Connect with us

Published

on

Dado Ruvic | Reuters

Chinese artificial intelligence firm DeepSeek rocked markets this week with claims its new AI model outperforms OpenAI’s and cost a fraction of the price to build.

The assertions — specifically that DeepSeek’s large language model cost just $5.6 million to train — have sparked concerns over the eyewatering sums that tech giants are currently spending on computing infrastructure required to train and run advanced AI workloads.

But not everyone is convinced by DeepSeek’s claims.

CNBC asked industry experts for their views on DeepSeek, and how it actually compares to OpenAI, creator of viral chatbot ChatGPT which sparked the AI revolution.

What is DeepSeek?

Last week, DeepSeek released R1, its new reasoning model that rivals OpenAI’s o1. A reasoning model is a large language model that breaks prompts down into smaller pieces and considers multiple approaches before generating a response. It is designed to process complex problems in a similar way to humans.

DeepSeek was founded in 2023 by Liang Wenfeng, co-founder of AI-focused quantitative hedge fund High-Flyer, to focus on large language models and reaching artificial general intelligence, or AGI.

AGI as a concept loosely refers to the idea of an AI that equals or surpasses human intellect on a wide range of tasks.

Much of the technology behind R1 isn’t new. What is notable, however, is that DeepSeek is the first to deploy it in a high-performing AI model with — according to the company — considerable reductions in power requirements.

“The takeaway is that there are many possibilities to develop this industry. The high-end chip/capital intensive way is one technological approach,” said Xiaomeng Lu, director of Eurasia Group’s geo-technology practice.

“But DeepSeek proves we are still in the nascent stage of AI development and the path established by OpenAI may not be the only route to highly capable AI.” 

How is it different from OpenAI?

Read more DeepSeek coverage

In a technical report, the company said its V3 model had a training cost of only $5.6 million — a fraction of the billions of dollars that notable Western AI labs such as OpenAI and Anthropic have spent to train and run their foundational AI models. It isn’t yet clear how much DeepSeek costs to run, however.

If the training costs are accurate, though, it means the model was developed at a fraction of the cost of rival models by OpenAI, Anthropic, Google and others.

Daniel Newman, CEO of tech insight firm The Futurum Group, said these developments suggest “a massive breakthrough,” although he shed some doubt on the exact figures.

“I believe the breakthroughs of DeepSeek indicate a meaningful inflection for scaling laws and are a real necessity,” he said. “Having said that, there are still a lot of questions and uncertainties around the full picture of costs as it pertains to the development of DeepSeek.”

Meanwhile, Paul Triolio, senior VP for China and technology policy lead at advisory firm DGA Group, noted it was difficult to draw a direct comparison between DeepSeek’s model cost and that of major U.S. developers.

“The 5.6 million figure for DeepSeek V3 was just for one training run, and the company stressed that this did not represent the overall cost of R&D to develop the model,” he said. “The overall cost then was likely significantly higher, but still lower than the amount spent by major US AI companies.” 

DeepSeek wasn’t immediately available for comment when contacted by CNBC.

Comparing DeepSeek, OpenAI on price

DeepSeek and OpenAI both disclose pricing for their models’ computations on their websites.

DeepSeek says R1 costs 55 cents per 1 million tokens of inputs — “tokens” referring to each individual unit of text processed by the model — and $2.19 per 1 million tokens of output.

In comparison, OpenAI’s pricing page for o1 shows the firm charges $15 per 1 million input tokens and $60 per 1 million output tokens. For GPT-4o mini, OpenAI’s smaller, low-cost language model, the firm charges 15 cents per 1 million input tokens.

Skepticism over chips

LinkedIn co-founder Reid Hoffman: DeepSeek AI proves this is now a 'game-on competition' with China

Nvidia has since come out and said that the GPUs that DeepSeek used were fully export-compliant.

The real deal or not?

Industry experts seem to broadly agree that what DeepSeek has achieved is impressive, although some have urged skepticism over some of the Chinese company’s claims.

“DeepSeek is legitimately impressive, but the level of hysteria is an indictment of so many,” U.S. entrepreneur Palmer Luckey, who founded Oculus and Anduril wrote on X.

“The $5M number is bogus. It is pushed by a Chinese hedge fund to slow investment in American AI startups, service their own shorts against American titans like Nvidia, and hide sanction evasion.”

Seena Rejal, chief commercial officer of NetMind, a London-headquartered startup that offers access to DeepSeek’s AI models via a distributed GPU network, said he saw no reason not to believe DeepSeek.

