Big technology companies are betting that a new wave of smaller, more precise AI models will be more effective when it comes to the needs of businesses in sectors like law, finance, and health care.
Despite bold statements, a lot of companies are failing to produce tangible results, according to Edward J Achtner, the head of generative AI for U.K. banking giant HSBC.
“Candidly, there’s a lot of success theater out there,” Achtner said on a panel at the CogX Global Leadership Summit alongside Ranil Boteju — a fellow AI leader at rival British bank Lloyds Banking Group — and Nathalie Oestmann, head of NV Ltd, an advisory firm for venture capital funds.
“We have to be very clinical in terms of what we choose to do, and where we choose to do it,” Achtner told attendees of the event, held at the Royal Albert Hall in London earlier this week.
Achtner outlined how the 150-year-old lending institution has embraced artificial intelligence since ChatGPT — the popular AI chatbot from Microsoft-backed startup OpenAI — burst onto the scene in November 2022.
The HSBC AI leader said that the bank has more than 550 use cases across its business lines and functions linked to AI — ranging from fighting money laundering and fraud using machine learning tools to supporting knowledge workers with newer generative AI systems.
One example he gave was a partnership that HSBC has in place with internet search titan Google on the use of AI technology anti-money laundering and fraud mitigation. That tie-up has been in place for several years, he said. The bank has also dipped its toes deeper into genAI tech much more recently.
“When it comes to generative artificial intelligence, we do need to clearly separate that” from other types of AI, Achtner said. “We do approach the underlying risk with respect to generative very differently because, while it represents incredible potential opportunity and productivity gains, it also represents a different type of risk.”
Achtner’s comments come as other figures in the financial services sector — particularly leaders at startup firms — have made bold statements about the level of overall efficiency gains and cost reductions they are seeing as a result of investments in AI.
Buy now, pay later firm Klarna says it has been taking advantage of AI to make up for loss of productivity resulting from declines in its workforce as employees move on from the company.
It is implementing a company-wide hiring freeze and has slashed overall employee headcount down to 3,800 from 5,000 — a roughly 24% workforce reduction — with the help of AI, CEO Sebastian Siemiatkowski said in August. He is looking to further reduce Klarna’s headcount to 2,000 staff members — without specifying a time for this target.
Klarna’s boss said the firm was lowering its overall headcount against the backdrop of AI’s potential to have “a dramatic impact” on jobs and society.
“I think politicians already today should consider whether there are other alternatives of how they could support people that may be effective,” he said at the time in an interview with the BBC. Siemiatkowski said it was “too simplistic” to say AI’s disruptive effects would be offset by the creation of new jobs thanks to AI.
Oestmann of NV Ltd, a London-based firm that offers advisory services for the C-suite of venture capital and private equity firms, directly touched on Klarna’s actions, saying headlines around such AI-driven workforce reductions are “not helpful.”
Klarna, she suggested, likely saw that AI “makes them a more valuable company” and was consequently incorporating the technology as part of plans to reduce its workforce anyway.
The result Klarna is seeing from AI “are very real,” a Klarna spokesperson told CNBC. “We publicize these results because we want to be honest and transparent about the impact genAI is having in the real world in companies today,” the spokesperson added.
“At the end of the day,” Oestmann added, as long as people are “trained appropriately” and banks and other financial services firm can “reinvent” themselves in the new AI era, “it will just help us to evolve.” She advised financial firms to pursue “continuous learning in everything that you do.”
“Make sure you are trying these tools out, make sure you are making this part of your everyday, make sure you are curious,” she added.
Boteju, chief data and analytics officer at Lloyds, pointed to three main use cases that the lender sees with respect to AI: automating back office functions like coding and engineering documentation, “human-in-the loop” uses like prompts for sales staff, and AI-generated responses to client queries.
Boteju stressed that Lloyds is “proceeding with caution” when it comes to exposing the bank’s customers to generative AI tools. “We want to get our guardrails in place before we actually start to scale those,” he added.
“Banks in particular have been using AI and machine learning for probably about 15 or 20 years,” Boteju said, signaling that machine learning, intelligent automation and chatbots are things traditional lenders have been “doing for a while.”
