On November 6, Sweetgreen announced that it was selling Spyce, its division that developed and made its Infinite Kitchen technology to automate the assembly of its bowls and salads. The acquirer is Wonder, the “restaurant and mealtime superapp,” as Fast Company dubbed it earlier this year.
With that, it’s time to eulogize Sweetgreen’s star-crossed life as a tech company. No more dreams of AI, blockchain, or robots.
Sweetgreen is to receive $100 million in cash and $86.4 million in Wonder stock—a positive return, given that it acquired Spyce in 2021 for a total cost of $70 million. Wonder, which is privately held, was valued north of $7 billion in May after it raised another $600 million. Sweetgreen, which went public four years ago, has a market capitalization of under $750 million.
After Sweetgreen’s disastrous Q2 2025 earnings report, I wrote that Infinite Kitchen represented “the first effort by the company to use technology to solve its biggest problem—operations—rather than mere magic dust sprinkles to make the company look like something it’s not.”
Now the company’s latest earnings are worse, and it doesn’t own what had felt like a competitive advantage.
“A lot of other companies are trying to figure out how to add automation to their experience and are not willing to start over,” Sweetgreen CEO Jonathan Neman told The Wall Street Journal in 2023 while showing off his first restaurant equipped with an Infinite Kitchen. “I’m willing to blow the whole thing up.”
The question, though, is when did Neman light the fuse that blew up Sweetgreen? Was it two years ago? Was it just November 6? Or was the bomb planted in the company’s earliest days, and it’s finally detonated? Sweetgreen’s stock is down another 12.5% as of Friday afternoon. (In response to queries, Sweetgreen directed me to Neman’s public statements.)
In this piece, we’ll explore:
- What we still don’t know about the sale of Infinite Kitchen
- Whether Neman could have taken a page from Pixar or Tesla to alter Sweetgreen’s course
- Why Neman has even harder decisions ahead to make Sweetgreen profitable
- How Sweetgreen’s positioning as a tech company ultimately failed it
Infinite Kitchen has been working
Sweetgreen remains committed to deploying Infinite Kitchen; it opened eight restaurants in Q3, and six included the tech. More are planned for 2026.
Rather than be responsible for developing and making the systems, Sweetgreen will buy them from Wonder at cost plus 5%, which Neman said was about $25,000, putting the Infinite Kitchen’s cost at $500,000.
In turn, Sweetgreen promised investors that the sale would shave $8 million annually from its general and administrative expenses. Those G&A costs are high. As the veteran restaurant operator and consultant Rick Vanzura noted on LinkedIn, Sweetgreen’s overhead was 17.9% of sales compared with Cava’s 10.8%.
But $8 million is just over 1% of expected 2025 revenues, meager savings for proprietary technology that Neman lauded again this week as having:
“consistently proven its ability to deliver faster throughput, improved accuracy and consistency, and elevated food quality.”
In Q3, restaurants with an Infinite Kitchen “continue to realize approximately 700 basis points of labor savings and nearly 100 basis points of [cost of goods sold] improvement compared to restaurants of similar age and volume.”
Why give up control of the tech driving 7% labor savings per quarter and 1% in food costs while it’s improving the product itself?
Why Sweetgreen sold its big tech bet
The Occam’s razor explanation appears to be that Sweetgreen really needed the money. Look at its cash on hand:
Q3 2025: $130M
Q2 2025: $168M
Q4 2021: $472M
In August, I anticipated that Sweetgreen would soon require fresh capital. I wondered whether the parties providing it would demand company control from Neman and his two cofounders in exchange.
This move cleverly sidestepped that possibility (for now) by selling the most valuable thing Sweetgreen owned that it could part with to a private company, buying Neman and company time to turn things around.
Neman still likely needs to do a more wrenching corporate restructuring that vastly reduces its overhead (read: major layoffs). The company’s new CFO reported that she’s launched a full review of the company’s restaurant expenses as well as its G&A. We’ll see if Neman can make some hard decisions to reinvent Sweetgreen.
