Connect with us

Finance

China’s latest AI model claims to be even cheaper to use than DeepSeek

Published

on

The World Artificial Intelligence Conference kicked off in Shanghai on Saturday, July 26, 2025.

Nurphoto | Nurphoto | Getty Images

BEIJING — Chinese companies are making smarter artificial intelligence models that are increasingly cheaper to use, echoing key aspects of DeepSeek’s market-shaking breakthrough.

Startup Z.ai, formerly known as Zhipu, announced Monday that its new GLM-4.5 AI model would cost less than DeepSeek to use. In contrast to the logic underlying existing AI models, Z.ai said its new GLM-4.5 is built on what’s known as “agentic” AI, meaning that the model automatically breaks down a task into sub-tasks in order to complete it more accurately.

The new model is also open sourced, meaning it is free for developers to download and use.

At about half the size of DeepSeek’s model, GLM-4.5 only needs eight Nvidia H20 chips to operate, Z.ai CEO Zhang Peng told CNBC on Monday.

That’s the chip Nvidia customized for China in order to comply with U.S. export controls. The chipmaker said this month that the U.S. will allow it to resume those China sales after a three-month pause, but it’s unclear when those shipments will begin.

Zhang said the company doesn’t need to buy more of the chips as it has enough computing power for now, but declined to share how much Z.ai spent on training the AI model. Details will be released later, he said.

Trump's AI strategy understands 'denying the chips doesn't work': CSIS

Back in January, DeepSeek had rattled global investors with its apparent ability to defy U.S. chip restrictions and create an AI model that not only rivaled U.S.-based OpenAI’s ChatGPT, but also undercut it in training and operating costs.

DeepSeek claimed training costs for its V3 model were less than $6 million, although some analysts said that figure was based on the company’s hardware spend of more than $500 million over time.

Z.ai said that for its new GLM-4.5 model, it would charge 11 cents per million input tokens versus 14 for DeepSeek R1; and 28 cents per million output tokens versus $2.19 for DeepSeek. Tokens are a way of measuring data for AI model processing.

Earlier this month, Alibaba-backed Moonshot released Kimi K2, which claimed to beat OpenAI’s ChatGPT and Anthropic’s Claude on certain coding capabilities. Kimi K2 charges 15 cents for every 1 million input tokens, and $2.50 per 1 million output tokens, according to its website.

Weekly analysis and insights from Asia’s largest economy in your inbox
Subscribe now

In late June, OpenAI named Zhipu in a warning about Chinese AI progress. The U.S. has also added the startup to its entity list that restricts American companies from doing business with it.

Z.ai launched in 2019 and is reportedly planning an initial public offering in Greater China.

The startup has raised more than $1.5 billion from investors including Alibaba, Tencent and Qiming Venture Partners, according to PitchBook. Aramco-backed Prosperity7 Ventures as well as municipal funds from the cities of Hangzhou and Chengdu are also among Z.ai’s backers, the database showed.

In the last few weeks, several other Chinese companies have announced new, open-source AI models. During the World AI Conference in Shanghai, Tencent released the HunyuanWorld-1.0 model for generating three-dimensional scenes for game development. Last week, Alibaba announced its Qwen3-Coder model for writing computer code.

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Finance

Why software stocks, 2026’s market dogs, have joined the rally

Published

on

ETF shelters from the Middle East War

Cybersecurity and enterprise software stocks have been market dogs in 2026, with fears that AI will wipe out a wide range of companies in the enterprise space dominating the narrative. But they snapped a brutal losing streak this past week, joining in the broader market rally that saw all losses from the U.S.-Iran war regained by the Dow Jones Industrial Average and S&P 500.

Cybersecurity has been “a victim of some of the AI-related headlines,” Christian Magoon, Amplify ETFs CEO, said on this week’s “ETF Edge.”

It wasn’t just niche cybersecurity names. Take Microsoft, for example, which was recently down close to 20% for the year. Its shares surged last week by 13%.

A big driver of the pummeling in software stocks was a rotation within tech by investors to AI infrastructure and semiconductors and some other names in large-cap tech, Magoon said, and since cybersecurity stocks and ETFs are heavily weighted towards software companies, they were left behind even as those businesses continue to grow on a fundamental basis.

