• Zhipu’s open-weight AI model GLM 5.2 wow developers, but its stock has plunged.
• Open-source software benefits from easy, cheap distribution. Open-weight AI is very different.
• Other companies often run Chinese open-weight models. That’s where the money is made.

The angst over China’s latest AI models is missing an important business fact: “open weight” AI is not the same thing as open-source software.

Open-source software, where the code is freely shared, can be an amazing business. Think Red Hat, which IBM bought for $34 billion. Open-weight AI models are different — and, so far, they’re proving to be a terrible business.

Take Z.ai, also known as Zhipu. It’s publicly traded, so we can see its finances. Last year, the Chinese company lost almost $500 million on revenue of about $107 million.

Zhipu is the lab behind GLM 5.2, an open-weight AI model that wowed the industry when it launched last month. You might expect the stock to have soared. Instead, Zhipu shares have plunged more than 40% over the past month.

MiniMax, one of the only other independent Chinese AI labs that’s publicly traded, lost $250 million last year on revenue of just $79 million. Its shares have fallen more than 50% in the past month.

“Open-weight models have a challenging path to making a profit,” William Blair analyst Arjun Bhatia wrote in a recent note to investors, in the understatement of the year.

Open-weight AI isn’t open-source software

The key difference comes down to economics.

Software can be distributed almost for free. Once it’s written, sending another customer a copy costs practically nothing. Profit margins improve as software companies grow.

AI doesn’t work that way. Every answer requires expensive chips, electricity, and data-center capacity. The next unit of software is nearly free; the next unit of intelligence is not.

Moonshot AI, another Chinese lab, illustrated the problem last week. Its new Kimi K3 open-weight model impressed the industry with frontier-level performance. But days after launch, the company had to halt new customer sign-ups because it didn’t have enough computing power to run the model.

If Moonshot were selling traditional software, adding millions of users would be relatively easy. Instead, every new customer increases the company’s infrastructure bill, capping its growth.

Someone else captures the profits

The way open-weight AI models are run, a process known as inference, makes the business situation worse.

Open-weight AI labs give outsiders their models’ trained numerical parameters, allowing them to download and run them. (Parameters are like tiny numerical dials inside a model’s brain that determine how these systems learn from data and what outputs they produce).

After that, these models are usually run by other companies, such as cloud giants Amazon, Microsoft, Google, Oracle, and Alibaba. There are also specialist providers such as Fireworks AI and Baseten, although they largely rent capacity from the big cloud companies.

Companies can also download these open weights and run the models themselves. Or, they can also use the Chinese model maker’s own inference service, but in the Western world, most corporate customers don’t do that for data security reasons.

Only that last option reliably generates real revenue for the model creator. In the other three cases, the AI lab that spent hundreds of millions of dollars building the model may receive little or no ongoing revenue.

That leaves the model makers in a difficult position. They’ve paid heavily to train the systems, then given away the key assets.

No wonder Alibaba’s stock is up about 13% over the past month, while AI labs Zhipu and MiniMax have been crushed.

Not Red Hat

“Unlike open-source software, open-weight models do not generate significant sums of revenue by selling support, services, and enterprise editions around the free offering (the Red Hat playbook),” William Blair’s Bhatia wrote.

“Instead, they primarily generate revenue by hosting the model and selling inference compute. But inference workloads will flow to whoever can operate the inference infrastructure most efficiently, and this is usually not the model provider,” the analyst added.

Raimo Lenshow, an analyst at Barclays, recently came back from China after researching the country’s AI sector. He reached a similar conclusion.

“Intense domestic competition has also led to more aggressive pricing competition,” the analyst told investors. “Some major models remain open-source or open-weight, accelerating the pricing pressure throughout the system. While this helps drive faster commercialization, it is also adding uncertainty to long-term profitability for those AI labs.”

So why give the models away?

Open technology has long been a strategy for challengers trying to catch market leaders. A late starter may not be able to match a leader’s customers or distribution, but it can spread its technology widely, attract developers, and make the leader’s product harder to sell at premium prices.

That may be exactly what China and its AI labs are trying to do. Open-weight models put pressure on OpenAI, Anthropic, and other US leaders by offering capable alternatives at lower prices. Even if the Chinese labs make little money themselves, they can force American competitors to cut prices and make it harder to recover the billions they spend training new models.

Bhatia said Chinese labs may be releasing open-weight models with “little regard for near-term profitability.” In his view, openness can turn advanced AI into a commodity, weakening the business model of US companies that keep their technology closed.

The financial payoff for the Chinese labs may come much later — or may be less important than the broader strategic benefit to China.

China’s leadership is now openly encouraging that strategy. In a recent speech in Shanghai, President Xi Jinping said countries should seize the opportunity to encourage “open-source, openness, collaboration, and sharing.”

