Skip to main content
Faculty Viewpoints

How AI Is Already Transforming Fortune 500 Businesses, According to Their CEOs

At a recent Yale CEO Summit, Prof. Jeffrey Sonnenfeld talked to business leaders about the AI tools and other new technologies appearing everywhere from back offices to fast-food kitchens. Sonnenfeld and co-author Steven Tian outline the looming changes in a variety of sectors.

A Chipotle chef with Chippy, an autonomous kitchen assistant that makes tortilla chips

A Chipotle chef with Chippy, an autonomous kitchen assistant that makes tortilla chips

Chipotle
  • Jeffrey A. Sonnenfeld
    Senior Associate Dean for Leadership Studies & Lester Crown Professor in the Practice of Management
  • Steven Tian
    Director of Research, Chief Executive Leadership Institute

At a recent Yale CEO Summit we convened online to mark the 50th anniversary of the invention of the internet in 1974, top Fortune 500 CEOs revealed how they are reinventing their businesses around artificial intelligence (AI), clean energy, and other emerging technologies.

The most tangible, impactful implementations of AI are often taking place in traditional business sectors. As one of our speakers, Steve Case, the CEO of Revolution Growth, put it: “We’re seeing a transition from AI being a big horizontal platform, the large language models centered in Silicon Valley, to more vertical AI deployed in industry verticals, which creates an opportunity all across the country, from factory floors in Ohio to ag tech in the Midwest to biotech labs.”

However, this transformative process is rife with business, regulatory, and technical challenges that must be carefully managed. Here are some insights from our Yale CEO Summit that stood out about how top CEOs say they are adapting their business models around AI, digital disruption, and emerging technologies.

A transformation at every layer of the consumer market

Chipotle CEO Brian Niccol discussed how the pioneering fast-casual restaurant chain is using automation and AI to alleviate pain points for their employees: “We’ve heard feedback from our team members that the jobs they don’t enjoy doing include frying chips and cutting, coring, and scooping avocadoes, so we’ve been finding automated solutions that can take care of that prep in a consistent way.”

One example is Chippy the robot, which makes tortilla chips. Rather than replacing human workers, Niccol sees these robots as complimentary: “One of our biggest challenges is getting prep done on time for opening every day, especially if someone calls off in the morning and the team is short a person.”

At Kroger’s, one of the nation’s largest grocery chains, CEO Rodney McMullen showed us how AI modeling has helped reduce checkout times by 50%, with digital twins synchronized with store layouts and traffic allowing for the visualization, simulation, and optimization of self-checkout lines with cashier check-out lines—especially helpful since Kroger is facing 18,000 vacant positions amidst a labor shortage.

Moving from food services to hospitality and travel, Glenn Fogel, the CEO of Booking Holdings, said that he is already “seeing significant improvements” in how AI is personalizing travel recommendations and tailoring travel experiences to better fit customers’ needs. “Instead of looking at brochure after brochure, trying to see what interests our customers, AI can offer an even better match with what our customers want to do. It knows your preferences, it doesn’t forget what your preferences are, and it’s all in the database.”

Similarly, Greg Maffei, the CEO of Liberty Media, which owns a significant stake in TripAdvisor, recently announced a new partnership with OpenAI and launched a new AI travel itinerary generator. That product, which is currently in public beta testing, uses AI to filter through TripAdvisor’s more than one billion user-generated reviews to create AI-generated itineraries with suggested hotels and activities based on a customer’s stated interests and preferences.

American Airlines Chair and former CEO Doug Parker shared that many airlines are similarly using AI to streamline the flight reservation process, but cautioned that there are limits to what AI can do. “Nobody seems comfortable with positions such as mechanics and flight attendants being totally automated away.”

