Modernizing for the mission: An interview with Jamie Wolff

Jamie Wolff has served in multiple roles across the federal government, from the US Navy to the Office of Management and Budget, and now leads the National Nuclear Security Administration’s (NNSA’s) modernization agenda as its chief information officer (CIO). Wolff brings a broad perspective on public sector technology to an agency whose work on the nation’s nuclear deterrent, counterterrorism, and energy management makes safety, security, and reliability mission critical.

The continued acceleration of technological change creates both opportunity and obligation for the NNSA: to bring AI, automation, and modern infrastructure to the mission without compromising security, resilience, or accountability. As part of McKinsey’s Rewiring the public sector series, Wolff sits down with McKinsey Partner Wasim Lala to discuss how the NNSA is balancing speed with security in its tech transformation, building a future-ready workforce, and reducing friction so that its scientists, engineers, and other employees can use the right tools to solve the right problems. This interview has been edited for length and clarity.

Wasim Lala: Right now, you are leading a very complex organization through a lot of complex technological changes. What are some of your top priorities right now, and how are those priorities shifting given how quickly technology is advancing?

Jamie Wolff: The complexity of the mission and the pace of change of technology is more critical right now than in the past. We have more work than we have had since the 1980s. There are more nuclear warhead programs in design and production than there have been for many years.

We have to reinvest in the organization, the infrastructure, the technology, the people—everything. We’re rebuilding the organization. We see this big growth and investment within the mission, but the technology is advancing at a pace that is more complex. We used to have a horizon of months within technology—18 or 24 months—and we used to know what was out there. Things are changing so fast and so significantly that month to month, it may be a different environment.

But we have to focus back on that mission. We bring the technology and information together to support our organization’s business and the mission. That’s where the CIO works.

Wasim Lala: Do you think there’s been a change in perception—or even, reality—of where the bottlenecks are in keeping up with the pace? Now that technology delivery is supremely accelerating, it seems that the onus is more on business owners to define what their problem statements are; the faster they understand how they want to solve the problem, the faster technology can support them.

Jamie Wolff: I have often found that the bottleneck is within the maturity of the organization the technology is supporting. With more progressive teams, you find that the technology delivery can be slow. They want more than you can give them at any given moment, but there are other teams where that is not so true.

Our science and technology people are very accepting of the pace of change. We have engineers and scientists working at Lawrence Livermore National Laboratory who are focused on some complex physics problems. For instance, how do we optimize a multivariate analysis for high explosives related to our work on nuclear weapons? This is the kind of project they had been thinking about for a long time. They thought they knew what the right approach would be but had not put the effort into solving the problem.

In real time, they ran this analysis using just an LLM [large language model]. Over several iterations, it optimized and produced a set of results such that PhD scientists, with their expert knowledge and experience, could say: “Those are reasonable results.”

These same users take that example that’s been proven to a certain level, move that workload over to one of the world’s fastest computers, and get a high-fidelity answer that drives that confidence level all the way up to where we need it. We’re demonstrating that cutting-edge technology today. These are the kinds of teams that see the technology, see the opportunity, and immediately want to begin using it.

Wasim Lala: For the mission-critical technology modernization programs that you’ve led, what has helped you accelerate?

Jamie Wolff: Last year, President Trump signed an executive order for a federal AI initiative, the Genesis Mission, which brings AI to science and technology research and development within the federal space. The Department of Energy is a key participant in this, and the NNSA also has a significant role.

So we have that guiding light, the articulation of a nationwide strategy, and we build our strategic plans against that. We also have implementation plans and projects that are defined to support it, with teams delivering those project plans right now. We have done that work to start driving solutions to actual problems and, more importantly, to align with the NNSA’s mission.

At the same time, we’re expanding our capacity and resiliency. We’ve become dependent on bandwidth to roll out new tools, so it has to be available all the time. We are making multiple automation tools accessible to our engineers and scientists, and we’re deploying them on both our unclassified and classified networks. With the architecture, the goal we’re working toward is having that unclassified environment mirror the classified one, so we can move information and work between the two.

We need to remove all our technology-related friction in the workplace. When we do, it enables the workforce to use the right tool for the right problem wherever they are, which is really mission critical.

