The drugmaker says 98% of its workforce has completed an AI certification program. Every division within the company is empowered to take ownership of its AI strategy, model usage, and investments, while Pfizer’s technology team develops and manages the enterprise AI backbone across those use cases.

“We find that as our employees get smarter about how to use AI, they are less resistant to applying it to key workflows,” says Hancock.

This broad embrace of AI helped Pfizer land the 48th spot on the Fortune AIQ 75 list, one of 36 firms added to the expanded ranking from 2025’s inaugural AIQ 50. Other newcomers include Lumen Technologies (43), Palo Alto Networks (49), UPS (56), Cisco (61), and Thermo Fisher (75).

Pfizer, while new to the list, has been using AI for more than a decade. Today, those applications include speeding up drug discovery through virtual simulations that screen molecular compounds and cut scientific computation, at times as much as 90% versus traditional lab work. Pfizer uses AI to monitor its manufacturing production, while the commercial team is able to use the tech to draft sharper pitches to physicians.

Hancock also stresses the importance of partnerships, working closely with biotech startup XtalPi on quantum computing, AI drug discovery with biotechnology firm PostEra, and AI research lab Boltz.

Lumen Technologies is building the physical network that supports AI by increasing the amount of fiber it has laid into the ground from 17 million miles today to a planned 58 million miles by the end of 2031. Earlier this year, the enterprise networking company was tapped by Anthropic to expand the AI startup’s fiber network across North America.

Lumen has developed an AI-powered digital twin, providing real-time visibility, predictive troubleshooting, and the ability to simulate changes before the company makes any network deployments.

Jim Fowler, Lumen’s chief technology and product officer, says all of the code that went into developing Lumen’s Multi-Cloud Gateway, a routing service that allows customers to move data directly between multiple cloud providers, was written by AI. Employees were able to more quickly guide, text, and approve that code, but are also generating revenue for the business.

“I think companies that just focus on the efficiency side of AI are going to find some short-term success, but long-term failures,” says Fowler.

Over the past year, networking-equipment company Cisco has had splashy product launches that include the October 2025 debut of the P200 chip, designed to connect data centers, and the G300 networking chip that was announced in February.

Internally, Cisco elevated Jeetu Patel to the combined role of president and chief product officer in May 2025, and then consolidated all products and engineering functions under his purview. Thimaya Subaiya, executive vice president of operations at Cisco Systems, says this organizational change has been critical for the engineering teams across the business to more closely collaborate.

Another big debut for Cisco was the AI agent platform MyAgent, which was rolled out to 90,000 employees in August. Subaiya says the team is gearing up for a new release of MyAgent by the end of October with newer technology and making it more robust in terms of response time and the clarity of its answers.

“I truly believe in the AI world, three months is equal to five years,” says Subaiya.

At UPS, massive troves of data make AI especially enticing. The shipping and logistics giant delivers an average of 20.8 million packages per day that represent about 1 billion signals of data. Bala Subramanian, chief digital and technology officer at UPS, says every one of those packages has virtual representation in the company’s digital twin.

This allows UPS to monitor the movement of these goods but also factor in real-time external signals like inclement weather, tariffs, and traffic patterns that could impact a package’s delivery time, and use AI to ingest that data, predict when issues could occur, and reroute as needed.

UPS is also using AI for compliance, helping customers create accurate international shipping documents so their goods won’t get held up at customs. Agentic AI that’s been deployed include a tracking agent that proactively retrieves shipping status details when a customer calls in and an onboarding agent that schedules time-definite deliveries. 

“AI provides us scale, it provides speed, and it provides us an incredible amount of insight,” says Subramanian.

As businesses and governments raise increased concerns about the cyber threats from more advanced AI systems, Palo Alto Networks has bolstered its position by participating in cybersecurity initiatives like Anthropic’s Project Glasswing and OpenAI’s Daybreak, which each debuted this spring. Palo Alto Networks is also a member of Nvidia’s Open Secure AI Alliance, which aims to share research and tools that can help companies identify, address, and disclose software vulnerabilities. 

This month, Palo Alto Networks also debuted a new subscription security service that uses advanced AI models, including Anthropic’s Claude Mythos and OpenAI’s GPT.

“We’ve seen attackers that are purposely using AI models to carry out attacks; the net result is more novel attacks, more sophisticated and much faster than what you’ve ever seen before,” says Lee Klarich, the chief product and technology officer at Palo Alto Networks. “The flip side is these models actually can really help defenders.”

Internally at Palo Alto Networks, CIO Meerah Rajavel has lauded a continued focus on fewer, more high-impact AI use cases. Rajavel and her team built Panda AI, a service-management agent that automates 83% of IT tickets, and has since been expanded to support AI agents that have been deployed across finance, HR, and legal.

“We are moving away from this notion of ticketing,” says Rajavel. Panda is advanced enough to decide if it can handle an employee inquiry, she adds, or determine that “it really needs to involve a human.”

Life-sciences company Thermo Fisher has rolled out over 100,000 enterprise licenses of OpenAI’s ChatGPT to its workforce and made Microsoft’s AI tools widely available, helping departments like marketing and legal improve workflows by as much as 15% to 20%.

Partnerships have also been key to Thermo Fisher’s strategy, working closely with OpenAI and AI chipmaker Nvidia, as well as with the Mayo Clinic.

In September, Thermo Fisher and the Mayo Clinic debuted a new company to create a biomedical database that’s intended to help researchers better understand how diseases can develop biologically years before symptoms are observed. Called Precure, the company will use AI to accelerate drug development and earlier diagnosis.

“There’s a lot of handoffs that make science work,” says Ryan Snyder, CIO at Thermo Fisher. Applying AI across all drug discovery research, he adds, allows for “ample opportunities for Thermo Fisher to influence and support that process.”