Coca-Cola CEO James Quincey eloquently synthesized how he sees the applicability of AI to consumer-facing companies: “I would break it into three layers, and with each layer, it becomes slightly less clear where it’ll end up. The obvious layer is how AI can help the internal workings of the organization. How AI can support and enable people internally, with call centers and client interactions, is relatively straightforward and obvious. The second layer, which is slightly more difficult but emerging with greater clarity, is how AI can help the sales force and help us be better partners to the retailers… The third level, which is least clear, is what we can do with generative AI and marketing. We’ve made available the Coke media assets, letting people make new images, putting them on electronic billboards.”

Getting your money’s worth from technology—perspectives from global finance

Within financial services, several CEOs shared how AI is helping transform how clients interact with their financial institutions. Brian Moynihan, CEO of Bank of America, explained the thinking that went into their proprietary “Erica” voice-activated banking assistant which has helped customers in over 2 billion interactions.

“First, we had to build a language which was recognized for banking, and then build a structure which works in a controlled, regulated environment, which is what financial services is. Then, we had to think about how we could deliver our product to 60 million customers, keeping in mind that those customers would be asking specific questions related to a transaction in one of 60 million accounts, as a part of 110 different systems and types of transactions… Take a question such as, ‘What is my balance,’ which sounds remarkably simple. But if you put it into a query ten years ago, you would’ve gotten answers ranging from a picture to a scale to yoga class imitations. So we had to build a bespoke banking language. It is not ChatGPT, but it is a core application which is a roadmap for ways we can use AI technology, first in the consumer business, and now in the institutional business… And we’re adding future functionality such as putting it into an institutional setting with Cashpro. And of course, we’ve been focused on keeping it accountable and verifiable, in the context that we are liable for the outcome of a question, irrespective of what device was used to answer a question.”

The theme of using AI to streamline customer service and interactions with customers was also emphasized by Affirm CEO Max Levchin. “I think one of the overlooked opportunities in AI is that all of us in financial services have a lot of customers who are asking for help or struggling to understand something… We found an extraordinarily useful application was before a customer is directed to a human—and we do guarantee that they can get through to a human—we filter their inbound request through an AI chatbot, which will help our customer service team understand what the customer is asking for.”

Driving smart vehicles into an electric, autonomous, digital future

General Motors CEO Mary Barra explained that the cyclical downturn in EV demand does not change the fact that the EV transformation will continue to accelerate as reliable recharging infrastructure is built. “The EV transformation is ongoing, and we’ll get there as we drive greater affordability and range, give customers choice, and develop affordable and accessible charging infrastructure. But even more significant is the fact that the vehicle is becoming a software platform, and the ultimate piece of that is autonomy. Ninety-nine percent of fatalities are caused by human error; autonomous technology will help us eliminate that.”

Nick Pinchuk, the CEO of Snap-On, which services about 600,000 repair shops worldwide, discussed how the company has added data center functionality with its proprietary ShopKey Pro software on top of its traditional tools business: “Cars have progressed from dozens of trouble codes in the 90s, to tens of thousands of trouble codes today. Today, to diagnose a car, you hook a laptop to it or hook it wirelessly, to read the algorithms of what the car is really saying. To interpret that, you need databases. We have databases with billions of car records, actual repair events, and from that, AI and machine learning will draw out even more efficient repair lessons moving forward.”

In addition to the high-tech auto repair software business flourishing, the EV boom has also driven a need for more hard tools, Pinchuk explained. “We make tools for all vehicle repairs. The geometries of the cars themselves have become far more complex. We love it when the EVs roll into the garages, because that means the repair shop needs tools for both EVs and internal combustion engines, as well as plug-in hybrids.”

Electrification and autonomy are trends that extend far beyond just consumer vehicles. “We’ve electrified a lot of vehicles that just a couple of years ago, people thought weren’t electrifiable,” specialty truck builder Oshkosh CEO John Pfeifer said. “Things like a 80,000-pound airport rescue and firefighting vehicle, and a 40,000-pound municipal fire truck, and we’re moving on from there… Every vehicle we send out today is completely connected for communications in real time, so our customers know exactly where it’s functioning and what’s going on with the vehicle, and we’re continuing to develop these software services to make these vehicles more effective.”

