The Mark Cuban Podcast

Episode: DeepSeek’s New Rival? Reflection AI Raises $2B in the U.S.
Date: October 18, 2025

Episode Overview

This episode of “The Mark Cuban Podcast” focuses on Reflection AI, an ambitious new player in the artificial intelligence space that has just raised an eye-watering $2 billion at an $8 billion valuation. The host explores how Reflection plans to use these funds, their commitment to open-source AI, comparisons and rivalries with other AI leaders (notably OpenAI, DeepSeek, Mistral, and more), and what this development means for the future of American and global AI innovation.

Key Discussion Points and Insights

1. Reflection AI’s Mission and Context

• Ambitious Fundraising: Reflection AI has raised $2B—remarkably, without yet releasing a flagship product.
• Open Source Focus: The company pledges to create “frontier open intelligence, accessible to all,” positioning itself as America’s answer to China’s DeepSeek and Quinn.
• Vision Statement: Technology’s greatest advancements, the host points out, come from openness and collaboration (Linux, the Internet, etc.). Reflection wants to drive value here for AI.

Quote [04:50]:
“We’re building frontier open intelligence, accessible to all.” — Reflection AI’s X (Twitter) post

2. The OpenAI Backdrop

• OpenAI was originally open-source and mission-driven, but shifted to a for-profit model, leading to dissatisfaction and debate in the AI community.
• Elon Musk notably criticized OpenAI’s direction, but even his own Xai is not truly open source.
• Reflection AI seeks to claim the “open AI” mantle, especially for the U.S., where leading open-source models have otherwise been eclipsed internationally.

Quote [07:40]:
“It feels like Reflection AI is taking up what OpenAI was supposed to do… the torch in America at least has been passed to Reflection.” — Host

3. Impressive Team and Background

• Reflection’s team boasts alumni from Google DeepMind and leading AI breakthroughs—Palm, Gemini, AlphaGo, AlphaCode, AlphaProof, ChatGPT, Character AI.
• CEO Misha Laskin previously worked on reward modeling for DeepSeek’s Gemini project.
• Despite the $8B valuation, the team remains lean (~60 staff), primarily engineers and researchers.

Quote [17:25]:
“We've assembled a team who've pioneered breakthroughs including Palm, Gemini, AlphaGo…” — Host, quoting Reflection AI’s statement

4. Technical Ambitions: Mixture of Experts (MoE) and LLMs

• Mixture of Experts Approach: Reflection will use the MoE method—multiple expert “modules” within the model, each specialized (e.g., in coding, psychology), and selectively activated for queries.
• This is seen as the next frontier, inspired by breakthroughs from DeepSeek, Quinn, and Kimi in China.
• Scale of Ambition: Plans to train on tens of trillions of tokens, aiming for “frontier scale” performance.
• A core argument is that solving autonomous coding, one of the hardest AI problems, demonstrates readiness to build general-purpose LLMs.

Quote [24:15]:
“We built something once thought possible only inside of the world’s top labs—a large-scale LLM and reinforcement learning platform capable of training massive mixture of experts at frontier scale.” — Misha Laskin (Reflection AI CEO)

5. Strategic Geopolitical and Business Context

• Global Stakes and American Competitiveness:
  Reflection frames its mission as critical to maintaining U.S. leadership in open-source AI, now largely challenged by China (DeepSeek, Quinn) and France (Mistral).
• Unique Market Need:
  Many enterprises and sovereign states want customizable, secure AI systems not tied to closed-source U.S. or Chinese models due to legal and data sovereignty issues.
• Urgency:
  The absence of an American open-source champion could mean a permanent competitive disadvantage for U.S. and allied economies.

Quote [29:10]:
“So you can either choose to live at a competitive disadvantage or rise to the occasion.” — Misha Laskin

6. Investor Lineup and Industry Support

• Key Investors: Nvidia (logical, since Reflection will need vast GPU compute), Disrupt, DST, 1789, B Capital, Lightspeed, GIC, Eric Wang, Eric Schmidt, Citi, Sequoia, CRV, and more.
• Industry Reactions:
  • David Sacks (White House AI and Crypto Czar, All-In Podcast): Elevates the significance of open-source AI for U.S. competitiveness.
  • Clem Delang (Co-Founder, CEO of Hugging Face): Celebrates Reflection’s emergence as crucial for U.S. open-source AI, but underlines the importance of high-velocity sharing of models and data.
• Community Buzz:
  There is considerable excitement and hope that Reflection’s massive funding and elite team can deliver on their open-source promise.

Quote [33:10]:
“It’s great to see more American open source AI projects… The global market will prefer the cost, customizability, and control that open source offers.” — David Sacks

Quote [34:02]:
“This is indeed great news for American open source AI. Now the challenge will be to show high velocity of sharing of open AI models and data sets…” — Clem Delang

Notable Quotes & Memorable Moments

• Reflection’s Vision – [04:50]:
  “We’re building frontier open intelligence, accessible to all.”
• Commentary on OpenAI’s Abandonment of Openness – [07:40]:
  “It feels like Reflection AI is taking up what OpenAI was supposed to do… the torch in America at least has been passed to Reflection.”
• Team Pedigree – [17:25]:
  “We've assembled a team who've pioneered breakthroughs including Palm, Gemini, AlphaGo…”
• Technical Milestone – [24:15]:
  “We built something once thought possible only inside of the world’s top labs—a large-scale LLM and reinforcement learning platform capable of training massive mixture of experts at frontier scale.” — Misha Laskin
• Wake-Up Call for U.S. AI – [29:10]:
  “So you can either choose to live at a competitive disadvantage or rise to the occasion.” — Misha Laskin
• Support from Industry Leaders – [33:10] and [34:02]:
  See quotes from David Sacks and Clem Delang above.

Critical Segments & Timestamps

| Timestamp | Segment | |------------|-----------------------------------------------------------------| | 00:01 | Episode theme intro: Reflection AI’s $2B raise & open source | | 04:50 | Reflection’s mission statement, “frontier open intelligence” | | 07:40 | Analysis: OpenAI, Xai, and America’s open-source ‘torch’ | | 13:51 | Background: Team, funding journey, and technical roots | | 17:25 | Player moves: Reflection’s DeepMind and OpenAI alumni | | 24:15 | Technical ambition: large-scale LLM and Mixture of Experts | | 29:10 | Geopolitics: “rise to the occasion” and U.S. AI competition | | 33:10 | David Sacks’ industry support | | 34:02 | Clem Delang’s Hugging Face commentary on velocity and sharing |

Conclusion

Reflection AI’s $2B raise signals a seismic shift in the battle for open-source AI leadership, especially as strategic competition heats up between the U.S., China, and Europe. With its elite team, massive backing, and technical ambition (frontier-scale Mixture of Experts LLMs), Reflection is uniquely positioned to fill the vacuum left by OpenAI’s pivot away from openness. The episode leaves listeners with a sense of anticipation: If Reflection delivers, it could reset the global open-source AI landscape in America’s favor.