But Wall Street now has become more bullish with the stocks at lower levels. Brent Thill, Jefferies tech analyst, said last week that the worst may be over for software stocks. “I think that this concept that software is dead, and then Anthropic and OpenAI are going to kill the entire industry, is just over-exaggerated,” he said on CNBC’s “Money Movers” on Wednesday.

Big Short” investor Michael Burry wrote in a Substack post on Wednesday that he is becoming bullish about software stocks after the recent selloff. “Software stocks remain interesting because of accelerated extreme declines last week arising from a reflexive positive feedback loop between falling software stocks and changes in the market for their bank debt,” he wrote.

The Global X Cybersecurity ETF (BUG), is down about 12% since the beginning of the year, with top holdings including Palo Alto Networks, Fortinet, Akamai Technologies and CrowdStrike. But BUG was up 12% last week. The First Trust NASDAQ Cybersecurity ETF (CIBR) is down 6% for the year, but up 9% in the past week.

Piper Sandler analyst Rob Owens reiterated an “overweight” rating on Palo Alto Networks which helped the stock pop 7% — it is now down roughly 6% on the year. Its peers saw similar moves, including CrowdStrike.

Stock Chart IconStock chart icon

hide content

Performance of Global X cybersecurity ETF versus S&P 500 over past one-year period.

Magoon said expectations may have become too high in cybersecurity, and with a crowding effect among investors, solid results were not enough to to push stocks higher. But the down-and-then-back-up 2026 for the sector is also a reminder that when stocks fall sharply in a short period of time, opportunity may knock.

“Once you’re down over 10% in some of these subsectors, you start to see the contrarians start to say, ‘well, maybe I’ll take a look at this,'” Magoon said.

He said AI does add both opportunity and uncertainty to the cybersecurity equation, increasing demand but also introducing new competition. But he added, “I think the dip is good to buy in an AI-driven world,” specifically because the risks to companies may lead to more M&A in cyber names that benefits the stocks.

For now, investors may look for opportunity on the margins rather than rush back into beaten-up tech names. “I think investors are still going to remain underweight software,” Thill said.

But Magoon advises investors to at least take the reminder to keep an eye on niches in the market during pronounced downturns. “The best-performing are often the least bought and do the best over the next 12 months versus late-in-the-game piling on,” he said.

While that may have been a mindset that worked against the last investors into cybersecurity and enterprise software in mid-2025 when the negative sentiment started building, at least for now, it’s started working for the stocks in the sector again.

Meanwhile, this year’s biggest winner is also a good example of what can be an extended trade in either a bullish or bearish direction. Last year, institutional ownership of energy was at multi-year lows, Magoon said, referencing Bank of America data. “Reverse sentiment can be a great indicator,” he said. 

But he cautioned that any selective buying of stocks that have dipped does have to contend with the risk that there is a potentially bigger drawdown in the market yet to come in 2026. That is because midterm election years historically have been marked by large drawdowns. “If you think it is bad right now, it could get a lot worse,” Magoon said. But he added that there’s a silver-lining in that data, too, for the patient investor. The market has posted very strong 12-month returns after midterm election drawdowns end. So, for investors with a longer-term time horizon and no need for short-term liquidity, Magoon said, “stick in there.” 

Sign up for our weekly newsletter that goes beyond the livestream, offering a closer look at the trends and figures shaping the ETF market.

Disclaimer

Choose CNBC as your preferred source on Google and never miss a moment from the most trusted name in business news.

Continue Reading

Finance

Violent downturns could test new ETF strategies, warns MFS Investment

Published

on

ETF Stress Tests: How funds are showing resilience in the face of uncertainty

New innovation in the exchange-traded fund industry could come at a cost to investors during extreme conditions.

According to MFS Investment Management’s Jamie Harrison, ETFs involved in increasingly complex derivatives and less transparent markets may be in uncharted territory when it comes to violent downturns.

