A startup that barely existed a year ago is now being valued at $25 billion. That number tells you almost everything you need to know about where AI investment priorities currently sit.
Reflection AI is in talks to raise $2.5 billion at a $25 billion pre-money valuation, according to a Wall Street Journal report published Wednesday. The figure marks a sharp increase from the $20 billion valuation the company was reportedly targeting earlier this month, and is more than three times the $8 billion valuation it carried when it raised $2 billion just months ago in October 2025. Nvidia contributed approximately $800 million in that prior round.
JPMorgan Chase is considering joining the new round through its Security and Resiliency Initiative - a program launched in December to back US companies tied to national security and critical infrastructure, with plans to invest up to $10 billion into venture-backed startups. Existing investor Disruptive is expected to participate again. No final terms have been announced.
What Reflection AI Actually Is
Founded in 2024 by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou, Reflection AI is a New York-based startup building open-source AI systems designed to be freely used and adapted by companies, research institutions, universities, and governments. Its core focus is automating software development - building AI systems that can write, test, and maintain code at scale.
The company is part of Nvidia's Nemotron Coalition, a broader effort to develop freely available US AI frameworks as a strategic alternative to China's DeepSeek models. The open-source positioning is deliberate. Lower costs, faster adoption across industries, and alignment with allied governments are all built into the model. Previous backers include Lightspeed Venture Partners, Sequoia Capital, B Capital, Citi, and former Google CEO Eric Schmidt.
The Valuation Trajectory Is Remarkable
In March 2025, Reflection AI was valued at approximately $545 million. By October 2025 it reached $8 billion. It is now targeting $25 billion - a roughly 46x increase in under a year. The company has accumulated more than $2 billion in total funding commitments despite being less than two years old and generating little meaningful revenue by most accounts.
That gap between valuation and revenue is not unusual in the current AI funding environment, but it is worth naming. Investors are pricing in future potential - specifically the potential to become critical AI infrastructure for US allies at a moment when geopolitical competition around AI systems is accelerating.
The Geopolitical Dimension
Reflection's pitch is not purely commercial. The company is actively expanding its global footprint in allied markets. It is working with South Korea's Shinsegae Group to develop Korean-language AI systems running on tens of thousands of Nvidia processors, building localized capabilities for Korean institutions and businesses. The company plans to replicate this framework in additional markets.
JPMorgan's potential involvement through its Security and Resiliency Initiative - rather than traditional venture channels - underlines the national security framing around this raise. When a major bank evaluates an AI startup through a national security lens, it signals that the line between commercial AI investment and strategic infrastructure is blurring fast.
What It Means for the Broader Market
Reflection AI's valuation jump is a useful indicator of where capital is flowing in 2026. Open-source AI with sovereign deployment potential, backed by the dominant chip manufacturer, with institutional investors treating it as critical infrastructure - that combination is commanding premium pricing regardless of near-term revenue.
For business leaders tracking AI developments, Reflection represents a category of AI company that is less about immediate product revenue and more about positioning within the geopolitical race for AI dominance. Whether the $25 billion valuation holds up to scrutiny over time depends entirely on execution. The capital and backing to try are clearly in place.
