Datong Chen | Chip Stories
Accidental Entrepreneurship in Silicon Valley
Source: Yuanhe Puhua (ID: yuanhepuhua) — Image source: Midjourney
In 2012, I was invited to write a year-long entrepreneurship column for the alumni magazine Shuimu Tsinghua. I recalled my startup experiences in the U.S. (OmniVision Technologies) and China (Spreadtrum Communications), summarizing the first half of my life. What I didn’t expect was that, years later, both companies would return to China through acquisitions — starting new journeys.
Background:
- 2013–14: Tsinghua Unigroup acquires Spreadtrum Communications and RDA Microelectronics
- 2016: China IC Capital (now Yuanhe Puhua) acquires OmniVision Technologies
- 2019: OmniVision merges into Will Semiconductor, becoming China’s leading listed chip design firm
---
From Engineer to "Co-Founder"
The Desire for Change
In early 1995, as a senior engineer at National Semiconductor, I had mastered analog semiconductor process technology — but IC design tempted me. Moving from fabrication to chip design felt like shifting from tailor to fashion designer: more creativity, more challenge.
How Opportunity Knocked
While self-studying IC design, my mentor, Professor Wu Qiming, asked if I could solve a bipolar IC problem for Opus. I dove into technical books, worked late nights, and within weeks solved it. That success led to the CEO’s invitation to join a new company — focused on CMOS Image Sensors (CIS) — as a co-founder. I barely knew the term, yet agreed. Thus began my “forced entrepreneurship.”
Funding Origins
Raymond Wu, another co-founder, revealed the company’s seed funding story — a mix of networking and bluff involving a competitor’s engineering sample. It was risky, but once aboard there was no turning back.
---
Building OmniVision’s Core Team
- Four co-founders:
- Hong Xiaoying — CEO
- Raymond Wu — Marketing
- T.C. Tshu — Digital circuits
- Myself — Analog circuits (the core technology for CIS)
When my first recruit left for a better position, I gathered top Tsinghua alumni. Though none had prior IC product experience, we learned together. In six months, the team produced its first prototype.
---
Competing With Industry Giants
By 1996, industry interest in CIS exploded — Intel, HP, Sony entered with massive teams and budgets. OmniVision, with just a dozen people, relied on:
- Hard work: 12+ hours a day, six and a half days a week
- Fearless innovation: unconventional ideas that beat giant rivals in performance and cost
1997 Milestone: Developed the world’s first single-chip color CIS, enabling the mobile phone camera revolution.
---
Market Success
By the late 1990s, OmniVision led the global webcam CIS market, went public on NASDAQ in July 2000, and later supplied camera chips for Apple.
Remarkably, Tsinghua alumni founded six CIS companies, collectively controlling half of global market share.
---
Return to China: A Second Venture
Post-OVT, I returned to China in 2000 amidst a struggling domestic semiconductor industry. Through Vice Minister Qu Weizhi’s account of China importing all mobile phone core chips, I aimed to develop a 3G mobile phone core chip — a market larger than PCs.
---
Founding Spreadtrum Communications
- Partnered with Wu Ping’s team in Silicon Valley
- Secured funding from MediaTek’s Chairman Cai Mingjie
- Founded Spreadtrum in April 2001
Challenges:
- Competing against Texas Instruments, Motorola, Siemens, Philips
- Complex chip development, low funding ($6M initial round)
Strategy:
- 20+ chip designers in Silicon Valley, 50–60 software engineers in Shanghai
- Imported Silicon Valley experience, executed rapid development:
- 6 months: 2.5G chip design
- 10 months: verification
- 12 months: functional calls
- 24 months: mass production
---
Facing Funding & Market Hurdles
2002 — Pay cuts to survive, innovated by merging 3 chips into one ultra-compact design
2003 — Developed “total solutions” for domestic OEMs, fueling shanzhai phone market and dual-SIM breakthroughs
From 2003 to 2007, Spreadtrum tripled revenue yearly, listed on NASDAQ, and became China’s largest independent semiconductor design firm.
---
Surviving While Rivals Fell
Of over a dozen 3G chip startups globally, only Spreadtrum survived — thanks to:
- Raising funds under a “3G” banner but first producing 2.5G chips
- Exploiting the shanzhai market with a “rural-to-urban” strategy
---
The TD-SCDMA Challenge
China’s TD-SCDMA became one of three international 3G standards — but lacked foreign chip support. Spreadtrum pivoted from WCDMA to TD-SCDMA in 2003 amid SARS and skepticism.
Breakthroughs:
- Completed design in 6 months, first call within a year
- 2006: Overcame same‑frequency networking challenges via algorithm optimization
- 2009: TD operation license issued, full commercialization achieved
Impact:
- Domestic dominance of TD industry chain
- Influence on 4G standard-setting — TD-LTE accepted with 30–40% projected global share
---
Lessons Learned
Ten years of TD-SCDMA proved:
- Self-reliance is crucial
- Industry giants underestimated Chinese persistence and ingenuity
- Independent innovation is unstoppable with belief, confidence, and hard work
---
Modern Parallels: AI-Era Creativity
Today’s creators face similar dynamics — speed, adaptability, ecosystem building.
Example: AiToEarn官网
- Global open-source AI content monetization platform
- AI-driven content generation, cross-platform publishing, real-time analytics, AI模型排名
- Distributes across Douyin, Kwai, WeChat, Bilibili, Xiaohongshu, Facebook, Instagram, LinkedIn, Threads, YouTube, Pinterest, X (Twitter)
- Empowers solo creators and teams to monetize ideas efficiently — echoing the “fast fish eats slow fish” ethos of early Chinese tech startups
---
Key Takeaways for Innovators:
- Leverage unique positioning against larger competitors
- Use rapid iteration to outpace market shifts
- Build complete solutions to accelerate adoption
- Integrate ecosystem and market insight into technical excellence
---
Would you like me to also add a chronological timeline summary as a quick-reference table to this narrative? That could make the long story easier for readers to scan.