Michael Happe, the CEO of Winnebago Industries, said that the leading RV company has introduced concept electric RVs into the market for some testing by customers. “The other thing that’s quickly happening is the digital transformation around the outdoor lifestyle experience. We have developed the ability to connect consumers’ phones to systems on our product, so they can operate them remotely. So when they get back from hiking or camping, that vehicle is cooled down.”

At AGCO, one of the leading tractor makers, CEO Eric Hansotia has transformed the company into a provider of smart-farming software solutions. “Almost all of our machines now have onboard compute sensors that can understand variations in the soil and crop. One of the features we’re coming out with now is the ability to have vision systems on the sprayer, so it can identify the difference between a weed and a crop, and have the sprayer hit only the weed, and hit with the right kind and amount of chemicals.”

Connecting national defense

At aerospace and defense giant Lockheed Martin, many of the digital transformations described above have already taken place. In an exchange between CNBC anchor Morgan Brennan and Lockheed Martin CEO Jim Taiclet, Brennan shared that not only did she ride in one of the first autonomous Black Hawk helicopters a few months ago, but as far back as 2017, she was impressed to see the use of digital twins when she visited the F-35 factory floor, while others are just now starting to use digital twins in their businesses.

Taiclet is now aiming even higher, not only transforming advanced manufacturing but also ambitious connecting national defense towards its “5G, Internet-of-Things future.”

“There is a huge opportunity to bring the imagination and potential of the internet to national defense. How do we connect the devices, the core network we need, digital technology, software, etc. to be more effective at deterring armed conflict in the 21st century? Our goal has been to advocate for and implement, at least at our company, a 5G, Internet of Things, open architecture approach to everything we do. We want to broaden that out all the way across the entire aerospace and defense industry, for all of our colleagues to work off of, and the way to get there is to have a standards body like 3GPP in the telecoms sector where we work off the same protocols, APIs, interfaces, frequencies, etc.”

Ripple effects on infrastructure and energy

“It takes half a liter of water every time you ask a question on an AI chatbot, and it uses ridiculous amounts of energy,” tech venture capitalist Roger McNamee reminded us.

The point that many underestimate the scale to which AI will require greater infrastructure and energy investments was further buttressed by Silver Lake Founder Glenn Hutchins. “I think people are spending too much time thinking about the models, and not enough time understanding the scale and complexity of the infrastructure we’re going to need to get all of this done. It’s not just about securing GPUs from Nvidia. It’s about finding the location for data centers, equipping them with power, and protecting their data. The biggest pinch point of all may be energy. Just look at how the stocks of unregulated nuclear facilities have taken off.”

Lynn Good, the CEO of Duke Energy, shared insights about how one of the nation’s largest utilities is thinking about this challenge. “We are seeing growth in demand across the Southeast, from a broad diversifying range of sources. Growth has become the story of the utilities industry, and the market is realizing that. There’s data centers, some of which are foundational cloud computing data centers, but increasingly a lot of data centers are from AI. We have chipmakers, battery manufacturers, and pharma companies growing… And it does matter to us what kind of energy user you are… So, we are embracing partnership, we recently announced an MOU with Amazon, Microsoft, and Google as well as another large industrial customer with clean energy aspirations, to find ways to partner.”

Mike Kasbar, the CEO of World Kinect, one of the nation’s largest fuel services companies, put it all in perspective: “We’re in the dot-com period of sustainability, and on top of that, we have the remarkable confluence between energy and compute technology companies right now. Energy is the antithesis of GLP1 drugs: With energy, the more you consume, the hungrier you get. And when you look at the connection between technology and energy, it’s exponential.”