A statement like that is more than a casual policy suggestion. Chinese technology companies are expected to align with the government’s strategic priorities.

After Xi put openness at the center of China’s AI strategy, companies such as Moonshot, Zhipu, and MiniMax are likely to face strong pressure to follow that direction — even if it makes their own path to profit much harder.

Sign up for BI’s Tech Memo newsletter here. Reach out to me via email at abarr@businessinsider.com.

• Curative’s CEO said the company canceled an annual $600,000 Salesforce bill.
• He said his company replaced the CRM with a tool that AI helped build in two months.
• He’s describing the SaaSpocalypse thesis: that AI coding tools will challenge software companies.

Curative CEO and founder Fred Turner believes in the SaaSpocalypse.

During an interview on the “20VC with Harry Stebbings” podcast, the health insurance executive was asked whether he bought the theory that software-as-a-service was dying.

“Yes,” Turner said, bluntly.

When asked why, he said, “I see the number of contracts that we’re canceling. We just recently canceled our Salesforce contract because we have an internal CRM that was vibecoded.”

That Salesforce contract cost Curative $600,000 a year, Turner said. He also said that Curative built its replacement in two months.

Overall, Curative plans to cut about 80% of its SaaS spending this year, according to Turner. The company is spending on AI instead, he said.

A Curative spokesperson told Business Insider that the startup had made a “notification of cancellation” for its Salesforce CRM contract.

“As Fred said publicly, Curative terminated its Salesforce CRM contract,” a Curative spokesperson said. “We do still use Slack,” the spokesperson said. Salesforce purchased Slack in late 2020 for $27.7 billion.

A Salesforce spokesperson told Business Insider that 150,000 companies still use its platforms and highlighted that its tools are built to navigate complex healthcare patient regulations, such as HIPAA.

“Our platform is built with trust and governance at its core,” a Salesforce spokesperson told Business Insider.

Salesforce’s CEO has pushed back on SaaSpocalypse fears

Turner is describing the exact SaaSpocalypse fear that spooked Wall Street at the start of 2026.

As AI coding agents quickly improved, investors feared that companies that bought software-as-a-service would turn to AI tools and build their own bespoke products. Software providers — such as Salesforce, Asana, DocuSign, ServiceNow, Adobe, and Workday — saw their stocks drop by 20% to 50% amid those snowballing concerns.

Salesforce’s CEO, Marc Benioff, has also strongly pushed back, saying that he’s still seeing “incredible demand” for his products. He’s also pointed out that Anthropic, the massive AI lab behind Claude, still uses Salesforce services.

“If there is a ‘SaaSpocalypse,’ it may be eaten by the ‘SaaS-quatch’ because there are a lot of companies using a lot of SaaS because it just got better with agents,” Benioff said during a February earnings call.

Turner acknowledged that replacing outside software with custom-built systems hasn’t been perfect. Maintenance is “definitely one of the most challenging pieces,” he said.

He also said Curative’s spending on Anthropic had surged as the company found more uses for AI.

“Our Anthropic cost over the last six or seven months has 6x’d every month, from a base of a couple of tens of thousands of dollars, now up to millions of dollars a month,” Turner said. “Eventually, we’re going to have to stop that spending increase because it’ll get unreasonable, but we just keep finding new things to do with it.”

For Curative, however, Turner said the economics would still work — even if Anthropic were to quintuple its prices.

He pointed to Gwen, a bespoke AI agent Curative uses to negotiate contracts with doctors and other healthcare providers. Turner said completing one contract before AI had cost the company an average of $1,500 to $2,000. Gwen’s average cost is about $70, he said.

“What we’ve done is said, ‘Well, now that we have the agent, we can do 10 times as many contracts this year as we could do last year,'” Turner said. “So, we’re going to do 10 times, and then we’re going to try and do 20 times, and we would just do a lot more volume than you could possibly have done with a human team.”

• Mike Recupero is the first VP of finance for Amazon’s Leo satellite business.
• Amazon is putting together a seasoned leadership team for Leo.
• Amazon expects Leo to generate meaningful growth and returns.

Amazon’s Leo satellite business has appointed its first dedicated finance VP, another sign the company is building the satellite venture into a more independent business.

Mike Recupero, who most recently served as Alexa’s finance chief, was named Leo’s VP of finance earlier this month, according to people familiar with the move. The newly created role makes him the first VP solely overseeing Leo’s finances.

Previously, Recupero spent about a year as GameStop’s CFO after serving as finance chief for several Amazon businesses, including Prime Video and North America retail.

Until now, Leo’s finances were overseen by executives who also managed Alexa, supported by more junior finance leaders.

