Up North AIUp North
Back to news

DeepMind Legend David Silver Raises $1B to Challenge LLM Orthodoxy

DeepMind Legend David Silver Raises $1B to Challenge LLM Orthodoxy. Claude Bots Turn $313 into $2.38M on Polymarket with 98% Win Rate.

Share

DeepMind Legend David Silver Raises $1B to Challenge LLM Orthodoxy

David Silver, the architect behind AlphaGo and AlphaZero, has secured $1 billion for his London-based startup Ineffable Intelligence at a $4 billion pre-money valuation [4][5]. The round, led by Sequoia's Patricia Huang with backing from Jensen Huang, Google, and Microsoft, represents Europe's largest seed round to date.

Silver's thesis directly challenges the current LLM paradigm. Instead of training on human-generated data, Ineffable Intelligence focuses on reinforcement learning and world models to create "endlessly learning superintelligence" through simulation and trial-and-error [4]. The approach aims to discover novel knowledge beyond the limits of human-created training data.

The company targets applications in robotics and critical infrastructure like power grids, betting that self-discovering AI systems will surpass language models trained on finite human knowledge. It's a bold rejection of the transformer-based approach that has dominated AI development since 2017.

Claude Bots Turn $313 into $2.38M on Polymarket with 98% Win Rate

AI trading bots powered by Claude have achieved extraordinary success on Polymarket, growing $313 to $2.38 million over four months through 26,738 verifiable on-chain trades [6]. The bots maintained an approximately 98% win rate using strategies including latency arbitrage, oracle arbitrage, and news-driven trades.

The operation used MCP (Model Context Protocol) orchestrators with 12 servers, demonstrating Claude's superior risk management compared to GPT-based alternatives [6]. The autonomous systems were netting over $1,000 daily, showcasing AI's natural advantage in high-frequency prediction markets where speed and pattern recognition matter more than human intuition.

This success story highlights a broader trend: AI systems excelling in structured, data-rich environments where traditional human advantages—like "gut feeling"—become liabilities. The verifiable, on-chain nature of these trades makes this one of the most transparent demonstrations of AI's trading capabilities to date.

'Vibe Coding' Revolution Transforms Software Development

Claude Code is pioneering "vibe coding"—a development approach that transforms natural language descriptions into full applications through terminal-based agentic workflows [7][8]. The system uses skills (Markdown orchestrations), persistent memory, and parallel sessions to enable multi-agent development teams that avoid the typical "AI slop" of single-agent systems.

Integration with MCP servers allows Claude Code to connect with external tools and APIs, creating sophisticated multi-agent setups that can handle production-level complexity [7]. Developers report 20x speed improvements while maintaining robust code quality and comprehensive testing suites.

The emergence of repositories like Claude How-To and oh-my-claudecode provides production templates for teams ready to implement these workflows [8]. This isn't just faster coding—it's a fundamental shift toward orchestrating AI agents rather than writing code line by line.

Nebius Converts Cold War NATO Base into Arctic AI Fortress

Nebius has partnered with Verne Global to deploy a 10MW NVIDIA H200/Blackwell GPU cluster in a former NATO fortress at Iceland's Keflavik [9][10]. The facility, located at Valhallarbraut 868 in the Ásbrú district, became operational around March 2025 and runs entirely on renewable hydro and geothermal power.

Engineers transforming Arctic NATO base into AI data center

The datacenter achieves remarkable efficiency with a PUE of 1.1-1.2 through free air cooling, while subsea cables including Greenland Connect and Danice provide low-latency bridges between Europe and North America [9]. The symbolic transformation of a Cold War military installation into an AI compute facility captures the broader geopolitical shift toward AI sovereignty.

This positions Iceland as a critical node in Europe's push for sovereign AI infrastructure, offering grid-stable compute powered entirely by renewable energy. The facility represents the kind of strategic AI infrastructure that will determine competitive advantage in the post-code era.

What This Means For Your Business

The stories emerging this week paint a clear picture: we're witnessing the final phase transition from the coding era to the orchestration era. Anthropic's Mythos leak shows AI capabilities advancing so rapidly that entire industries—cybersecurity being the canary in the coal mine—face fundamental disruption within months, not years. Meanwhile, Claude's trading success and the rise of "vibe coding" demonstrate that AI systems now outperform humans in structured, high-frequency decision-making environments.

For business leaders, the message is stark: competitive advantage is shifting from those who can build software to those who can orchestrate AI systems effectively. David Silver's $1B bet against LLMs and toward self-learning systems signals that even today's cutting-edge models may soon look primitive. Companies still debating whether to adopt AI are asking the wrong question—the real question is whether you're building judgment and orchestration capabilities fast enough to stay relevant.

The Nordic infrastructure play in Iceland isn't just about compute—it's about sovereignty in an era where AI capability determines national and corporate power. As these systems become more capable, the organizations that survive will be those that learned to dance with AI rather than compete against it. Key takeaway: Code is becoming free, but the judgment to orchestrate AI systems effectively is becoming the only sustainable moat.

See what we're exploring →

Sources

  1. https://fortune.com/2026/03/26/anthropic-says-testing-mythos-powerful-new-ai-model-after-data-leak-reveals-its-existence-step-change-in-capabilities
  2. https://mashable.com/article/claude-mythos-ai-model-anthropic-leak
  3. https://investing.com/news/stock-market-news/cybersecurity-stocks-plunge-as-anthropics-claude-mythos-leak-sparks-ai-fear-4584897
  4. https://techfundingnews.com/david-silver-deepmind-1b-ineffable-intelligence
  5. https://www.implicator.ai/deepminds-david-silver-raises-1b-for-london-ai-lab-in-europes-biggest-seed-round
  6. https://emergingai.substack.com/p/from-300-to-238-million-in-4-months
  7. https://medium.com/@druce/claude-code-claude-skills-and-the-vibe-coding-revolution-d8bf0c7845f9
  8. https://natesnewsletter.substack.com/p/the-claude-code-complete-guide-learn
  9. https://northwiseproject.com/nbis-keflavik-iceland
  10. https://www.businesswire.com/news/home/20250311595160/en/Verne-Strikes-10MW-Deal-with-Nebius-to-Further-Expand-Europes-AI-Capacity

Stay ahead of AI

No spam. Unsubscribe anytime.

Want to go deeper?

Reading the news is one thing. Exploring the frontier is another. See what we're building.