The rapid development of artificial intelligence (AI) is transforming the global technology landscape, with open-source and open-weight AI models emerging as major forces in the industry’s future. Misha Laskin, co-founder and CEO of Reflection AI, believes that making advanced AI models more accessible could reshape technological innovation, business competition, and the global AI ecosystem.
In a recent interview, Laskin discussed Reflection AI’s mission to advance open intelligence, the growing competition between the United States and China, and the economic opportunities surrounding open-weight AI models. He emphasized that open AI technologies could become foundational infrastructure, much like Linux, Android, and the internet.
From Theoretical Physics to Artificial Intelligence
Laskin’s journey into artificial intelligence began with an interest in theoretical physics. He was attracted to physics because of its ability to explain complex phenomena through fundamental principles and its significant contribution to technological advancements.
His interest gradually shifted toward AI as he recognized the transformative potential of deep neural networks and reinforcement learning. The success of AlphaGo, which demonstrated that AI systems could achieve superhuman performance through learning and experimentation, further strengthened his interest in the field.
Before establishing Reflection AI in 2024 with co-founder Ioannis Antonoglou, Laskin worked at Google DeepMind, contributing to the development of advanced AI systems, including Gemini.
His experience in reinforcement learning and large language models helped shape Reflection AI’s initial vision of developing AI systems capable of performing complex cognitive tasks rather than simply responding to conversational prompts.
Reflection AI’s Mission to Make Intelligence Open and Accessible
Founded in 2024, Reflection AI aims to make advanced artificial intelligence more accessible through open-weight models and supporting technologies.
According to Laskin, the company initially focused on developing AI systems capable of coding and performing autonomous tasks. However, its strategy evolved as the global AI landscape changed and Chinese companies began releasing increasingly competitive open models.
Reflection AI subsequently expanded its ambition to strengthen the open AI ecosystem, particularly in the United States.
Laskin believes that open models are strategically important because countries and organizations should not become entirely dependent on AI infrastructure developed and controlled by a single nation.
He compared the importance of open AI to the historical impact of Linux, Android, and internet protocols, which became widely adopted technological foundations across the world.
Understanding Open-Source AI and Open-Weight Models
One of the central topics of Laskin’s discussion was the distinction between open-source and closed-source artificial intelligence.
He explained the difference through an analogy between renting and owning a property. Closed AI systems generally require users to access models through services controlled by their providers. Open-weight models, on the other hand, allow users to download model weights, deploy models on their own infrastructure, and customize them for different applications, subject to their licensing terms.
This approach provides businesses and developers with greater flexibility, control, and independence.
However, Laskin also acknowledged that open does not necessarily mean completely free. Although a model may be available under a free license, organizations still need computing resources, infrastructure, deployment tools, and technical expertise to operate it effectively.
He also highlighted the distinction between open weights and fully open-source AI. While releasing model weights makes it possible for developers to use and modify a model, fully open-source approaches may additionally involve making training data, development methods, and other technical details available.
Reflection AI’s broader approach is to release as much supporting research and technology as possible while recognizing that certain proprietary datasets cannot be publicly distributed.
The Growing AI Competition Between the United States and China
Laskin identified the growing competition between the United States and China as one of the major reasons behind Reflection AI’s strategic direction.
He pointed to the emergence of powerful Chinese open models, particularly DeepSeek, as an important development that changed the competitive landscape.
According to Laskin, the availability of capable open models has increased competition, encouraged innovation, and contributed to lower AI service costs.
He argued that the United States needs a strong domestic open AI ecosystem to ensure that businesses, governments, researchers, and international partners have access to competitive alternatives.
He also discussed the potential role of government support in China’s AI industry, noting that the scale of investment required to develop advanced AI models raises questions about the financial and institutional support behind these companies.
For Reflection AI, developing competitive open models is not simply a commercial opportunity. It is also part of a broader effort to ensure that advanced AI infrastructure remains accessible to organizations across different countries.
Why Open AI Models Could Transform the Technology Industry
Another important point raised by Laskin was the potential economic impact of open-weight AI models.
He believes that artificial intelligence should increasingly be viewed as a fundamental utility, similar to electricity, water, and internet connectivity, rather than exclusively as a premium technology product.
As open models become more capable and widely available, the cost of accessing AI intelligence could decline significantly.
This could create opportunities for businesses to develop customized AI applications without depending entirely on expensive proprietary systems.
Laskin also suggested that the economic value of the AI industry may increasingly shift toward the infrastructure and services surrounding models. These include computing resources, model deployment, customization, enterprise integration, and AI-powered autonomous systems.
For businesses, this could mean greater control over operational costs, data security, and the development of specialized AI solutions.
Reflection AI and the Future of Open Intelligence
Reflection AI’s first model, Beam, represents an important step toward its broader open-intelligence mission. The company says the model is designed to deliver strong performance while reducing the computing resources required for reasoning and complex tasks.
The company’s strategy also focuses on building an ecosystem around its models through developer partnerships, fine-tuning frameworks, and tools for creating autonomous AI agents.
With backing from major investors, including Nvidia, Sequoia Capital, and Citigroup, Reflection AI is positioning itself as an important participant in the growing open AI market.
The company faces significant competition from established AI laboratories and emerging model developers worldwide. Nevertheless, its focus on accessibility, customization, and independent AI infrastructure reflects a broader shift in how organizations may adopt artificial intelligence.
Misha Laskin’s vision highlights an important question facing the global technology industry: should advanced artificial intelligence remain controlled by a relatively small number of companies, or should it become an accessible resource that businesses, governments, and researchers can independently develop and customize?
Through Reflection AI, Laskin is advocating for a future in which open intelligence plays a central role in technological progress.
As competition between open and proprietary AI models continues to intensify, the development of accessible, efficient, and customizable AI systems could become one of the defining factors shaping the next generation of artificial intelligence.