Generative AI, on the other hand, is a more nascent technology, according to the Lloyds exec. The bank is increasingly thinking about how to scale that technology — but by “using the current frameworks and infrastructure we’ve got,” rather than by moving the needle significantly.
Boteju and Achtner’s comments tally with what other AI leaders of financial services have said previously. Speaking with CNBC last week, Bahadir Yilmaz, chief analytics officer of ING, said that AI is unlikely to be as disruptive as firms like Klarna are suggesting with their public messaging.
“We see the same potential that they’re seeing,” Yilmaz said in an interview in London. “It’s just the tone of communication is a bit different.” He added that ING is primarily using AI in its global contact centers and internally for software engineering.
“We don’t need to be seen as an AI-driven bank,” Yilmaz said, adding that, with many processes lenders won’t even need AI to solve certain problems. “It’s a really powerful tool. It’s very disruptive. But we don’t necessarily have to say we are putting it as a sauce on all the food.”
Johan Tjarnberg, CEO of Swedish online payments firm Trustly, told CNBC earlier this week that AI “will actually be one of the biggest technology levers in payments.”But even so, he noted that the firm is focusing more of the “basics of AI” than on transformative changes like AI-led customer service.
One area where Trustly is looking to improve customer experience with AI is subscriptions. The startup is working on an “intelligent charging mechanism” that would aim to figure out the best time for a bank to take payment from a subscription platform user, based on their historical financial activity.
Tjarnberg added that Trustly is seeing closer to 5-10% improved efficiency as a result of implementing AI within its organization.
Against a volatile market backdrop, the software maker’s stock has gained 45% and is the best performer among companies valued at $5 billion or more, according to FactSet. The closest tech names are VeriSign, up 33%, Okta, up 30%, Robinhood, up 29%, and Uber, up 29%.
“When you think about macroeconomic concerns, you as a company need to be more efficient, and this is where Palantir thrives,” said Bank of America analyst Mariana Pérez Mora.
Palantir has set itself apart in the software world for its artificial-intelligence-enabled tools, gaining recognition for its defense and software contracts with key U.S. government agencies, including the military. In the fourth quarter, its government revenues jumped 45% year-over-year to $343 million.
Read more CNBC tech news
Companies have faced immense volatility in 2025 as tariffs threaten to jeopardize global supply chains and halt day-to-day manufacturing operations by hiking costs. Those fears have brought the broad market index down about 7% this year, while the tech-heavy Nasdaq Composite has slumped 11%.
At the same time, the Trump administration has clamped down on government spending, giving Tesla CEO Elon Musk‘s Department of Government Efficiency freedom to slash public sector costs. Some administration officials have touted shifting dollars from consulting contracts to commercial software providers like Palantir, said William Blair analyst Louie DiPalma.
“Palantir’s business model is highly aligned with the priorities of the Trump administration in terms of increasing agility and being very quick to market,” he said.
That’s put Palantir in the league with major contractors such as Lockheed Martin and Northrop Grumman, which have outperformed in this year’s downdraft. Many companies in the space are also looking to partner with the firm and tend to flock to defense during recessionary times, DiPalma said.
Stock Chart IconStock chart icon
Palantir vs. the Nasdaq Composite
CEO Alex Karp has also been a vocal supporter of American innovation and the company’s central role in helping prop up what he called the “single best tech scene in the world” during an interview with CNBC earlier this year. Karp also told CNBC that the U.S. needs an “all-country effort” to compete against emerging adversaries.
But the ride for Palantir has been far from smooth, and shares have been susceptible to volatile swings. Shares sold off nearly 14% during the week that Trump first announced tariffs. Shares rocketed 22% one day in February on strong earnings.
Its inclusion in more passive and quant funds over the years and the growing attention of retail traders has added to that turbulence, DiPalma said. Last year, the company joined both the S&P and Nasdaq. Palantir trades at one of the highest price-to-earnings multiples in software and last traded at 185 times earnings over the next twelve months. That puts a steep bar on the stock.
Kurt Sievers, chief executive officer of NXP Semiconductors NV, during the Federation of German Industries (BDI) conference in Berlin, Germany, on Monday, June 19, 2023.