One example: Our IT folks are using the Department of the Army’s tenant and are currently operating within it. We didn’t have to create our own. Both organizations signed the MOUs [memorandums of understanding]—then the agreements and contract—and replicated nothing. This happened in days.

Wasim Lala: You’ve served at the OMB [Office of Management and Budget] and in many parts of the federal government. What do you think the government’s AI priorities should be, and what can agencies do to make sure they’re getting the value?

Jamie Wolff: Among the most important things are what are you trying to achieve with AI, and how are you trying to achieve it? AI use should be focused on the highest-priority challenges, and the problem should be defined. Many organizations, and certainly federal agencies, have a mission to manage processes, and those processes can often be optimized using automation. Even before that, you’re digitizing and optimizing the process, bringing in automation to manage process flow. Then there’s the opportunity to put AI as a knowledge layer on top of it.

There’s also a lot of business efficiency to gain by using AI in specific parts of the process. From a procurement perspective, you start with the analysis and end up with a contract, and there are a lot of steps in between. How do we find efficiency in that process? And because our contracting professionals constantly change, how do you maintain and build that knowledge so it’s not lost when the person who did the work walks out the door? That’s where AI comes in and can add great value.

Wasim Lala: Speaking of procurement, how do you speed up that process? How do you identify the right solutions and the right partners for yourself at the NNSA?

Jamie Wolff: The competition is incredibly important, and it is built into our system. It is the law, but it’s also how we find the best value for the American taxpayer. There are certainly ways that we can optimize it. We need to work with the functional expert and identify points in the process that are duplicative.

But I think there’s also an end state that we don’t really understand. There’s a case study in Harvard Business Review about Walmart using AI to negotiate with their suppliers, in which they set the parameters of the contract they’re willing to accept, and they use AI to negotiate.

Within the federal government, how can we begin to take advantage of those kinds of innovations within any procurement team? Can we say, “A person already gave the parameters that are acceptable, and now it’s up to automation to execute them?”

Wasim Lala: In the world where you live, though, it can be a rare skill to understand the capabilities that you’re sourcing for.

Jamie Wolff: Yes. But we buy a lot of toilet paper, too. We purchase a lot of commodities; every organization does. So how do you get rid of the things you don’t need to put a lot of brain power into, then put the experts on the most important things? We have a lot of specialized capabilities that we focus on, and that’s where I’d like to put the most attention.

I’m guessing that the procurement process for toilet paper is fairly simple. But many simple cases will build up to create friction in the system.

Wasim Lala: Historically, I’ve seen a lot of private sector organizations outsource product ownership to their vendors and ask for essentially turnkey solutions. Now, with advances in AI, there’s a greater need for organizations to understand exactly what your products will be. I’m seeing more organizations build that product mindset in-house.

Jamie Wolff: Absolutely. When I started at the NNSA, we had a concept of moving toward a managed service provider, and I decided it was not a good idea for us. Part of it is the mission, right? The mission is owned by the government. In this specific space, it’s IT management as well as the securing of some of the most sensitive systems and information in the world. I’m not turning that over to somebody else; that is my responsibility.

I think there are bits and pieces that we can move out to find greater cost savings. But you have to be selective. More of this will be brought internal as we understand our mission a little bit more and move out some of the commodity services. This is already happening today.

Wasim Lala: A favorite quote of mine is that AI will become almost a common capability—and data will be where you get your proprietary advantage. Do you agree? Will data be the real source of advantage, and how do you think about data with respect to AI?

Jamie Wolff: This goes back quite a while. In 2018, during the first Trump administration, the federal government said that data was a strategic asset. Back then, I had the opportunity to work on the implementation of that strategy.

We were realigning government to take advantage of data and information, which are two sides of the same coin. I think it’s incredibly important; we organize ourselves against information and storytelling. We need to have that information available and also need information that’s understandable.

How do you have both? Typically, in technology, we’ll use some kind of translation because both versions are right. When we create truly enterprise-wide solutions, you can’t have thousands or even millions of translations. You need a common taxonomy. NNSA is doing that right now. It’s one of the places where we are working with McKinsey to develop a common ERP [enterprise resource planning] environment for all of our sites on a classified network, so they have the autonomy on the unclassified side.