“Those would be something that you’d want to keep an eye on as volatility ramps up,” the firm’s head of ETF capital markets told CNBC’s “ETF Edge” this week. “As innovation continues to increase at a rapid pace within the ETF wrapper, [it’s] definitely something that we advise our clients to be really front-footed about… Lack of transparency could absolutely be an issue if we’re going to start seeing some deep sell-offs.”

His firm has been around since 1924 and is known for inventing the open-end mutual fund. Last year, ETF.com named MFS Investment Management as the best new ETF issuer.

“It’s important to do due diligence on the portfolio,” he said. “Having a firm that has deep partnerships, deep bench of subject matter experts that plays with the A-team in terms of the Street and liquidity providers available [are] super important.”

Liquidity as the real issue?

Harrison suggested the real issue is liquidity, particularly during a steep sell-off.

“We’ve all seen the news and the headlines around potential private credit ETFs. That picture becomes much more murky,” he added. “It’s up to advisors, to investors [and] to clients to really dig in and look under the hood and engage with their issuers.”

He noted investors will have to ask some tough questions.

“What does this look like in a 20% drawdown? How does this liquidity facility work? Am I going to be able to get in? Am I going to be able to get out? And if I’m able to get out, am I able to get out at a price that’s tight to NAV [net asset value], and what’s the infrastructure at your shop in terms of managing that consideration for me,” said Harrison.

Amplify ETFs’ Christian Magoon is also concerned about these newer ETF strategies could weather a monster drawdown. He listed private credit as a red flag.

“If your ETF owns private credit, I think it’s worth taking a look at, kind of what the standards are around liquidity and how that ETF is trading, because that should be a bit of a mismatch between the trading pace of ETFs and the underlying asset,” the firm’s CEO said in the same interview.

Magoon also highlighted potential issues surrounding equity-linked notes. The notes provide fixed income security while offering potentially higher returns linked to stocks or equity indexes.

“Those could potentially be in stress due to redemptions and the underlying credit risk. That’s another kind of unique derivative,” Magoon said. “I would very closely look at any ETF that has equity-linked notes should we get into a major drawdown or there be a contagion in private credit or something related to the banking system.”

Choose CNBC as your preferred source on Google and never miss a moment from the most trusted name in business news.

Continue Reading

Finance

Anthropic Mythos reveals ‘more vulnerabilities’ for cyberattacks

Published

on

Jamie Dimon, chief executive officer of JPMorgan Chase & Co., right, departs the US Capitol in Washington, DC, US, on Wednesday, Feb. 25, 2026.

Graeme Sloan | Bloomberg | Getty Images

JPMorgan Chase CEO Jamie Dimon said Tuesday that while artificial intelligence tools could eventually help companies defend themselves from cyberattacks, they are first making them more vulnerable.

Dimon said that JPMorgan was testing Anthropic’s latest model — the Mythos preview announced by the AI firm last week — as part of its broader effort to reap the benefits of AI while protecting against bad actors wielding the same technology.

“AI’s made it worse, it’s made it harder,” Dimon told analysts on the bank’s earnings call Tuesday morning. “It does create additional vulnerabilities, and maybe down the road, better ways to strengthen yourself too.”

When asked by a reporter about Mythos, Dimon seemed to refer to Anthropic’s warning that the model had already found thousands of vulnerabilities in corporate software.

“I think you read exactly what is it,” Dimon said. “It shows a lot more vulnerabilities need to be fixed.”

The remarks reveal how artificial intelligence, a technology welcomed by corporations as a productivity boon, has also morphed into a serious threat by giving bad actors new ways to hack into technology systems. Last week, Treasury Secretary Scott Bessent summoned bank CEOs to a meeting to discuss the risks posed by Mythos.

JPMorgan, the world’s largest bank by market cap, has for years invested heavily to stay ahead of threats, with dedicated teams and constant coordination with government agencies, Dimon said.

“We spend a lot of money. We’ve got top experts. We’re in constant contact with the government,” he said. “It’s a full-time job, and we’re doing it all the time.”

‘Attack mode’

Choose CNBC as your preferred source on Google and never miss a moment from the most trusted name in business news.

Continue Reading

Trending