“Even if it’s off by a certain factor, it still is coming in as greatly efficient,” Rejal told CNBC in a phone interview earlier this week. “The logic of what they’ve explained is very sensible.”

However, some have claimed DeepSeek’s technology might not have been built from scratch.

“DeepSeek makes the same mistakes O1 makes, a strong indication the technology was ripped off,” billionaire investor Vinod Khosla said on X, without giving more details.

It’s a claim that OpenAI itself has alluded to, telling CNBC in a statement Wednesday that it is reviewing reports DeepSeek may have “inappropriately” used output data from its models to develop their AI model, a method referred to as “distillation.”

“We take aggressive, proactive countermeasures to protect our technology and will continue working closely with the U.S. government to protect the most capable models being built here,” an OpenAI spokesperson told CNBC.

Commoditization of AI

However the scrutiny surrounding DeepSeek shakes out, AI scientists broadly agree it marks a positive step for the industry.

Yann LeCun, chief AI scientist at Meta, said that DeepSeek’s success represented a victory for open-source AI models, not necessarily a win for China over the U.S. Meta is behind a popular open-source AI model called Llama.

“To people who see the performance of DeepSeek and think: ‘China is surpassing the US in AI.’ You are reading this wrong. The correct reading is: ‘Open source models are surpassing proprietary ones’,” he said in a post on LinkedIn.

“DeepSeek has profited from open research and open source (e.g. PyTorch and Llama from Meta). They came up with new ideas and built them on top of other people’s work. Because their work is published and open source, everyone can profit from it. That is the power of open research and open source.”

WATCH: Why DeepSeek is putting America’s AI lead in jeopardy

Why China's DeepSeek is putting America's AI lead in jeopardy

– CNBC’s Katrina Bishop and Hayden Field contributed to this report

Continue Reading

Technology

Apple scores big victory with ‘F1,’ but AI is still a major problem in Cupertino

Published

on

By

Apple scores big victory with 'F1,' but AI is still a major problem in Cupertino

Formula One F1 – United States Grand Prix – Circuit of the Americas, Austin, Texas, U.S. – October 23, 2022 Tim Cook waves the chequered flag to the race winner Red Bull’s Max Verstappen 

Mike Segar | Reuters

Apple had two major launches last month. They couldn’t have been more different.

First, Apple revealed some of the artificial intelligence advancements it had been working on in the past year when it released developer versions of its operating systems to muted applause at its annual developer’s conference, WWDC. Then, at the end of the month, Apple hit the red carpet as its first true blockbuster movie, “F1,” debuted to over $155 million — and glowing reviews — in its first weekend.

While “F1” was a victory lap for Apple, highlighting the strength of its long-term outlook, the growth of its services business and its ability to tap into culture, Wall Street’s reaction to the company’s AI announcements at WWDC suggest there’s some trouble underneath the hood.

“F1” showed Apple at its best — in particular, its ability to invest in new, long-term projects. When Apple TV+ launched in 2019, it had only a handful of original shows and one movie, a film festival darling called “Hala” that didn’t even share its box office revenue.

Despite Apple TV+ being written off as a costly side-project, Apple stuck with its plan over the years, expanding its staff and operation in Culver City, California. That allowed the company to build up Hollywood connections, especially for TV shows, and build an entertainment track record. Now, an Apple Original can lead the box office on a summer weekend, the prime season for blockbuster films.

The success of “F1” also highlights Apple’s significant marketing machine and ability to get big-name talent to appear with its leadership. Apple pulled out all the stops to market the movie, including using its Wallet app to send a push notification with a discount for tickets to the film. To promote “F1,” Cook appeared with movie star Brad Pitt at an Apple store in New York and posted a video with actual F1 racer Lewis Hamilton, who was one of the film’s producers.

(L-R) Brad Pitt, Lewis Hamilton, Tim Cook, and Damson Idris attend the World Premiere of “F1: The Movie” in Times Square on June 16, 2025 in New York City.

Jamie Mccarthy | Getty Images Entertainment | Getty Images

Although Apple services chief Eddy Cue said in a recent interview that Apple needs the its film business to be profitable to “continue to do great things,” “F1” isn’t just about the bottom line for the company.

Apple’s Hollywood productions are perhaps the most prominent face of the company’s services business, a profit engine that has been an investor favorite since the iPhone maker started highlighting the division in 2016.

Films will only ever be a small fraction of the services unit, which also includes payments, iCloud subscriptions, magazine bundles, Apple Music, game bundles, warranties, fees related to digital payments and ad sales. Plus, even the biggest box office smashes would be small on Apple’s scale — the company does over $1 billion in sales on average every day.