We can use that environment to pull common information for reporting and analysis, but we can’t do that if the taxonomy’s not the same. This is an area that we must invest in; it’s basic to what we do. It’s normal technology as well as more sophisticated technology that lays the groundwork for artificial intelligence.

Wasim Lala: We’ve talked a lot about technology, but an important part of all of these changes is the human element. How do you think about the workforce of the future, particularly as a significant portion of the NNSA’s workforce will hit retirement age in the near future?

Jamie Wolff: The Department of Energy is focused on this now: specifically, on some differences between the whole workforce and the IT workforce. Broadly, NNSA has been growing; the majority of the workforce started within the past five years. We’re hiring talent wherever you can find talent, which typically tends to be younger.

Younger employees have the mobility in their career, and federal agencies—as the employer—have a great opportunity to bring them in, then help them embrace the mission and feel that commitment. Then they tend to stay because once you get a midcareer or late-career employee, you’re probably going to keep them. It’s slightly different on the IT side. We have high turnover, which is part of our model. I don’t get nervous if our turnover is 25 percent.

In fact, that’s where I want it to be. That means that our workforce is moving up, moving laterally, changing organizations, changing teams; that’s actually healthy in technology. As technology is modernizing and the workforce is turning over quickly, that creates the opportunity to focus on new technology areas or new sectors.

Wasim Lala: How do you think about public trust within the NNSA’s context and its use of AI?

Jamie Wolff: We use these tools to improve efficiency and decision-making in support of the mission and are not putting nuclear devices at the whim of AI. One of our missions is building and maintaining nuclear weapons. We’re augmenting the capability of true experts in the field and making them more efficient and effective. We are looking at new opportunities to improve production and logistics—those kinds of things. How do we bring AI on top of that?

We have a couple of use cases related to systems maintenance, which we do periodically because these things tended to fail in the past. We wrote rules that said, “Monthly, I’m going to do this and yearly, I’m going to do that.” But what if we don’t have to do that anymore? What if we can get enough information out of the machine to predict maintenance and potential failures?

This is incredibly important because it may change the paradigm of how we operate. We still have the monthly maintenance cards, but at some point we’ll move away from that. From a production perspective, that changes how we produce the pieces and parts that go into nuclear weapons, but it does not by any means put that weapon at risk. The weapon is too important and there’s too high of a consequence if something goes wrong.

Wasim Lala: With respect to tech spending, I feel like we’re in a paradigm shift. Before AI, everyone used to think about it as IT as a fixed cost. Now, predictions suggest that AI costs could become almost 30 percent of IT budgets—if not more—in one to two years. How are you thinking about this, and who, between IT and your customers, will bear that cost?

Jamie Wolff: Within that, there is the normal technology management model of enterprise level and commodity level. AI, including cloud services, is a commodity; it’s not unique. But that line between enterprise level and commodity level is hard to identify, and I think it changes frequently. For us at NNSA, our model is based on an appropriation from Congress. If I don’t have an appropriation to provide a service to another entity, then that’s an entity to which I have to allocate cost.

If they’re building on top of it and in a different part of the organization or even another federal agency, then the cost is theirs to bear. This is the complexity that McKinsey has helped us with: trying to figure out the framework that creates transparency where we can put cost, spending, and everything in the right buckets. We’re going to be working on this—the efficiency of these processes, the administrative burden of cost management—for a while.

Wasim Lala: Looking ahead five to ten years, what do you think will be the key facets that differentiate public sector organizations that take full advantage of cutting-edge technology from those that do not?

Jamie Wolff: At NNSA, our technology modernization has a rhythm and a process that we truly understand. To be successful, you have to continue on that journey toward a truly digital enterprise. The AI landscape is very murky. There’s technology on the horizon that I think is interesting, but it’s unclear how valuable it could be to an enterprise. Quantum seems amazing, but does it impact an organization?

Ten years from now, what I don’t want is for the technology to be doing more of the same. I don’t want to be doing the same thing better. We’ve got to find something different. We’ve got to be able to add value to the organization—take the low-value work and paper pushing out. And with our people, raise their skill sets and refocus them on the things that truly matter.