AI’s ambitions depend on something less glamorous than the next breakthrough model: a reliable supply of electricity. As technology companies invest in nuclear power, what happens when Silicon Valley’s product cycles meet an industry built around decades of planning and a rigorous safety culture?

On this episode of Tech Transforms, Carolyn Ford welcomes Joyce Connery, owner of Connery Strategies, distinguished fellow at the Atlantic Council’s Nuclear Energy Policy Initiative, and former chair of the Defense Nuclear Facilities Safety Board. Joyce traces her career from nuclear nonproliferation work in Kazakhstan to the National Security Council and a decade overseeing safety at US defense nuclear facilities.

Carolyn also brings something personal to the conversation: nuclear power makes her nervous. Rather than dismissing that reaction, Joyce explores how fear, experience, transparency, and trust shape the public’s willingness to embrace nuclear energy.

Together, they discuss how AI demand is changing nuclear investment, the opportunities and challenges surrounding small modular and advanced reactors, and why cybersecurity must be treated as a safety function. Joyce explains why digital systems, physical operations, and security need shared ownership—and why innovation cannot come at the expense of the safety culture that supports them.

In this episode:
• How technology companies are changing nuclear financing, timelines, and priorities.
• Why public trust requires more than assurances that a technology is safe.
• Where small modular and advanced reactors fit into the energy conversation.
• Why cybersecurity, instrumentation, and nuclear safety must work together.
• How AI could support reactor modeling and operations while preserving human oversight.
• Why leaders should measure progress by electricity delivered, rather than designs announced.

Plus, Joyce weighs in on fictional power sources and the enduring absurdity of surviving a nuclear blast inside a refrigerator.

Show Notes:

Guest Email: Joyce@connerystrategies.com

LinkedIn: https://www.linkedin.com/in/joyce-connery-770199/

Relevant Links and Resources:

• Atlantic Council profile: https://www.atlanticcouncil.org/expert/joyce-connery/

• Defense Nuclear Facilities Safety Board: https://www.dnfsb.gov

• Harvard Belfer Center seminar with Joyce Connery: https://www.belfercenter.org/event/perspectives-role-and-challenges-safety-oversight-defense-nuclear-facilities

Drawing on a career that began on the help desk and eventually led to running a software company, James shares how firsthand experience with users shaped his approach to technology leadership. He argues that organizations should begin modernization by understanding what employees and customers are actually trying to accomplish, rather than starting with a software platform.

Carolyn and James explore the value of composable architectures that allow organizations to integrate best-of-breed technologies and replace individual components as needs change. That flexibility becomes increasingly important as AI accelerates software development and introduces new capabilities that organizations may want to adopt without rebuilding their entire technology environment.

The conversation also examines one of the biggest sources of inefficiency inside large organizations: the handoffs between teams. Processes that cross HR, finance, procurement, IT, security, and other functions often rely on emails, spreadsheets, shared mailboxes, and disconnected systems. James explains how organizations can create a more unified user experience while allowing individual teams to continue working within the systems that support their functions.

Security is another critical part of the discussion. Rather than adding security requirements after a system has been designed, James advocates building security, governance, access controls, and auditability into the architecture from the beginning.

They also tackle shadow workflows and the continued reliance on spreadsheets. Instead of blaming employees for creating workarounds, James encourages leaders to understand why those tools emerged in the first place and use them as signals of where processes may need improvement.

Finally, the conversation turns to vendor lock-in and why it represents more than a financial concern. Organizations that become overly dependent on a single technology provider can lose the ability to respond quickly to new technologies, changing requirements, and emerging AI capabilities. A composable approach can help preserve that agility.

For leaders wondering where to begin, James offers a straightforward first step: inventory your processes, establish a baseline, understand the outcomes you want to achieve, and then determine where modernization can deliver the greatest impact.

Show Notes:

LinkedIn: https://www.linkedin.com/in/jameswdavies/
Website: https://kineticdata.com

The discussion dives into the Pentagon’s evolving AI consolidation efforts under CTO Emil Michael, the growing influence of the Chief Digital and Artificial Intelligence Office (CDAO), and the rapid deployment of generative AI tools across government-furnished devices through genai.mil. Trac shares firsthand observations about how AI adoption is changing the way government teams review work, collaborate, and make decisions — while also raising concerns about overreliance on AI-generated output, information overload, and the expanding trust gap between humans and machines.
The conversation then shifts into one of the most provocative topics in cybersecurity today: Anthropic’s controversial Mythos model — an advanced cybersecurity AI reportedly so powerful at identifying vulnerabilities that Anthropic restricted access to a select group of organizations. Carolyn and Trac unpack the implications of private companies controlling potentially transformative cybersecurity capabilities, the governance vacuum surrounding advanced AI models, and the blurred lines between national security, corporate influence, and AI competition.

They also tackle the Department of Defense’s proposed acceleration of $152 billion in AI and technology spending, debating whether rapid investment is necessary to keep pace with innovation or whether “fast money with slow governance” risks creating chaos, opportunism, and massive waste. Throughout the discussion, Trac emphasizes the importance of infrastructure, workforce upskilling, governance, and ethical oversight — warning that technology is evolving faster than institutions are prepared to manage.

The episode closes with a nuanced and thoughtful conversation about diversity in technology and defense environments. Rather than focusing on political rhetoric, Trac reframes the discussion around “thought diversity,” arguing that resilient systems require teams built from varied experiences, perspectives, and backgrounds. Together, Carolyn and Trac explore the difficult balance between merit, inclusion, operational readiness, and the evolving realities of workforce policy in defense and cybersecurity.

This episode is equal parts AI strategy briefing, cybersecurity ethics discussion, and real-time reflection on how quickly technology, governance, and society are colliding in the defense space.

Show Notes Links:
Trac's website: https://tracybannon.tech/
Trac's email: trac@tracybannon.tech
LinkedIn: https://www.linkedin.com/in/tracylbannon/
Anthropic Mythos leak: https://fortune.com/2026/03/26/anthropic-says-testing-mythos-powerful-new-ai-model-after-data-leak-reveals-its-existence-step-change-in-capabilities/
DoD $152B spending: https://federalnewsnetwork.com/defense-news/2026/02/dod-plans-to-spend-entire-152-billion-from-reconciliation-bill-in-one-year/
Judge blocks Anthropic ban: https://www.washingtontechnology.com/companies/2026/03/judge-blocks-dods-ban-anthropic-calls-it-first-amendment-retaliation/412451/
DEI executive order: https://www.washingtontechnology.com/companies/2026/03/federal-contractor-dei-initiatives-singled-out-latest-trump-executive-order/412456/

In this special Tech Transforms + ATARC crossover episode, Carolyn Ford talks with Frank Indiviglio, Chief Technology Officer at NOAA, about how AI, data science, and high-performance computing are transforming weather forecasting, climate modeling, and ocean research. Frank shares how NOAA uses tools like REMUS underwater vehicles, balances open data with cybersecurity, and drives innovation through cross-sector collaboration.

Resources & Links:
🌐 Learn more about NOAA’s AI initiatives: noaa.gov/ai
🔗 Connect with Frank Indiviglio on LinkedIn: Frank Indiviglio

Tune in for insights on the future of AI, oceans, and innovation!