The Three AI Reckonings Your Board Isn't Discussing (But Should Be)

Executive Summary

While Boards debate AI adoption timelines and compliance frameworks, three seismic shifts are unfolding beneath the surface that will determine whether your organization thrives or becomes strategically obsolete in the next five years. These are not hypothetical scenarios for your risk committee to model. They are happening now, moving faster than most companies or countries can comprehend, and they require immediate Board-level attention.

I advise Boards and CEOs navigating this convergence. Let me be direct about what keeps me up at night:

(1) foreign open-weight AI undercutting U.S. industrial capacity,
(2) AI-plus-human produced, tailored software fabrics that also redefine enterprise cybersecurity, and
(3) the paradox that AI agents are creating more work instead of eliminating it. Each of these shifts individually would constitute a strategic inflection point.

Together, these three trends represent what I call “tech tectonics”, the kind of foundational disruption that separates companies that shape the future from those that react to it after their competitors have already moved.

For Boards and CEOs, your fiduciary duty now extends to understanding these dynamics and making hard choices about how your organization will respond. The question is whether you are ready to govern accordingly.

Foreign Open-Weight AI: Long-Term Economic Undercutting at Machine Speed

Let me start with the threat that most Boards are not yet treating with the urgency it deserves. Foreign open-weight AI models are systematically undercutting U.S. AI propositions, and the long-term consequences follow a pattern we have seen before with devastating effect.

This is not a new playbook. We watched it happen with computer chips. We watched it happen with solar panels.

A foreign competitor offers a cheaper alternative. U.S. companies adopt it because the cost savings are immediate and the strategic risks seem distant. Domestic production capacity atrophies. We forget how to manufacture these critical technologies ourselves. And then, once we have become dependent and our industrial memory has faded, the full costs emerge. The foreign models start phoning home. The long-term harm presents itself. By then, rebuilding domestic capacity is exponentially harder than maintaining it would have been.

History offers an ironic mirror here. In our early days, the United States stole loom intellectual property from the British when we were just getting started. We undercut their markets. We built an industrial base on technology we did not invent. Nations engage in economic warfare constantly. This is another round, happening at a speed faster than most countries or companies can process.

I called this out back in January 2025, before DeepSeek 1.0 debuted, when certain U.S. AI companies were operating under the assumption that their closed models could not be distilled by others to leapfrog ahead. The Silicon Valley consensus was that open-weight AI would never pose a serious competitive threat to proprietary systems. Then it happened, roughly ten days after I raised the alarm. The distillation techniques worked. The open-weight models delivered comparable performance at a fraction of the cost. And suddenly, the strategic assumptions underlying billions of dollars in AI infrastructure investment were in question.

For Boards, this creates a cascade of implications that extend far beyond your AI vendor contracts. If we all defect and wait for someone else to fix it at a later date, we are postponing the pain we will eventually have to endure to possibly fix it, if we can fix it at all. Moreover, the timing of this shift from U.S.-compute intensive AI models to foreign open-weight AI models may cause shockwaves through planned, new data center buildouts. Open-weight AI models are significantly less compute-intensive to deliver results. Companies may choose to operate them locally rather than relying on cloud infrastructure. The massive investments in AI-specific data centers may face utilization rates far below projections.

So, what can we do about this? We need to accelerate U.S. production of competitive open-weight AI models. Several already exist and are good.

What we lack is not technical capability but scalable business models and the funding mechanisms to support an ecosystem of U.S.-operated AI development that can compete on cost and performance with foreign alternatives.

This may require creative thinking about how we fund strategic technologies. Consider the C-SPAN model. When cable television was innovative technology, a small fee on cable stations funded C-SPAN so citizens could access direct-from-source videos of what Congress was or was not doing. Or consider the Universal Service Fund. When cell phones were innovative technology, a small percentage of cell fees paid for broadband connectivity to schools across all fifty states. Congress may have misused that fund later, but the initial model worked.

I am not suggesting we replicate either model exactly. But we cannot rest on our laurels, nor can we pretend this situation is fine.

Even if you are not in politics or government, your Board has a responsibility to point out the longer-term harms and advocate for action to advance more open-weight models both in the U.S. and regionalized ones around the world. Because if we do nothing, in five to six years we will face a real situation normal, all fouled up, to use the truest of military acronym jargon.

The strategic question for your Board is this: are you building AI capabilities on a foundation that you control, or are you creating dependencies that will become vulnerabilities the moment geopolitical winds shift? If the answer makes you uncomfortable, you have a fiduciary exposure that demands immediate attention.