The appointment reflects how Amazon is assembling a seasoned leadership team around Leo, formerly known as Project Kuiper.

Over the past two years, Leo has also recruited former GitLab chief revenue officer Chris Weber as vice president of sales and marketing and former T-Mobile executive Clint Patterson as chief marketing officer. VP of technology Rajeev Badyal leads the overall Leo business and reports to Panos Panay, SVP of devices, Alexa, and Leo.

Recupero will primarily oversee Leo’s multibillion-dollar infrastructure buildout, including satellite manufacturing and launches, as well as the integration of Globalstar, the satellite communications company Amazon is acquiring for $11.6 billion, the people said.

Amit Singh has replaced Recupero as Alexa’s finance lead.

Leo is one of Amazon’s biggest long-term bets beyond its core retail and cloud businesses. CEO Andy Jassy previously said Leo already had a series of revenue commitments from enterprise and government customers and is expected to generate meaningful growth and returns for Amazon. The company said earlier this month that Leo has completed 14 missions and launched 396 satellites so far, making it the third-largest satellite constellation in orbit.

Wall Street is also growing more bullish on the business. Bank of America recently estimated Leo could generate $20 billion to $25 billion in annual revenue by 2032 and eventually be worth $200 billion to $275 billion.

The growing enthusiasm has not eliminated the risks. A Blue Origin rocket that Amazon plans to use for future Leo missions exploded during a ground test in June. Leo’s VP Badyal sought to reassure employees at the time, saying such setbacks are an expected part of spaceflight and that the company would continue pressing ahead.

Amazon declined to comment.

Have a tip? Contact this reporter via email at ekim@businessinsider.com or Signal, Telegram, or WhatsApp at 650-942-3061. Use a personal email address, a nonwork WiFi network, and a nonwork device; here’s our guide to sharing information securely.

• Corner, a social mapping startup that helps people find local spots, has doubled its funding.
• The New York-based startup has raised $7.5 million, the company confirmed.
• It’s one of many new social media startups trying to break through.

Finding a really good restaurant or bar takes effort.

Corner, a startup looking to make that process easier, has raised more capital to stake its claim in the map game.

Corner, which was founded in 2022, is a social-mapping app targeting Gen Z users who want to meet up in real life. In the app, users can leave reviews, share photos, and curate lists. Locations are added to Corner’s map only when a user saves it.

The platform has more than 250,000 users — primarily in New York and Europe — across 425 cities, the company said.

The startup has nearly doubled its total funding since 2025.

In September, Corner told Business Insider that it had raised $3.75 million across two rounds from VC funds including Abstract Ventures, Tapestry, and 1517. The startup also has angel investors, like Partiful’s cofounder Shreya Murthy.

The New York-based startup has raised new funding since, with total funding reaching $7.5 million, Corner confirmed. The startup is gearing up to raise a Series A round.

Corner isn’t the only app trying to make waves in the social-mapping and restaurant-saving categories.

Beli, a rival app for saving and sharing restaurants, had raised $12 million in venture capital as of June 2025, the company previously told Business Insider. Apps like Amo’s Bump, a social app for tracking your friends, are also gaining steam. Meanwhile, Corner has to compete with larger players like Google Maps, Apple Maps, and Yelp.

Corner has carved out a following on social media as it grows its platform and business. Its digital magazine has over 91,000 followers on Instagram. Eliza Wu and Jake Xia, Corner’s cofounders, have also become faces of the brand on social.

A map with personalized recommendations

Last year, Corner rolled out an overhauled search feature that leverages AI to suggest recommendations based on Corner reviews and content.

“Our proprietary dataset powers true personalization instead of virality-based recommendations,” said Corner CEO Wu.

Corner’s been experimenting with new features, too.

Recently, I’ve noticed Corner will let me know when a reservation opens up at a restaurant I’ve saved. The app also now lets you add bookings and events, provides a round-up of places that people — including your friends — are raving about, and offers age- and location-based suggestions on where to go out.

“Of everything else that I’ve seen, and I’ve seen this idea tried a dozen times in the last 10 years, this was the best execution I’d seen,” Abstract Ventures’ Ramtin Naimi, a Corner investor, told Business Insider in September.

• The FDA said an earlier test involving a Taylor Farms lettuce sample produced a “false positive.”
• The agency later clarified that there is still “overwhelming epidemiological data” supporting a recall.
• Amid the confusion, a question remains: should I eat salad or not?

As the FDA and Taylor Farms go back and forth on the source of the cyclosporiasis outbreak in the US, Americans still have one pressing question: Will my salad give me explosive diarrhea?

On Sunday, the FDA said a test pointing to Taylor Farms as a potential source of Cyclospora was a false positive.

