The EDA Century Battle: Cadence vs Avant – A Tale of Corporate Espionage and Legal Warfare

Background

EDA, the primary tool for chip design, is often called the "mother of chips" and carries immense technological value, making it a highly lucrative industry where dominant players fiercely protect their innovations.

In the 1980s, commercial computing spurred a boom in chip design and CAD software, leading to a golden era for new companies.

By the early 1990s, two giants monopolized the market: Synopsys dominated front‑end technology with about 60% share, while Cadence controlled back‑end and verification technologies with roughly 80%.

Note: Roughly, the EDA CAD market can be divided into three parts: front‑end (frontend, e.g., Verilog simulation), back‑end (backend, e.g., place & route), and verification (DRC/LVS, etc.).

Turn of Events

In early 1991, four former Cadence employees—Stephen Wu, Liao Yu‑zeng, Zhuo Aike, and Cai Maike—left Cadence to found a new EDA company, Arcsys.

Two years later Arcsys launched its own place‑and‑route product, ArcCell, winning major customers such as Intel and Motorola, striking a severe blow to Cadence.

Cadence, suspecting code theft, launched an investigation and, by the end of 1992, assigned General Manager Xu Jianguo to lead a campaign—codenamed AK47—to out‑innovate and out‑market Arcsys within 47 weeks.

Xu Jianguo, known for his aggressive, battlefield‑like management style, formed a "B‑team" of technical and marketing staff, aiming to surpass Arcsys technically and crush it commercially.

Competition

On the market side, Xu personally visited customers who had switched to Arcsys, probing every detail of product differences and offering various incentives to win them back.

Technically, as chip design entered sub‑micron and deep‑sub‑micron eras, older routing methods were being replaced, and Xu pressured R&D to innovate faster than Arcsys.

Arcsys, lacking Cadence’s deep resources, struggled financially; by 1993 its sales were $1.7 million with a $2.2 million loss, pushing the company to the brink.

Turning Point

Internal conflict erupted at Cadence when Xu’s harsh management led to widespread dissatisfaction, culminating in a clash with another senior manager, James Solomon, in late 1993. The company’s president sided with Solomon, hired a new general manager, and Xu resigned.

Xu’s departure proved advantageous for Arcsys, which offered him 550,000 shares (valued at roughly $20 million three years later). He joined Arcsys in March 1994.

After Xu’s move, Arcsys continued to poach Cadence talent and possibly steal code, intensifying the rivalry.

For example, an employee named Mickey left Cadence for Arcsys, sending a 6 MB email from his workstation containing Cadence’s core QPlace source code to Arcsys the day before his departure.

Similar incidents followed, and despite Cadence’s attempts, concrete evidence of theft remained elusive until a seemingly trivial bug revealed the truth.

Critical Evidence

One ordinary employee, while using Arcsys software, discovered the following line of code:

Error a: color not found in this file.

The correct syntax should be:

Error：a color not found in this file.

This minor syntax error, left unchanged by the original author, proved that Arcsys had directly copied Cadence’s code, providing Cadence with decisive proof.

Armed with this evidence, Cadence initiated a large‑scale raid on Avant (formerly Arcsys) in December 1995, a rare move in Silicon Valley.

The legal process was lengthy; finally, in 2001, four of six defendants received one‑ to two‑year prison sentences, and Avant was ordered to pay $195 million in damages—the largest corporate‑to‑corporate IP settlement in Silicon Valley at the time.

Aftermath

From 1996 to 2001, while the lawsuit unfolded, the industry continued to evolve. Mentor Graphics re‑entered the market with hierarchical verification, gaining significant share.

Cadence’s morale plummeted; its senior manager resigned, expressing disgust with the protracted legal battle.

Avant’s board members were incarcerated, the company faced massive fines, and its civil suit with Cadence persisted, eroding sales.

Eventually, Synopsys announced an $800 million acquisition of Avant, ending the century‑long EDA feud.

In the acquisition, $100 million was allocated to several Avant directors to ensure they would not hold future management positions at Synopsys; Xu Jianguo personally received about $40 million.

Conclusion

The decisive turning point in this cross‑century Silicon Valley showdown originated from a simple programmer’s bug, illustrating how seemingly insignificant details can dramatically shift the balance in corporate espionage and legal battles.

For programmers, the lesson is clear: copying code is risky, but copying even the bugs can be fatal.