The Self-Regenerating Software Fabric: Cybersecurity's Next Paradigm

Now let me turn to a shift that is happening in real time in some of the most high-stakes environments on the planet, and it has profound implications for how your organization thinks about software development and cybersecurity.

We may be living through a historical inflection point in how software is written and updated. Enterprises of the future will use AI combined with human teaming to refresh their entire software stack from scratch on a continuous basis.

We've already seen vibe-coded competitors of well-known software products arise, perhaps lacking full functionality and other viable product-related issues yet proof points that AI is transforming how software is written and maintained. Enterprises in 2027-onwards will no longer think of individual products or applications. Instead, they will operate a single comprehensive AI-augmented software fabric tailored specifically for their enterprise, one that they refresh with new features, new security controls, and new elements unique to their operations every five to six days.

Again, this is not speculative. I am seeing organizations in active conflict zones do exactly this right now, both for the functionality it provides and for the security advantages of operating on a constantly changing software fabric that is not known to attackers. Traditional cybersecurity assumes a relatively static software environment where defenders catalog known vulnerabilities, apply patches, and monitor for exploitation attempts. But what happens when the software itself is regenerated faster than attackers can map it?

The path to get there involves what I call a war of the GANs, or Generative Adversarial Networks.

Here is how it works: companies regularly rebuild their software fabric as one cohesive element, including security controls, business logic, and integration layers. They roll this out following rapid testing that is roughly ninety percent automated and supervised by digital agents, with ten percent oversight by human supervisors who provide contextual judgment and ethical guardrails.

The company retains a set of known test cases, including test cases specifically designed to verify that the fabric rebuild maintains critical functionality and security baselines. Then, in parallel, the moment this fabric is approved for deployment, the company deploys its own GANs to attack it. These adversarial systems probe for exploits, holes, vulnerabilities, and edge cases. The lessons from these internal attacks train the next rebuild of the fabric, creating a continuous improvement loop where the software evolves faster than external threats can adapt.

The net result is that the future of software is not individual products, but rather a single fabric tailored for each enterprise, constantly evolving, and in parallel constantly being tested by the company itself for risks and vulnerabilities. This represents a fundamental shift from static defense to dynamic resilience.

For Boards, this raises several governance questions that your technology committees need to address now, not after your competitors have already moved:

First, does your organization have the AI-augmented development capacity to operate in this new model, or are you still managing software as a portfolio of discrete products with annual release cycles? Second, do your cybersecurity investments assume a static environment, or are they architected for continuous regeneration? Third, and most critically, do you have the human-machine partnerships in place to provide the judgment and oversight that prevents this model from creating new vulnerabilities even as it addresses old ones?

I want to be clear about the risks. Vibe coding, where AI generates code based on natural language prompts without rigorous human oversight, can introduce subtle bugs and security flaws that are difficult to detect. This is why I emphasize AI combined with human teaming, not AI replacing human judgment. The humans in this model are not writing every line of code. Instead, humans are setting the strategic direction, defining the test cases, interpreting the results of adversarial testing, and making the final call on whether a regenerated fabric is ready for deployment.

This is decision elasticity in action. Your organization needs the capacity to regenerate its software fabric rapidly while maintaining the wisdom to ensure that speed does not compromise security or functionality. The companies that master this balance will operate inside the decision cycle of their competitors and adversaries. Those that do not will find themselves defending static systems against adaptive threats, a losing proposition in any contested environment.

The AI Agent Paradox: More Automation, More Work

Finally, let me address a dynamic that is confounding many of the executives I advise, and it has significant implications for how you think about AI adoption and workforce strategy.

AI agents are not taking your jobs. They are creating more work instead. Are we surprised? I have been raising this issue for years, pointing to historical precedents that should have prepared us for this outcome.

When computers first arrived, everyone predicted we would need fewer workers. Instead, we needed more people to keep the computers working, more cybersecurity professionals to protect them, and more analysts to interpret the data they generated. Work did not disappear. It transformed and multiplied.

When computer animation debuted, industry observers worried that we would need fewer animators. Surprise. We now employ roughly ten times as many animators as we did when computer animation first emerged. The technology did not eliminate the profession. It expanded the scope of what was possible and created demand for more human creativity to fill that expanded canvas.