Taylor Farms said on Sunday that the FDA had apologized to the salad juggernaut, but its voluntary recall of iceberg lettuce grown and processed in central Mexico in the US market was still in place.

In a separate statement shared Monday, the FDA said: “To clarify, this false-positive lab sample DOES NOT change the basis for FDA’s ongoing outbreak investigation or the overwhelming epidemiological data supporting the current voluntary recall by Taylor Farms.

It added its “traceback investigation and outbreak data continue to converge on shredded iceberg lettuce from Taylor Farms locations in central Mexico.”

So far, so confusing.

Dr. Tyler Evans, an infectious disease doctor and the author of “Pandemics, Poverty, and Politics,” said that deciding whether or not you want to eat salad depends on your risk tolerance, as well as where you live.

To help you decide, we broke the risk down into three tiers: low, medium, and throwing caution to the wind.

Since the outbreak was detected in May, more than 8,260 confirmed cases of cyclosporiasis have been reported to state and federal health departments, according to data compiled by the Georgetown University Health Security Operations Center.

The bulk of the cases have been reported in Michigan, Ohio, Indiana, Illinois, New York, North Carolina, and Colorado, but that could simply be because those states have more robust public health surveillance than others.

The cyclospora parasite thrives in hot summer weather. Public health authorities say at worst, this outbreak could last until the end of August, but that’s a long, sweaty stretch of weather to face without any leafy greens in your diet.

Evans is based in California, where there isn’t much cyclosporiasis being reported, and said he just had salad at a Mexican restaurant.

“I’m a very risk-tolerant infectious disease doctor,” he said.

If you’re older, immunocompromised, pregnant, under the age of 2 (with a developing immune system), or have a medical condition, you may want to play things a little safer until the fall.

Lowest risk of infection

To ensure you will definitely not contract cyclosporiasis, choose only vegetables that are cooked to at least 158 degrees Fahrenheit, the temperature that kills this parasite. If you like berries, make jam. Craving veggies? Try grilled asparagus.

If you’re in a high-risk category, “I would completely refrain from not just lettuce or iceberg lettuce, but probably all leafy greens for the foreseeable future,” Evans said.

Last week, Business Insider spoke to a GI doctor who recommended some high-fiber swaps for those who can’t face fresh produce right now.

Medium risk

Buy whole heads of lettuce, remove and discard the outer layer of leaves, and wash them thoroughly under cold running water before consuming. You can also shop for lettuce grown locally at farmers’ markets. Keep an eye on your local case numbers to inform your decisions.

“In states or areas where there’s a higher incidence, then I think using the mantra to peel it, boil it, cook it, or forget it is a good idea,” Evans said.

Throwing caution to the wind

Eating raw, unwashed leafy greens, especially bagged pre-chopped lettuce and salad kits, which have been the most common source of cyclosporiasis outbreaks in recent years, may be risky.

Other vegetables that have been contaminated with the parasite in the past include raspberries, snap and snow peas, and fresh herbs (cilantro, basil, parsley). It’s possible that none of these foods are involved in the current outbreak, and that the risk is limited to only leafy greens, but we don’t know for sure.

“Once the temperatures start dipping, we should be in the clear again,” Evans said. “But the systems can do a lot better. There’s some degree of acceptable risk in terms of the seasonality outbreaks, but we can do better in terms of detecting and responding to these things.”

• Current and former students of 10 elite schools are receiving payments from a $284 million financial aid settlement.
• Recipients are expected to receive an average of $2,000.
• It follows an antitrust lawsuit that accused schools of participating in a “price-fixing cartel.”

Check your bank account: you might be a couple of thousand dollars richer.

On July 20, the first batch of payments to current and former students who attended 10 elite schools was distributed from a $284 million settlement. It follows a 2022 lawsuit filed by students who accused the schools of participating in “a price-fixing cartel” that favored wealthy students and reduced financial aid packages for students, overcharging recipients “by at least hundreds of millions of dollars,” according to the lawsuit.

The schools denied wrongdoing and agreed to make settlement payments to resolve the claims. According to the settlement website, current and former students who attended the following schools will receive payments of an average of $2,000: Brown, University of Chicago, Columbia, Dartmouth, Duke, Emory, Northwestern, Vanderbilt, Rice, and Yale.

Payments are being sent out through ACH, PayPal, Zelle, and Venmo. Those who opted for a physical check are expected to receive it by July 24.

Current and former students are expected to receive payments if they were enrolled in one of the listed schools’ undergraduate programs full-time, received need-based financial aid from the school, and whose tuition and room and board costs were not fully covered by the financial aid they received.

Have a story to share? Contact this reporter at asheffey@businessinsider.com.