On Artificial Intelligence: The Next AI Battle Isn’t About Bigger Models

My passion has always been healthcare. As a healthcare technology executive rather than a clinician, I have spent more than three decades focused on how technology can improve healthcare delivery. Our teams developed interconnected healthcare systems long before the internet had the capacity to support today’s data volumes. Along the way, we were awarded more than 30 U.S. patents spanning healthcare technology, expert systems, encryption, augmented reality and machine learning. When ChatGPT was introduced on November 30, 2022, I immediately recognized that artificial intelligence had crossed another major inflection point. Nearly four years later, after hundreds of articles, blogs, developing architecture and countless thousands of hours studying the field, I am more convinced than ever that the opportunity continues to expand at an extraordinary pace.

Only a year ago, the artificial intelligence conversation revolved around one question: Who has the biggest frontier model?

Companies such as OpenAI, Anthropic, Google, and xAI invested billions of dollars building increasingly capable large language models behind proprietary walls. Their strategy was straightforward: build the most advanced systems, keep the model weights private, and deliver intelligence through cloud-based services.

A Shift in Strategy

During the past several weeks, two announcements illustrate how quickly the landscape is changing. Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, released Inkling, its first open-weight model designed for developers to customize and fine-tune. At nearly the same time, China’s Moonshot AI introduced Kimi K3, another powerful open-weight model that immediately attracted global attention for its performance and low operating cost.

For investors, these announcements are important not because one model may outperform another on benchmarks, but because they signal a broader change in strategy.

Two Categories of Intelligence

Today, the AI ecosystem is evolving into two broad categories.

The first consists of Frontier Models. These are the largest and most capable proprietary systems built by companies such as OpenAI, Anthropic, Google DeepMind, and xAI. They represent the cutting edge of artificial intelligence and require enormous investments in computing infrastructure, specialized talent, and data. Users access these models primarily through cloud services while the underlying model weights remain proprietary.

The second category consists of Open Models, often called Open-Weight Models. Rather than accessing intelligence exclusively through a vendor’s cloud, organizations can download these models, fine-tune them for specific domains, and deploy them within their own infrastructure. That distinction is becoming increasingly important for industries such as healthcare, financial services, defense, and government where privacy, regulation, intellectual property, or national security make keeping sensitive data inside organizational boundaries essential.

Neither approach is inherently better. They solve different problems. Frontier models maximize capability. Open models maximize adaptability.

Many of the industry’s most influential leaders, including Jensen Huang of NVIDIA and Elon Musk, have publicly supported continued innovation in both frontier AI and open AI ecosystems. Frontier models continue to push the limits of capability, while open models accelerate adoption by allowing developers and enterprises to customize AI for their own applications.

In many ways, the relationship resembles the early evolution of computing. Mainframes did not disappear because personal computers emerged. Cloud computing did not eliminate on-premises computing. Different architectures evolved because different problems required different solutions. Artificial intelligence appears to be following a similar path.

What It Means for Family Offices

For family offices, this distinction matters because the investment opportunity extends far beyond the companies building the largest models.

As AI matures, increasing value may shift toward organizations that apply these models to solve real-world problems in healthcare, manufacturing, engineering, logistics, financial services, and scientific research. The next wave of value creation may come less from building another frontier model and more from building the applications, workflows, and domain-specific intelligence that run on top of them.

This is particularly relevant in highly regulated industries. Organizations may increasingly prefer models they can inspect, customize, and deploy within secure environments rather than sending sensitive information to third-party cloud services. That is one reason the rapid advancement of open-weight models deserves attention from long-term investors.

Artificial intelligence is still in its early innings. The headlines often focus on which company released the latest model or achieved the highest benchmark score. Those milestones are important, but they are not the entire story.

The larger story is that artificial intelligence is no longer one market. It is becoming an ecosystem of platforms, infrastructure, models, and applications.

Frontier models will continue expanding the boundaries of what artificial intelligence can achieve. Open models will help distribute those capabilities across industries, enterprises, and developers around the world. For long-term stewards of capital, understanding both sides of that ecosystem may prove more valuable than trying to predict which single model wins the next benchmark.

About the Author

Noel J. Guillama is Healthcare Editor for Family Office Networks and Chairman of HealthScoreAI, Inc. He is a healthcare technology executive, entrepreneur, inventor, and investor with more than three decades of experience spanning healthcare delivery, health information technology, artificial intelligence, insurance, and capital markets, that included two IPOs. He holds 37 U.S. patents and writes on the intersection of healthcare, artificial intelligence, technology, and long-term investment trends.

Mr. Guillama also served for more than two decades in leadership roles with two public endowments, including Director, Treasurer, Chairman of the Investment Committee, Vice Chairman, and Chairman, helping oversee investment portfolios totaling more than $500 million in assets. This combination of healthcare operating experience, technology innovation, and long-term institutional investment stewardship informs his perspective on the intersection of healthcare, artificial intelligence, innovation, and capital allocation.