By: Sofiia Krazhan

Edited by: Eric Chen

The AI bubble — a period of inflated investment and unrealistic expectations in artificial intelligence — has grown following the widespread adoption of ChatGPT and similar models by other software designers and AI companies. The top 10 companies in the S&P 500 have collectively accounted for 40% of the index’s total gains, while representing over 20 percent of world market cap. These companies have heavily invested in AI infrastructure and software for their products, from Alphabet’s Google Gemini integrated in Chrome to Amazon’s development of their chatbot and other AI applications. With this integration now extending into the defense sector of the US military, pilot drones and weapons manufacturing are increasingly based on AI integration. This raises a critical question: what will happen to the defense sector if the AI bubble bursts?

In 2008, a bubble caused by rising prices of real estate corresponding to mortgage-backed securities crashed the US economy when subprime mortgages were integrated into mortgage bonds as diversified assets. The foundation went under and many people lost their homes and jobs. The current AI bubble differs in that we are not speaking of the same physical capital that existed in mortgage-backed securities. Instead, we are speaking of the software and hardware of AI — the platforms themselves being developed and the data centers behind them. These data centers require water and electrical grids from entire towns, cities, and even states in the United States. Not only are AI companies investing into development of large data centers for their operations, they are also investing in the computing chips for data processing. OpenAI promises to buy over $1 trillion dollars of NVIDIA chips, with projected IPOs anticipated as soon as 2026 or 2027. However, AI itself is ultimately an aggregator of information. As much as one can ask AI questions and feed it problems, there is only so much information it can accept and regurgitate back, because of that, AI is going to reach a plateau of its development. The distinction matters is while the commercial LLM sector may be vulnerable to bubble dynamics driven by plateau effects, military AI applications operate under different constraints, meaning that any economic correction in generative AI need not affect the continued expansion of defense capabilities. Yet despite these limitations, the defense sector has increasingly integrated AI into weapons and drone systems, suggesting that government investment may insulate military applications from commercial AI sector constraints.

Pilot drones and weapons manufacturing are more and more based on AI integration. Defense companies are either contracted with AI companies like OpenAI or they are developing their own internal AI models specific to the modern equipment they are aiming to develop. This includes applications like facial recognition, cyber security, and drone-based weapons. The defense sector has been consistently producing products for the United States since the end of the second world war and throughout the entire Cold War. The space race itself was a technological breakthrough, and the same pattern has continued to apply to AI as long as government contracts exist.

A new generation of college students and graduates will see a fall in the hype and employment around the AI sector. When the companies with the larger market share sneeze, their smaller competitors catch pneumonia. For example, if Claude from Anthropic becomes the only major AI model in popular use, other AI systems are going to fizzle out as non-legitimate companies because they were riding the hype wave of OpenAI.

The software and hardware sustainability of AI will widely depend on which companies survive. OpenAI and Sam Altman ventures are going to survive because he has developed good relationships in the White House and across the tech sector. However, smaller companies in San Francisco, for example, that are not directly applicable to a public sector will face the bubble bursting. These companies may declare bankruptcy, develop into other ventures, or be bought up by the biggest competitors.

The most pertinent issue is whether the defense sector will be left untouched by the AI bubble. Whenever something technologically advances, there is usually a defense reason for it — the space race, the internet and literally anything else. The defense industry generates revenue, pumps it into tech innovation and development, and funnels all that money back into defense. The government contracts are going to keep coming from companies like Lockheed Martin and others. As long as government contracts include innovation, the creation of jobs in the defense industry, and the contract of companies that create drone warfare, AI innovation is going to continue.¹⁶

Defense companies that are currently growing are going to continue growing as conflict progresses in areas of the world. As technology advances, defense companies will be contracted to develop new weaponry for the United States, strengthening national defense capabilities and establishing the defense sector as a continued engine of innovation. Countries like Ukraine that depend on Western weapons supplied via government channels will continue to receive these systems. In spite of continuous growth of defense companies utilizing AI, the primary issue that the AI market is going to face is the sustainability of the data centers that tech giants hope to install across leading states like Texas and Tennessee. AI companies do not necessarily have the same breakthroughs and developments as defense tech. The sustainability of AI as a weight on the environment and on human usage is going to depend on which companies survive. If the commercial AI sector faces a plateau, defense companies may redirect their focus from physical training of soldiers — which can take up to 10 years and hundreds of thousands of dollars — and instead invest in training them to use drones that function with AI and are supervised by humans.

The traditional view of military AI as a minor budgetary line item is being upended by a structural shift in how the Pentagon and Silicon Valley interact. While direct Department of Defense AI funding remains a modest $1.8 billion, broader investment in AI-dependent command and control systems is projected to hit $21 billion by 2025. This government spending is being mirrored—and amplified—by a surge in private capital: venture-backed defense startups raised $7.7 billion by October 2025, more than doubling the previous year’s total. Unlike the commercial sector, where skepticism regarding return on investment (ROI) and power constraints has begun to cool the “frontier model” hype, the military value proposition remains concrete and existential. The imperative to maintain a technological edge over adversaries like China creates a non-cyclical demand for autonomous systems and faster decision-making. We are transitioning toward a hybrid ecosystem where the military “imports” frontier models from leaders like OpenAI and Anthropic, then optimizes them for specialized defense use cases. This suggests that defense is no longer a niche sector, but a primary engine of the AI economy, providing a high-floor market that persists even when consumer sentiment wavers.

Whether the AI bubble bursts or not, the defense sector is going to continue to advance. Government contracts will sustain defense manufacturing regardless of commercial AI sector weakness. However, if smaller AI companies and infrastructure face sustainability issues, the cost and availability of AI-integrated weapons systems could be affected. The defense industry has historically been the chicken that came first and laid the egg — it generates the revenue that funds tech innovation. As long as this pattern continues, defense sector development and weapons sales to countries like Ukraine will remain supported by government investment and international demand.