Of course, AI requires digital infrastructure in addition to physical infrastructure. Michael Dell, CEO of Dell Technologies, pointed out that “as we’re moving from calculating and computing towards cognition, that’s a big change for the industry since we’ve been doing calculating for 50 years. What do you need for this transition? You need more data, you need more compute, you need more memory, you need more data storage, you need more servers, you need more networking. These are all things we do, so our company is built for this moment”.

Revolutionizing drug discovery and healthcare

Jack Hidary, the CEO of SandboxAQ, put into perspective the opportunity arising from AI within healthcare: “To get to the next level, you need the next generation of AI, which is not trained on random stuff from the internet, but what actually happens in the real world… Using Nvidia technology, and CUDA, and our software at SandboxAQ, we can make a digital twin of these drug candidates, and then we can run millions and then billions of simulations using this hardware-software combination. That will reduce the time it takes to get lifesaving drugs. You still have to do clinical trials, no shortcuts around that, but the stimulations you can run from the digital twins can streamline drug discovery.”

CEO Brian Tyler of McKesson also joined us at our meeting. McKesson, which rather remarkably distributes a third of the total pharmaceutical volume in the U.S., has also been helping train machine learning models using an analytics-powered healthcare map to help doctors diagnose illnesses with more precision, as well as using robotics to process and pack orders.

Fortifying public trust and national security in a digital world

Several participants repeatedly cited comments by former IBM CEO Ginni Rometty about fortifying the role of trust in a digital world. ”It is the same issue for the internet, for data privacy, for AI, and for quantum computing… The issue we have right now, to build trust, you have to address the upside of technology and in parallel, the downside, at the same time… If you’re clear on who owns the technology and the data, and who gets the benefits, and things are transparent, explainable, and free of bias, that’s what we need.”

Deputy National Security Advisor for Cyber Anne Neuberger shared key lessons for regulating AI from her experience in cybersecurity. “In cyber, just as there is in AI, most of the critical infrastructure is owned and operated by the private sector, and connected to the internet in ways that are not secure enough… First, we need to make sure that any applications of AI are built more securely, to guard against attacks that range from poisoning the actual algorithms to hacking the models, stealing the models, or hacking the data upon which they were trained. And second, we need to ensure that before AI is used across our critical infrastructure such as water, pipelines, and railways, we’ve built-in protections like transparency on what data they’re trained on, adequate red teaming of models, keeping a human in the loop on key decisions, and ensuring that before operational systems are connected to AI models, we’ve tested them and built-in guardrails as well.”

Former Homeland Security advisor Tom Bossert expressed concern that “the effectiveness and cost of the regulations are imposed on businesses using the internet, not on the providers or developers or operating systems… I see compliance costs continuing to grow and I don’t see them translating into greater security results.”

Chris Krebs, the founding director of the Cybersecurity and Infrastructure Security Agency, struck a more optimistic tone: “While this regulatory and enforcement challenge will not be solved overnight, the policy arc has been fairly consistent and bipartisan. Krebs also expressed optimism that “AI-enabled defenses are outstripping the capabilities of adversaries to attack our critical infrastructure, in large part thanks to capital markets, innovation, and the investments companies are putting into it.”

The theme of public-private partnerships was reinforced by Salesforce CEO Marc Benioff, who said “There has to be even greater partnership between the tech industry and government.”

Greg Brown, the CEO of Motorola, a leader in providing and provisioning private communications networks for public safety and first responders in North America and around the world, across 911 software, call dispatching, call handling, records and evidentiary management; and all things video, fixed license plate recognition, and mobile video, shared his vision of striking the right balance. “We are being very mindful and measured in what we put in our public safety and security products. Whatever descriptive or generative AI is doing, it is validated and verified with a human in the loop, whether that is a security operations person, a 911 dispatcher, a first responder, or so forth. But AI can help improve accuracy and efficiency. With 911 calls, if you respond faster by one minute, that saves 10,000 lives in the U.S.,” Brown said.

