A $2.1 billion bet on American AI sovereignty has materialized in a most unexpected place: a Brooklyn startup with zero public products. Reflection, founded just last year, has already secured a staggering $8 billion valuation without releasing a single AI model into the wild.
The rush to back the venture stems from mounting anxiety over China’s dominance in open-source artificial intelligence. Companies like DeepSeek have surged ahead by offering fully transparent code that clients can examine and customize, leaving Washington scrambling for a domestic counterweight. That geopolitical tension has transformed a fledgling team of former Google DeepMind researchers into one of the most heavily funded AI ventures on the planet.
Misha Laskin and Ioannis Antonoglou, both architects of early Gemini development at Google, launched Reflection in 2024. The duo quietly assembled a war chest from Silicon Valley’s most powerful institutions. Nvidia, Sequoia, and Lightspeed all claimed stakes, alongside 1789 Capital, where Donald Trump Jr. serves as a partner. The political throughline hardly seems accidental.
Rather than racing to ship a chatbot or enterprise platform, the company has directed billions toward constructing a massive data center in South Korea. Simultaneously, its engineers are crafting bespoke AI systems tailored specifically for Korean enterprises, a departure from the one-size-fits-all approach favored by most Western labs.
This deliberate opacity has not deterred investors. For an administration eager to prove American competitiveness, a well-funded, politically connected startup building infrastructure abroad signals strategic heft, even in the absence of demonstrable technology. Whether Reflection can translate its enormous capital reserves and elite talent into functioning systems remains the defining question. The answer will ripple far beyond Brooklyn.
