DeepSeek Blows Past Its Funding Target as Reflection Previews a 501B Open-Weight Model
Tuesday, October 6, 2026 · 11 min
DeepSeek is close to raising at least 80 billion yuan ahead of a planned 2027 IPO, and Reflection announced Beam, a 501-billion-parameter open-weight model with weights promised later this month. Plus, Ars Technica on attacks that pivot between agent protocols, and a $140 billion plan for off-grid AI data centers in Japan.
Listen
Show notes
DeepSeek is close to raising at least 80 billion yuan ahead of a planned 2027 IPO, and Reflection announced Beam, a 501-billion-parameter open-weight model with weights promised later this month. Plus, Ars Technica on attacks that pivot between agent protocols, and a $140 billion plan for off-grid AI data centers in Japan.
In this episode
- DeepSeek set to raise $15 billion in Tencent-backed funding | The Straits Times — The Straits Times
# DeepSeek set to raise $15 billion in Tencent-backed funding | The Straits Times Published: 2026-10-06T06:02:39Z Source: straitstimes.com Language: en ## Story DeepSeek set to raise $15 billion in Tencent-backed funding | The Straits Times # DeepSeek set to raise at least $15 billion in Tencent-backed funding DeepSeek’s funding sets the stage for one of the most highly anticipated Chinese…
- Introducing Beam: Reflection's 501B open-weight model — Reflection — Reflection
Reflection's Beam: A 501B Apache 2.0 Frontier Model | AI news # Reflection's Beam: A 501B Apache 2.0 Frontier Model | AI news October 6, 2026 25 min read ## Reflection's Beam: A 501B Apache 2.0 Frontier Model Reflection has open-sourced Beam, and the numbers are the point. Five hundred and one billion total parameters with 23 billion active, pretrained on 23.8 trillion tokens,…
- MCP for agent-to-agent comms may be the riskiest protocol you've never heard of — Ars Technica
“AI agents give attackers a fresh set of connections to walk across,” Douglas McKee, director of vulnerability intelligence at Rapid7, told Ars. “Someone plants text in content, an agent will read it then pass it along to another agent as a normal delegated task, and that second agent runs it because it trusts whoever handed it the work. Every piece in that chain did exactly what it was designed…
- Japan launches $140B AI infrastructure plan - TechRadar — Efosa Udinmwen
Japan launches $140B AI infrastructure plan — but it won't connect to the power grid | TechRadar # ‘The keyword is speed to power’: Dell, Apollo, and Jera launch a $140 billion plan to build up to 4GW of AI data centers across Japan and Asia By Efosa Udinmwen Published 5 October 2026 3 min read Japan's data centers consume just 1.7GW, but this could change soon (Image credit: Getty…