AI及前沿科技論壇

The promise of autonomous agents is massive, but the reality is often chaotic. Free automation still comes with five-figure monthly token bills and real security liabilities. Boards are asking what happens once an agent can move money as easily as it runs a workflow, and regulators are asking the same question. The 2026 playbook: build hard boundaries, stop token runaway, keep capital safe, and grow inside whatever rules land.

Physical AI is transforming how machines perceive, reason, and act in the real world. From autonomous vehicles to industrial robotics, embodied intelligence is moving from lab to deployment at unprecedented speed. This session explores the technologies, world models, simulation, and edge AI, driving this revolution, the economics of scaling physical systems, and the investment frameworks needed to evaluate platforms that blur the line between software and hardware.

This session introduces Google’s latest computing advancements - featuring flexible capacity models, deep observability, and enterprise-grade SLOs built for massive quantitative workloads. The session also shares how a leading global market maker leveraged these new TPU capabilities to scale high-performance financial modeling, eliminate compute bottlenecks, and sharply accelerate time-to-alpha.

The internet's supply of high-quality human-generated training data is running dry, and synthetic data is not a magic bullet. Enterprise-grade synthetic pipelines are becoming a genuine competitive moat for those who build them right. This session tracks the $2.25 billion synthetic pretraining data market, reveals how to manufacture the next generation of AI training data.

The panel focuses on the critical path for AI to move from technological breakthroughs to industrial collaboration. It brings together in-depth dialogue on cutting-edge robotics (brain, cerebellum, and body), world models and LLMs, the AIDC industry chain, and financial innovation. The forum will explore how embodied intelligence can move beyond R&D-intensive spending toward real-world deployment; how world models reshape cognition and decision-making; how AIDC supports computing infrastructure; and how financial capital accelerates industry-chain integration. Aimed at industry decision-makers, technology experts, and investors, it provides a cross-sector exchange platform to identify the next growth curve of the AI industry chain.

The era of expensive AI is ending, and the economic fallout is rapidly separating the winners from the obsolete. As inference costs collapse, the advantage is shifting from model builders to application scale-ups. This panel decodes the new economics of AI: how cheap compute is destroying infrastructure margins, accelerating Big Tech acquisitions, and rewriting the blueprint for which startups actually capture the trillion-dollar alpha.

What drives real adoption across borders? Is the future built on massive compute, fine-tuned domain models, or localized compliance? Bringing together leaders across enterprise AI workflows, specialized banking models, and digital asset infrastructure, this panel explores the practical realities of scaling cross-border. Panelists discuss navigating regional rules, securing AI execution layers, and deploying models that balance speed with compliance.