NXP Semiconductor Inc. fell about 8% on Monday after the chip company announced that CEO Kurt Sievers will step down as part of its latest earnings.
Here’s how the company did, versus LSEG consensus estimates:
Earnings per share: $2.64 adjusted vs. $2.58 expected
Revenue: $2.84 billion vs. $2.83 billion expected
Sievers will retire at the end of the year, with Rafael Sotomayor stepping in as president on April 28, 2025.
The company beat expectations on the top and bottom lines but cited a “challenging set of market conditions” looking forward.
“We are operating in a very uncertain environment influenced by tariffs with volatile direct and indirect effects,” Sievers said in an earnings release.
Sales in NXP’s first quarter declined 9% year over year.
The company posted $1.67 billion in auto sales during the first quarter, trailing analyst estimates of $1.69 billion.
Read more CNBC tech news
NXP Semi said that second-quarter sales would come in at a midpoint of $2.9 billion, ahead of the $2.87 billion that analysts were projecting. Second-quarter adjusted EPS will be $2.66, in line with analyst estimates.
The company logged first-quarter net income of $490 million, which was a 23% year-to-year drop from $639 million.
NXP’s net income per share was $1.92 compared to $2.47 during the same time a year ago. A drop of 22%.
This is breaking news. Please refresh for updates.
Microsoft President Brad Smith speaks during signing ceremony of cooperation agreement between the Polish Ministry of Defence and Microsoft, in Warsaw, Poland, February 17, 2025.
Kacper Pempel | Reuters
The U.S. cannot afford to fall behind China in the race to a working quantum computer, Microsoft President Brad Smith wrote Monday.
President Donald Trump and the U.S. government need to prioritize funding for quantum research, or China could surpass the U.S., endangering economic competitiveness and security, Smith wrote.
“While most believe that the United States still holds the lead position, we cannot afford to rule out the possibility of a strategic surprise or that China may already be at parity with the United States,” Smith wrote. “Simply put, the United States cannot afford to fall behind, or worse, lose the race entirely.”
Microsoft’s position is the latest sign that research into quantum computing is starting to heat up among big tech companies and investors who are looking for the next technology that could rival the artificial intelligence boom.
Smith is calling for the Trump administration to increase funding for quantum research, renew the National Quantum Initiative Act and expand a program for testing quantum computers by the Defense Advanced Research Projects Agency, or DARPA. The Microsoft executive is also calling on the White House to expand the educational pipeline of people who have the math and science skills to work on quantum machines, fast-track immigration for Ph.D.s with quantum skills and for the government to buy more quantum-related computer parts to build a U.S. supply chain.
Microsoft did not detail how China surpassing the U.S. in quantum computing technology would endanger national security, but a National Security Agency official last year discussed what could happen if China or another adversary surprised the U.S. by building a quantum computer first.
The official, NSA Director of Research Gil Herrera, said that if such a “black swan” event happened, banks might not be able to keep transactions private because a quantum computer could crack their encryption, according to the Washington Times. A working quantum computer could also crack existing encrypted data that is usually shared publicly in a scrambled fashion, which could reveal secrets on U.S. nuclear weapon systems.
In February, Microsoft announced its latest quantum chip called Majorana, claiming that it invented a new kind of matter to develop the prototype device. Last year, Google announced Willow, a new device the company claimed was a “milestone” because it was able to correct errors and solve a math problem in five minutes that would have taken longer than the age of the universe on a traditional computer.
While the computers people are used to use bits that are either 0 or 1 to do calculations, quantum computers use “qubits,” which end up being on or off based on probability. Experts say that quantum computers will eventually be useful for problems with nearly infinite possibilities, such as simulating chemistry, or routing deliveries.
But the current quantum computers are far away from that point, and many computer industry participants say it could take decades for quantum computers to reach their potential.
Microsoft’s chip, Majorana, has eight qubits, but the company says it has a goal of least 1 million qubits for a commercially useful chip. Microsoft needs to build a device with a few hundred qubits before the company starts looking at whether it’s reliable enough for customers.