s $1.1 billion acquisition of AI firm Sana, will help employees book travel, manage expenses and automate common IT tasks like employee onboarding and off-boarding. 

Such an expansion of Workday’s AI agent offerings puts it in competition with longtime IT service management provider ServiceNow, which has AI agents of its own.

“Are we going to overtake ServiceNow day one? No,” Bhusri told Belle. “But we thought, ‘Why not us? We have this incredible platform in Sana, so why don’t we do the agentic workflows as well?’”

Lights! … Camera! … Prompt! One of the films debuting this week at cinema’s yearly conclave cost just $500,000 to make. About $400,000 of that was AI compute costs. WSJ Leadership Institute's Isabelle Bousquette spoke with Higgsfield AI, the San Francisco-based startup behind "Hell Grind," a 95-minute fully AI-generated film.

AI has been a hot topic on the French Riviera and Hollywood, with industry players and technologists questioning how much of the moviemaking process can or should be outsourced to the tech and what it means for jobs and human creativity.

According to the company, AI-powered filmmaking is more than just writing a prompt.

“You have to understand camera composition, which shots are changed. Like you can’t have two close-ups back to back, you have to start with an establishing shot,” said Adilet Abish, an in-house director and creative producer at Higgsfield. “You still need those filmmaking skills.”

Yes, there is still prompting involved, but this is not your 2024-version. 

Every prompt had to be extremely long and detailed. Each one would typically start with a prefix that defined requirements like style (8k IMAX, photorealistic), lighting (natural light only, “contre-jour” backlight, camera on shadow side) and the type of camera it should look like it was being shot on (“cine lens,” 180-degree shutter motion blur).