Nvidia-Backed Reflection AI Seeks 2.5 Billion Dollar Funding Round at 25 Billion Dollar Valuation

What Happened

Reflection AI, a New York-based startup founded by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou, is in talks to raise $2.5 billion at a $25 billion valuation, according to multiple reports from late March 2026. The round would represent a threefold increase from the company’s previous $8 billion valuation, which was set when Nvidia invested approximately $800 million in an earlier round. Tech Startups reported that JPMorgan Chase and existing investor Disruptive are expected to join the round.

Why It Matters

The valuation trajectory is remarkable even by AI industry standards. Reflection was valued at $545 million as recently as last year, making the jump to $25 billion a roughly 46-fold increase in under 12 months. If completed, the deal would rank among the largest funding rounds ever tied to an open-source AI effort and would signal that investors view open-weight model development as a viable commercial strategy rather than a philanthropy exercise.

JPMorgan Chase’s involvement through its Security and Resiliency Initiative adds a geopolitical dimension. The initiative, launched in December 2025, plans to invest up to $10 billion across venture-backed startups tied to national security and critical infrastructure. The bank’s interest in Reflection reflects a growing view in Washington and on Wall Street that US-developed open-source AI models serve a strategic function as alternatives to Chinese competitors like DeepSeek.

Technical Details

Reflection’s core focus is automating software development with AI systems that can write, test, and maintain code at scale. The company was founded in 2024 and has grown rapidly from its initial DeepMind research roots. It builds open-weight models designed to run efficiently on Nvidia hardware, positioning itself within a small group of Nvidia-backed startups working to establish a network of freely available AI systems that companies, research labs, and universities can use and adapt.

The open-weight approach means Reflection publishes model architectures and trained parameters, allowing external developers to fine-tune and deploy the models on their own infrastructure. This contrasts with the closed-API model used by OpenAI and Anthropic, where customers access models through paid endpoints but cannot inspect or modify the underlying weights.

Who’s Affected

The competitive landscape for open-source AI development has intensified significantly in 2026. China’s DeepSeek demonstrated with its R1 model that competitive AI performance could be achieved with a fraction of typical training budgets, challenging the assumption that frontier AI requires Western-scale capital expenditure. Meta continues to expand its Llama model family, which has accumulated hundreds of millions of downloads. Mistral AI has established a foothold in regulated European industries after raising more than $1 billion.

For Nvidia, backing Reflection serves a dual purpose: it supports the open-source ecosystem that drives demand for Nvidia GPUs, and it counters the narrative that Chinese labs can match Western AI capabilities at lower cost. Reflection’s models are explicitly designed to run on Nvidia hardware, creating a flywheel between Nvidia’s chip sales and Reflection’s adoption.

What’s Next

The round has not yet closed, and the $25 billion valuation remains a target rather than a done deal. Reflection has generated no meaningful revenue to date, making the valuation entirely a bet on future capability and market position. The company’s competitive advantage hinges on whether its open-weight models can match the performance of closed competitors while attracting a developer ecosystem large enough to justify the premium.

Nvidia’s backing provides both capital and preferential access to GPU supply, but the gap between valuation and revenue remains a risk factor if AI investment sentiment shifts. The valuation represents the pre-money estimate and does not yet include the fresh capital from the funding round, meaning the post-money figure could exceed $27 billion. Security vulnerabilities identified in similar open-weight systems through algorithmic jailbreaking techniques also present a technical risk that Reflection will need to address as its models see wider enterprise deployment.