2025 AI Landscape: Turning Points, Models, and Market Frenzy

Overview

In this episode of Crossing, we sit down with Zhuang Minghao, host of 'Tulong Zhishu' (屠龙之术), to reflect on the major AI and technology events of 2025. We delve into the rapid technological iterations and product disruptions witnessed throughout the year, examining both the excitement and the uncertainties of this era of innovation.

Guest

Zhuang Minghao is the host of 'Tulong Zhishu' (屠龙之术), a podcast focusing on technology and industry insights.

What You'll Learn

• Why 2025 was a 'turning point' year for AI, with both upward growth in infrastructure and potential market bubbles.
• The divergence and consensus in AI technical routes between China and the US, including the impact of open-source strategies.
• The competitive landscape of large models, from DeepSeek R1 to Sora 2, and the race for AGI.
• The rise of AI Agents, their current limitations, and their potential in vertical markets and on-device applications.
• How the capital markets are reacting to the AI boom, contrasting the enthusiasm in public markets with the more cautious sentiment in venture capital.

Topics Covered

• The 'turning point' year of 2025: AI growth and market speculation
• Large language models: US vs. China strategies, open-source impact
• Multimodal AI and world models: Sora 2, Genie, and the path to AGI
• The 'Agent Year': Manus, vertical applications, and on-device Agents (e.g., HarmonyOS HMAF)
• Capital markets: Public market frenzy vs. venture capital realities

Why This Matters

This episode offers a comprehensive retrospective on a critical year in AI, providing valuable insights into market dynamics, technological advancements, and strategic directions for a global audience of AI professionals and investors.