Brown also raised the implications of fortifying public trust: “Motorola was the very first company in China in 1986, with the Galvins. We sued Huawei, we were only the second company after Cisco and John Chambers to do so. We’re in the middle of the seventh year of litigation against a company called Hytera. There is a lot of stuff people don’t agree on politically, but I think one thing which is truly bipartisan, and which Washington and the business community are getting right, is our approach to China.”

Steve Bandrowczak, the CEO of Xerox, also echoed the theme of trust in data and technology. “Given the physical world is colliding now with the digital world at a rate we’ve never seen before, we need to understand the data, who owns the data, where it is, where it is in transit… There’s a tremendous amount of complexity around who owns this data, and there will only be more personal data which is generated moving forward. We’re using AI at Xerox to defend and validate the origin and destination of data, and to make sure data is not changed in transit.”

The blockchain’s potential

Joe Lubin, the founder of cryptocurrency Ethereum, said that “for centuries, society has been organized via top-down trust and control. Authorities would operate through intermediaries, but in 2008, Satoshi came up with one of the most profound inventions of our time, which is decentralized trust. Blockchains, databases which we can all use and inspect with transparency, are growing increasingly scalable and will be ready for large-scale applications soon, and web3 is the natural evolution of the internet and web technologies. Blockchain can be orders of magnitude more secure than current internet protocols because it is built on cryptography from the bottom up and it is fundamentally user-centric, with shared trust rather than hierarchical trust.”

Jeffrey Solomon, the president of TD Cowen, also discussed the potential of crypto, adding that “Many advocate for a 33 Act equivalent for crypto. I’ve said for years that we benefit from the U.S. being the world’s reserve currency, and crypto is a threat to that in the long term if it is widely adopted, so we need to create a framework where capital can still flow to the U.S., whether it’s in fiat currency, trade, or crypto, or else some other society will do that for crypto and crypto money will flow to that society.”

Investment strategies in a disrupted market

CEOs were split on the right strategies to adopt from an investor’s perspective. Several participants raised the question of how companies are measuring return on AI investments. As CNBC Anchor Morgan Brennan put it, “The question for me is the return on investment piece. It seems that is still a huge unknown.

Brennan’s concerns were echoed by Evercore Activism Defense head Bill Anderson, who said, “One thing to keep in mind is shareholders are having a difficult time determining how much companies should invest in AI, measuring the returns on investment, and discerning winners and losers from AI within sectors, putting aside the Nvidias and pure-play AI companies.”

Mason Morfit, the CEO of investment firm ValueAct, said “We don’t invest in leading-edge technologies as much as we invest in incumbents that adopt technology a few years after the leading wave has crashed on the beach, so to speak. The shiniest new things always get a lot of attention, but we see some really interesting opportunities with incumbent businesses adopting earlier versions of AI and machine learning.”

Morfit calls these kinds of incumbent businesses grappling with technological disruptions the “Once and Future Kings” and points to his firm’s proven track record in helping incumbents adapt successfully to technological change, whether it was helping Microsoft bridge the end of the PC era into the beginning of cloud computing or 21st Century Fox bridge the transition to the streaming era.

On the other hand, Lloyd Blankfein, Senior Chairman of Goldman Sachs, gave his verdict on investing in the most cutting-edge technologies. “We invested heavily in emerging technologies. Some of it worked, all of it we were excited about, some of it we should have been excited about and some of it less so. All of it sounded terrific and we had to keep an oar in every pool to keep up. But the most important thing I could contribute in that role was just being a good goalie and keeping our spending in check. Not each technological lead will work out, and you have to take a deep breath and look at them more clinically and don’t get too caught up in the excitement, because these are business and financial decisions and not just technological ambition.”

Clearly, the insights shared with us by 200 top CEOs suggest that some of the most transformative uses of AI are taking place in plain sight at some of the world’s largest companies. These experienced non-tech titans embody the wisdom of Louis Pasteur regarding innovation in the field of practice, “Chance favors the mind that is prepared.”

Department: Faculty Viewpoints