One of the most tempting bets on artificial intelligence is hiding in plain sight. The company is a household name, a technology powerhouse that has long revolved around cloud and AI. It also has the highest dividend yield in the tech sector.
While there’s a reason the market is skeptical of Big Blue’s position on AI, this omission is short-sighted. Shares of his chipmaker Nvidia (NYSE:NVDA) are up 172% this year on the success of chips used in generative artificial intelligence applications. AI software giants Microsoft (MSFT) and Alphabet (GOOGL) each grew 40%.
IBM? Down 8%.
This stunning decline comes as IBM (IBM) has rebuilt itself over the past three years under the steadfast leadership of CEO Arvind Krishna. IBM may have more AI savvy than almost any other company, but it’s still less than doubling its stock’s expected revenue, while Nvidia is nearly 20 times more. IBM also has a dividend yield of over 5%.
Ads – scroll to continue
The company has been working on AI applications for at least 40 years. In his 1997, two years before Nvidia went public, his IBM supercomputer called Deep Blue defeated chess world champion Garry Kasparov in his six matches. In 2011, IBM’s Watson supercomputer famously defeated a human champion. dangerous. Among Watson’s antagonists was the legendary Ken Jennings, who is now the show’s host. “For my part, I welcome our new computer champion,” Jennings wrote in his final report. dangerous and answer.
Yes, IBM made a mistake. After Jeopardy’s stunt, the company made a big push to use Watson in healthcare, drug discovery, and other applications. It never worked, and IBM reportedly sold its Watson Health business for $1 billion in 2022. Some people interpret the sale of Watson Health as IBM giving up on AI, but that’s not the case at all.
Last week, I had a lengthy conversation with Krishna about the company’s approach to the AI business. He had a lot to say.
First, IBM recently launched a completely new version of Watson called Watson X. This new product has his three parts. Watson.ai works with customers to create new models, or data sets. Watson.data acts as a data store, making the company compete with his Snowflake
Ads – scroll to continue
(SNOW) etc. Watson.governance also monitors AI models to ensure they are accurate, accountable, and do not contain false or offensive information.
What IBM is trying to do is create the next ChatGPT. “Public models are incredibly powerful,” says Krishna. “What is Google, what is Facebook?”
,
What Microsoft is doing fits that mode perfectly. They are building a very large model that works for everyone. ”
But Krishna believes consumer AI applications are only a small part of the opportunity. “It’s like an iceberg,” he says of the situation where chatbots like Microsoft Bing and Google Bard are above the surface. “There are more and more use cases that do not benefit from the large-scale public model.”
Ads – scroll to continue
IBM’s strategy is to help customers create their own AI applications and squeeze more value out of their data. In some cases, the company combines open source models with proprietary data. For some of our customers, IBM has built a private model just for their data.
While IBM has no plans to build general large-scale models like ChatGPT, Krishna said it is building a family of domain-specific datasets. For example, Public He creates a chemical model based on information in the domain.
“Take one of our partners in the chemical industry, Dow, Mitsui Chemicals, BASF, for example, they have their own data on how chemicals are manufactured,” he says. “Will some of them incorporate their own data into the public model? We can provide faster answers to internal questions and come up with new formulations, which will accelerate their business model.”
Ads – scroll to continue
In one case, Krishna said, IBM is working with banks to create better compliance and audit data for its own employees, one of the goals being to ensure proper controls are in place. to the regulatory authority.
IBM has built 20 of these domain models. The list is not only for chemistry and banking, but also for writing code and making IT operations more efficient. To improve weather modeling, there are geospatial models that IBM is adapting to combine with NASA’s climate data. The company, which owns open source software giant Red Hat, does not shy away from using an open source training model. IBM, in particular, has partnered with Hugging Face, an AI startup with a library of 130,000 models.
A complicating factor for investors is that IBM doesn’t disclose how much of its revenue is related to AI. Krishna says it is impossible to calculate the number. “Our mainframe has an AI circuit. So is the mainframe business AI? In the future, we plan to use AI for storage backup. AI is in action. All cybersecurity is going to use AI. By the end of the next five years, AI will be in everything.”
Ads – scroll to continue
Krishna cites PwC estimates that AI could create $16 trillion in value to the economy by 2030. And Krishna has long held the view that IBM will grow his mid-single-digit earnings over the long term.
What the CEO doesn’t say directly is that AI could open up new opportunities for IBM and make more of a contribution to revenue. “It’s a potential facilitator,” he says.
Investors should read between the lines.
write destination Eric J. Savitz (eric.savitz@barrons.com)