The same pattern is repeating with AI agents, and recent reporting confirms what many small-to-midsize business owners are experiencing firsthand. For agents to provide value, they require constant supervision, frequent correction, and integration work that often exceeds the time savings they provide. This is turning small-to-midsize businesses toward using AI vendors to manage these systems for them, which creates a different kind of risk. You are outsourcing a core part of your business operations to a third party whose incentives may not align with yours and whose systems you do not control.

For Boards, this paradox demands a more sophisticated approach to AI adoption than the simplistic cost-reduction narratives that dominate many business cases.

Yes, AI agents can automate certain tasks. But automation does not equal elimination. It shifts the nature of work from execution to oversight, from routine tasks to exception handling, from individual productivity to system orchestration.

This has workforce implications that your human capital committees need to address. Are you training your employees to work effectively with AI agents, or are you assuming the technology will simply replace them? Are you building the organizational capacity to supervise, correct, and improve AI systems, or are you outsourcing that capability to vendors? Are you measuring AI success by the number of tasks automated or by outcomes improved?

It also has strategic implications for how you think about competitive advantage. If every company in your industry adopts the same AI agents from the same vendors, you have not created differentiation. You have commoditized a capability. The companies that win will be those that develop adaptive approaches to human-machine teaming, build organizational muscle memory around AI oversight, and use AI to amplify human judgment rather than replace it.

Decision elasticity requires humans and machines working together, each amplifying the other's strengths. Pivotability requires organizations that can adapt faster than their competitors, which means having humans in the loop who can recognize when an AI system is optimizing for the wrong objective and course-correct before strategic damage occurs.

We live in interesting times, to borrow the apocryphal curse. The question for your Board is whether you will treat AI adoption as a technology implementation project or as a fundamental transformation of how your organization creates value, manages risk, and competes in contested markets.

An Invitation to Strategic Leadership

These three reckonings, the foreign open-weight undercut, the self-regenerating software fabric, and the AI agent paradox, are not isolated challenges. They are interconnected dynamics that will reshape industries, redefine competitive advantage, and determine which organizations thrive in an era of tech tectonics.

If your Board is still treating AI as a compliance checkbox, cybersecurity as an IT issue, and workforce strategy as a cost-reduction exercise, you are already behind.

The companies that will lead in the next five years are those whose Boards understand that these are strategic imperatives requiring governance innovation, operational discipline, and the courage to make hard choices without perfect information.

My work as a senior advisor is dedicated to helping Boards develop the capacity to navigate these challenges proactively rather than reactively. This is not about implementing a specific solution or following a vendor roadmap. It is about building the organizational muscle memory to wrestle with complexity, embrace uncertainty, and preserve human agency even as you leverage machine capabilities.

The stakes are existential. The nature of work, the meaning of competitive advantage, and the future of entire industries are in flux.

For Boards and strategic-minded CEOs, the question is whether you will shape that future or react to it after others have already moved. Fortune favors the brave, and I strongly recommend a proactive leadership approach. Your fiduciary duty now extends into domains that traditional governance frameworks were not designed to address. The question is whether you are ready to lead and to govern accordingly.

Additional readings:

• Decision Advantage in the Age of Distributed Power: What U.S. Intelligence Executives Want Every Board
• Pragmatic, Practical Actions (Not Fear): We Can and Should Take Actions to Reduce Harms, Risks, and Concerns in the Present
• Rethinking AI Success: From Imitating Us to Helping Us Improve
• Scaling Resilience Through Strategic Partnerships: How Boards Build Pivotable Ecosystems Amid Tech Tectonics
• From Boardroom to Operations: Building Human-Machine Partnerships That Balance Speed with Wisdom
• Fiduciary Duty in the Gray Zone: What Boards Must Know About Converging Geopolitical and Technology Risks

Dr. David Bray is the 2026 inaugural recipient of the Marconi Society's Excellence in AI for global leadership. He is both Chair of the Accelerator and a Distinguished Fellow at the non-partisan Stimson Center as well as Principal and CEO at LeadDoAdapt Ventures, Inc. David previously served as a non-partisan Senior National Intelligence Service Executive, as Chief Information Officer of the Federal Communications Commission, and IT Chief for the Bioterrorism Preparedness and Response Program. Business Insider named him one of the top “24 Americans Changing the World” and he has received both the Joint Civilian Service Commendation Award and the National Intelligence Exceptional Achievement Medal. The U.S. Congress invited him to serve as an expert witness on AI in September 2025. He also advises corporate Boards and CEOs on navigating the convergence of AI, cybersecurity, and geopolitical risk.