But movies are the only services component that can get celebrities like Pitt or George Clooney to appear next to an Apple logo — and the success of “F1” means that Apple could do more big popcorn films in the future.

“Nothing breeds success or inspires future investment like a current success,” said Comscore senior media analyst Paul Dergarabedian.

But if “F1” is a sign that Apple’s services business is in full throttle, the company’s AI struggles are a “check engine” light that won’t turn off.

Replacing Siri’s engine

At WWDC last month, Wall Street was eager to hear about the company’s plans for Apple Intelligence, its suite of AI features that it first revealed in 2024. Apple Intelligence, which is a key tenet of the company’s hardware products, had a rollout marred by delays and underwhelming features.

Apple spent most of WWDC going over smaller machine learning features, but did not reveal what investors and consumers increasingly want: A sophisticated Siri that can converse fluidly and get stuff done, like making a restaurant reservation. In the age of OpenAI’s ChatGPT, Anthropic’s Claude and Google’s Gemini, the expectation of AI assistants among consumers is growing beyond “Siri, how’s the weather?”

The company had previewed a significantly improved Siri in the summer of 2024, but earlier this year, those features were delayed to sometime in 2026. At WWDC, Apple didn’t offer any updates about the improved Siri beyond that the company was “continuing its work to deliver” the features in the “coming year.” Some observers reduced their expectations for Apple’s AI after the conference.

“Current expectations for Apple Intelligence to kickstart a super upgrade cycle are too high, in our view,” wrote Jefferies analysts this week.

Siri should be an example of how Apple’s ability to improve products and projects over the long-term makes it tough to compete with.

It beat nearly every other voice assistant to market when it first debuted on iPhones in 2011. Fourteen years later, Siri remains essentially the same one-off, rigid, question-and-answer system that struggles with open-ended questions and dates, even after the invention in recent years of sophisticated voice bots based on generative AI technology that can hold a conversation.

Apple’s strongest rivals, including Android parent Google, have done way more to integrate sophisticated AI assistants into their devices than Apple has. And Google doesn’t have the same reflex against collecting data and cloud processing as privacy-obsessed Apple.

Some analysts have said they believe Apple has a few years before the company’s lack of competitive AI features will start to show up in device sales, given the company’s large installed base and high customer loyalty. But Apple can’t get lapped before it re-enters the race, and its former design guru Jony Ive is now working on new hardware with OpenAI, ramping up the pressure in Cupertino.

“The three-year problem, which is within an investment time frame, is that Android is racing ahead,” Needham senior internet analyst Laura Martin said on CNBC this week.

Apple’s services success with projects like “F1” is an example of what the company can do when it sets clear goals in public and then executes them over extended time-frames.

Its AI strategy could use a similar long-term plan, as customers and investors wonder when Apple will fully embrace the technology that has captivated Silicon Valley.

Wall Street’s anxiety over Apple’s AI struggles was evident this week after Bloomberg reported that Apple was considering replacing Siri’s engine with Anthropic or OpenAI’s technology, as opposed to its own foundation models.

The move, if it were to happen, would contradict one of Apple’s most important strategies in the Cook era: Apple wants to own its core technologies, like the touchscreen, processor, modem and maps software, not buy them from suppliers.

Using external technology would be an admission that Apple Foundation Models aren’t good enough yet for what the company wants to do with Siri.

“They’ve fallen farther and farther behind, and they need to supercharge their generative AI efforts” Martin said. “They can’t do that internally.”

Apple might even pay billions for the use of Anthropic’s AI software, according to the Bloomberg report. If Apple were to pay for AI, it would be a reversal from current services deals, like the search deal with Alphabet where the Cupertino company gets paid $20 billion per year to push iPhone traffic to Google Search.

The company didn’t confirm the report and declined comment, but Wall Street welcomed the report and Apple shares rose.

In the world of AI in Silicon Valley, signing bonuses for the kinds of engineers that can develop new models can range up to $100 million, according to OpenAI CEO Sam Altman.

“I can’t see Apple doing that,” Martin said.

Earlier this week, Meta CEO Mark Zuckerberg sent a memo bragging about hiring 11 AI experts from companies such as OpenAI, Anthropic, and Google’s DeepMind. That came after Zuckerberg hired Scale AI CEO Alexandr Wang to lead a new AI division as part of a $14.3 billion deal.

Meta’s not the only company to spend hundreds of millions on AI celebrities to get them in the building. Google spent big to hire away the founders of Character.AI, Microsoft got its AI leader by striking a deal with Inflection and Amazon hired the executive team of Adept to bulk up its AI roster.

