Senior alumnus Ben Yi Chou is a 1990 (Class of '79) graduate of the Department of Computer Science and Information Engineering (CSIE) at National Chiao Tung University. Bringing his extensive venture capital experience in the high-tech sector back to his alma mater, he shared his industry insights on the artificial intelligence, especially this wave of AI brings to the computer science field. For most students, this was not just a lecture on mega trends in the AI industry, but a practical masterclass on "how computer scientists can reposition themselves to seize opportunities under the new industrial AI revolution." 

Alumnus Chou is currently a partner at GRC SinoGreen Fund, where he has long been deeply involved in early-stage high-tech investments across both the US and China. His expertise spans semiconductors, AI infrastructure, network communications, new materials, and advanced manufacturing. Leveraging his engineering background, he has carved out a path as a "hands-on venture capital fund manager," bridging the gap between technological R&D, commercialization, startup investment, and risk management.

A Career Bridging Technology and Venture Capital

After graduating from the CSIE Department at Chiao Tung University, Ben Yi Chou went to the United States to obtain a Master's degree in Management Information Systems from Boston University. Since entering the venture capital industry in 2001, he has accumulated 25 years of professional experience.

As a private equity venture capital fund manager, he manages early-stage venture capital for numerous domestic and international institutional investors, including sovereign wealth funds, global institutional investors, global fortune 500 companies, and family offices. 

His investment footprint covers Silicon Valley, Taiwan, and Mainland China. He is responsible for defining VC fund investment strategies, analyzing industry trends, and executing investment projects. He actively participates in boards of directors to guide corporate governance for post investment management. Throughout his career, he has successfully completed more than 30 investment exits via domestic or international IPOs or M&As, delivering exceptional investment returns for his investors.

The "Big Bang" of Generative AI

Alumnus Chou shared his perspective of AI revolution by a wealth of data and case studies to illustrate how the generative AI revolution has unfolded. Since the Google team published the landmark paper "Attention Is All You Need" in 2017, it has triggered a "Big Bang"-level evolution in Generative AI, bringing disruptive changes to various industries akin to a new Industrial Revolution. Individuals or enterprises that effectively leverage AI tools will experience massive productivity gains, completely outclassing traditional working models.

“AI is by no means a short-term bubble”. Global tech giants and leading AI firms—from NVIDIA, OpenAI, Anthropic, Google, and Microsoft in the US, to DeepSeek, Zhipu, Alibaba, Tencent, and ByteDance in China—are committing tens or hundreds of billions of dollars in capital expenditures.  These are the world's smartest minds and leading tech enterprises by far, and “they are going "All in AI."” Products and technologies are iterating rapidly at a weekly pace, bringing jaw-dropping advancements such as open-source and closed-source large language models (LLMs) varied by the most intelligent or the lowest-cost for AI inference, Long-context windows and multi-modality, model distillation, post-training, and fine-tuning for all sizes of models.  And enterprises are deploying their private AI inference systems locally to safeguard their business know-how and information security. 

Redefining Computer Architecture for the AI Era

In the realm of AI datacenters, NVIDIA is leading the industry in completely redefining AI infrastructure. From a computer science perspective, this means centering the system around GPU computing and rewriting the rules of "computer architecture" from scratch:

• Insufficient data access performance?Stack it with HBM (High Bandwidth Memory) and use Chiplet technology to co-package it directly with the GPU.
• GPU scaling bottlenecked by Moore's Law?Deploy ultra-high-speed NVLink protocols and switches to interconnect all GPUs within a rack, enabling fully unified, synchronous GPU computing.
• Motherboard materials unable to keep up with the speed?Transition directly to active copper cables for interconnects.
• GPU clusters not large enough to support the latest LLM training or massive token inference services? Utilize high-performance superNIC (network interface card) with RDMA (Remote Direct Memory Access) protocols to link all GPUs across clusters.
• Slow token throughput? Inadequate system memory? Cost too high? Heterogeneous GPU/computing environments?Develop multi-tier cache systems, intermediate system software, and optimized inference frameworks.

Overnight, the pillars of Compute, Network, Storage, and Power are being torn down and rebuilt. We are incredibly fortunate to participate in this new "AI Wild West," welcoming an era where computing power grows tenfold or a hundredfold.

Future Outlook: From Infrastructure to Exponential Application Growth

Looking ahead, Alumnus Chou believes that as the underlying AI infrastructure matures, AI token services will become as ubiquitous and readily accessible as electricity. Consequently, the top-layer application layer will experience explosive growth.

Just as Apple's iPhone, Amazon's AWS, Tencent's WeChat, and Google's Android ultimately reaped the rewards of the mobile 3G/4G/5G and fiber-optic broadband network infrastructure built during the 2000 dot-com bubble, the future AI world will witness the birth of the next trillion- or ten-trillion-dollar mega-enterprises.

Massive multi-trillion-dollar market opportunities will emerge across several fronts:

• Enterprise process renovation:Truly embedding model token services into business workflows, driven by the sensational emergence of agentic AI (like Open Claw) and the astonishing development of AI coding tools (such as Cursor and Claude Code).
• The AI financial infrastructure:High-frequency micro-transactions, automated financial flows, and asset tokenization on the blockchain executed by AI agents.
• The World Model: Physical & Biological Worlds:Enabling AI to enter the physical realm (via 3D spatial recognition) and the biological realm (AI for Science), paving the way for autonomous driving, embodied robotics, novel materials, and next-generation drug discovery and development.

Conclusion: Embracing the "AI Wild West"

Finally, addressing the question of "Is AI a bubble?", Alumnus Chou shared his personal perspective:

"Driven by capital market euphoria, the asset prices of AI may indeed experience a bubble. However, the transformation and impact brought by AI technology itself will be as profound as the Industrial Revolution." 

He concluded with an old Chinese saying, "A top scholar emerges every three years, but a master performer takes ten," to encourage the students of the CSIE Department. In this "AI Wild West," the rules are being rewritten. By embracing innovation, welcoming change, and masterfully utilizing AI tools and techniques, a single individual can potentially achieve what previously required the work of 10 or even 100 people.

His final advice to the students: Step out, embrace change, connect with the latest and best technologies.  Take the open-source software community, for example, that AI knows no boundary. Seize this new industry and this brand-new starting line—if you are a true geek, the entire world will see you!