The logic underpinning the Spyce sale may be irrefutable, but there’s still a lot we don’t know. To wit:
- Whether Wonder can also license the Infinite Kitchen tech
- How long Sweetgreen’s cost-plus deal lasts
- Whether those terms also apply to future Spyce innovations
I don’t expect we’ll get direct answers, but this is what investors in particular should be thinking about and monitoring.
Sweetgreen’s “Sliding Doors” moment No. 1: The Pixar path
In 1989, Pixar, six years before the debut of Toy Story, decided to sell its RenderMan technology to other companies. Pixar needed cash, especially if it was going to fulfill its vision of making feature-length computer-generated animated movies. The gambit worked. Pixar retained control of the tech, has enhanced it repeatedly over the years, and major motion pictures from other studios still rely on RenderMan.
Could Sweetgreen have decided to license the Infinite Kitchen technology to its competitors rather than sell the tech to one of them and then become a licensee? Doing so could have helped bring down the costs of Infinite Kitchen and spurred further innovation, as Wonder now hopes to do.
Given how hot the private markets are for robotics tech, could Sweetgreen have engineered some complex financial deal to get funding for Spyce to scale it without having to sell it? I don’t think that’s too outlandish an idea.
Alas, public market investors haven’t been patient with Neman (the stock is down almost 90% since it went public).
This would have been bold and visionary in 2021 after Sweetgreen acquired Spyce, or in 2023 when Neman talked of his willingness to “blow the whole thing up” and energized investors with the Infinite Kitchen’s potential.
Making that call in late 2025, when consumers appear to be cooling to bowl-based meals (the “slopcession,” or “slopapocalypse,” as it were), would have been risky. But the siren call of those labor and cost savings could have won it some customers and allowed it to control its destiny.
Sweetgreen’s “Sliding Doors” moment No. 2: The Tesla way
In 2014, Elon Musk open-sourced Tesla’s electric vehicle patents. This, too, was a bold move for a still shaky, unprofitable company. Musk did it to accelerate the auto industry’s adoption of EVs, which it did.
At the time, Tesla’s market cap was approximately $28 billion.
Today it’s $1.4 trillion.
What if Sweetgreen had open-sourced Spyce’s patents? Would it have sparked a wave of innovation in automated restaurant tech? This is less likely than if it merely licensed systems to rivals, as the restaurant industry is far more atomized than the car business.
But the move would have been a bravura stroke that, at the least, would have bolstered Neman’s narrative that Sweetgreen is a different kind of company.
Live by the tech narrative, die by the tech narrative
Not long after Sweetgreen went public in November 2021, Kristen Hawley, a Fast Company contributor, wrote in her food and tech newsletter, Expedite, the uncomfortable truth that “salad doesn’t scale like software.”
Now we can confirm that restaurant automation hardware to make salads and bowls doesn’t scale like software either.
Companies need a story, a vision to sell to investors, media, and customers. It’s why Tesla backers voted to give Elon Musk his potential $1 trillion pay package this past week. For Sweetgreen, its story has long been that this was a tech company—no, a platform—that sold healthy salads and bowls rather than a restaurant company that used tech like, um, every other restaurant chain.
I wanted to find the first instance of Sweetgreen publicly presenting itself as a tech company, and I believe it initially did so on the occasion of its 2011 Sweetlife festival, thanks to a planned integration with a then-buzzy social app:
“We look at ourselves less as a restaurant group than a think tank,” cofounder Nathaniel Ru told Mashable. “We’re more of tech startup than a restaurant business.”
Fourteen years later, Sweetgreen is a restaurant business. Its future success will be determined by continuing to improve its operations, developing new menu items, and marketing itself as a “lifestyle brand,” as Neman told investors, that’s “focused on creating culture through distinct brand moments.” Again, like every other restaurant chain. As I understand it, the company still has a tech team, but so does everyone else. The tech dream may die hard at Sweetgreen HQ, but die it should.
In other words, the troubled company’s tech Cinderella story is over. Sweetgreen’s enchanted digital coach has become a garden-variety analog pumpkin.