Apple, on the other hand, hasn’t announced any big AI hires in recent years. While Cook rubs shoulders with Pitt, the actual race may be passing Apple by.

WATCH: Jefferies upgrades Apple to ‘Hold’

Jefferies upgrades Apple to 'Hold'

Continue Reading

Technology

Musk backs Sen. Paul’s criticism of Trump’s megabill in first comment since it passed

Published

on

By

Musk backs Sen. Paul's criticism of Trump's megabill in first comment since it passed

Tesla CEO Elon Musk speaks alongside U.S. President Donald Trump to reporters in the Oval Office of the White House on May 30, 2025 in Washington, DC.

Kevin Dietsch | Getty Images

Tesla CEO Elon Musk, who bombarded President Donald Trump‘s signature spending bill for weeks, on Friday made his first comments since the legislation passed.

Musk backed a post on X by Sen. Rand Paul, R-Ky., who said the bill’s budget “explodes the deficit” and continues a pattern of “short-term politicking over long-term sustainability.”

The House of Representatives narrowly passed the One Big Beautiful Bill Act on Thursday, sending it to Trump to sign into law.

Paul and Musk have been vocal opponents of Trump’s tax and spending bill, and repeatedly called out the potential for the spending package to increase the national debt.

On Monday, Musk called it the “DEBT SLAVERY bill.”

The independent Congressional Budget Office has said the bill could add $3.4 trillion to the $36.2 trillion of U.S. debt over the next decade. The White House has labeled the agency as “partisan” and continuously refuted the CBO’s estimates.

Read more CNBC tech news

The bill includes trillions of dollars in tax cuts, increased spending for immigration enforcement and large cuts to funding for Medicaid and other programs.

It also cuts tax credits and support for solar and wind energy and electric vehicles, a particularly sore spot for Musk, who has several companies that benefit from the programs.

“I took away his EV Mandate that forced everyone to buy Electric Cars that nobody else wanted (that he knew for months I was going to do!), and he just went CRAZY!” Trump wrote in a social media post in early June as the pair traded insults and threats.

Shares of Tesla plummeted as the feud intensified, with the company losing $152 billion in market cap on June 5 and putting the company below $1 trillion in value. The stock has largely rebounded since, but is still below where it was trading before the ruckus with Trump.

Stock Chart IconStock chart icon

hide content

Tesla one-month stock chart.

— CNBC’s Kevin Breuninger and Erin Doherty contributed to this article.

Continue Reading

Technology

Microsoft layoffs hit 830 workers in home state of Washington

Published

on

By

Microsoft layoffs hit 830 workers in home state of Washington

Microsoft CEO Satya Nadella speaks at the Axel Springer building in Berlin on Oct. 17, 2023. He received the annual Axel Springer Award.

Ben Kriemann | Getty Images

Among the thousands of Microsoft employees who lost their jobs in the cutbacks announced this week were 830 staffers in the company’s home state of Washington.

Nearly a dozen game design workers in the state were part of the layoffs, along with three audio designers, two mechanical engineers, one optical engineer and one lab technician, according to a document Microsoft submitted to Washington employment officials.

There were also five individual contributors and one manager at the Microsoft Research division in the cuts, as well as 10 lawyers and six hardware engineers, the document shows.

Microsoft announced plans on Wednesday to eliminate 9,000 jobs, as part of an effort to eliminate redundancy and to encourage employees to focus on more meaningful work by adopting new technologies, a person familiar with the matter told CNBC. The person asked not to be named while discussing private matters.

Scores of Microsoft salespeople and video game developers have since come forward on social media to announce their departure. In April, Microsoft said revenue from Xbox content and services grew 8%, trailing overall growth of 13%.

In sales, the company parted ways with 16 customer success account management staff members based in Washington, 28 in sales strategy enablement and another five in sales compensation. One Washington-based government affairs worker was also laid off.

Microsoft eliminated 17 jobs in cloud solution architecture in the state, according to the document. The company’s fastest revenue growth comes from Azure and other cloud services that customers buy based on usage.

CEO Satya Nadella has not publicly commented on the layoffs, and Microsoft didn’t immediately provide a comment about the cuts in Washington. On a conference call with analysts in April, Microsoft CFO Amy Hood said the company had a “focus on cost efficiencies” during the March quarter.

WATCH: Microsoft layoffs not performance-based, largely targeting middle managers

Microsoft layoffs not performance-based, largely targeting middle managers

Continue Reading

Trending