Source: Content compiled from semiwiki.
At the beginning of this article, let’s share a few facts:
Fact 1: How many exhibits related to Electronic Design Automation (EDA) are there in the Computer History Museum? Zero.
Fact 2: The average starting salary for software engineers at Netflix is $219,000, while for Cadence, it is $125,000; the starting salary for Cadence hardware engineers is $119,000 (source: levels.fyi).
Fact 3: The revenue of the EDA industry has historically accounted for 2% of the semiconductor industry’s revenue over the past 25 years, only recently rising to 3%.
It is evident that the EDA industry is severely underestimated.
The Cause of These Facts
I initially investigated the issue of value capture in EDA as a research project because I was puzzled as to why this critical technology has historically only accounted for 2% of the semiconductor industry’s revenue. As an outsider, I was unsure if this ratio was reasonable. How EDA is underestimated or overestimated (and charged too low or too high) is itself a question worth exploring, but it is not surprising that EDA practitioners wish to capture a larger share. When I presented this 2% figure at various conferences to draw attention to my research, the feedback was remarkably consistent:
When asked about this matter, employees from large firms, especially senior engineers, strongly expressed their views that the value of technology is underestimated. They believe the entire industry should be compensated more, and they themselves should be compensated more. “10%!” someone exclaimed. Well, that might be too greedy. Some engineers questioned what salespeople were doing that led to the stagnation of the industry’s revenue. Overall, employees from large firms were reluctant to disclose details, but they hoped I would share my findings once the research was completed because they wanted to know the truth. In fact, everyone wants to know, but no one is willing to talk about it openly.
Smaller vendors (salespeople, engineers, founders, sometimes even individuals wearing multiple hats) unanimously blamed the big companies: “It’s all their fault! These companies drive down prices and squeeze us. It’s so unfair! Not only are we suffering, but the entire industry is undervalued! You should ask them what they are doing!” Their manner of accusation, tone, and structure were astonishingly similar, as if rehearsed by some secret alliance whose sole purpose was to vent their grievances. While small EDA companies were eager to share their experiences with me, they felt insignificant because industry-level revenue and profits are determined by the three giants.
Being an outsider has its advantages. My ignorance gave me the courage to throw any random question to anyone willing to engage. When I posed this question to a panel of executives, Joe Costello, the former CEO of Cadence, replied, “Profit margin? I haven’t thought about that. I think about the value we provide through our products. I have always believed that if we create value for our customers, we will naturally receive value in return.” Oh my.
Wally Rhines, the former CEO of Mentor, had a more tempered response, consistent with his views in his writings. He believes that EDA maintains a healthy long-term relationship with customers: “I am confident that EDA salespeople will work with semiconductor customers to provide the software they need every year, even during downturns in the semiconductor industry, keeping average spending within budget” (“Forecasting Trends in Semiconductor Business After Moore’s Law,” Rhines, 2019). After a brief conversation with Wally, I was convinced he is a staunch market believer, thinking we are approaching a good equilibrium state (even if we are not, the market can always self-correct any temporary deviations). However, first, there may be multiple equilibrium states, and we are uncertain whether we are in an ideal state that can sustain EDA innovation; secondly, even if there is a force that can lead us to a better state, who can predict how long it will take? In the long run, the market will self-correct, but “in the long run, we are all dead” (Keynes, 1923).
It seems that EDA practitioners are more focused on value creation while neglecting value capture, expecting customers to automatically recognize their products at a reasonable price. It turns out my suspicion was not unfounded. Charles Shi, a senior analyst at Needham, pointed out at the DAC 2024 conference that many EDA practitioners believe that as long as they create value, they will automatically receive value, but unfortunately, this is not the case. Therefore, my goal is to explore why EDA has consistently occupied 2% of semiconductor revenue and how this figure is achieved. Theoretically, we know that this outcome is the result of macro market structure and micro business practices working together, but I am very interested in what these practices are and how they facilitate value capture. More specifically, I want to know how companies maximize profits through sales. The next question is, who should I talk to, and what questions should I ask, which actually covers the entire process. The people to engage include salespeople, engineers, CAD teams, and procurement personnel.
Preview of Research Findings
After formal and informal discussions with a dozen decision-makers, the answers quickly emerged. Although the current number of participants may not be sufficient to guarantee the publication of an academic paper, and the project is still ongoing, I believe it is meaningful to share some key findings. Most of these findings should not come as a surprise. Think of it as a mirror (perhaps slightly distorted)—we occasionally need to look in the mirror, even though the reflection should not be too shocking.
Here is a brief overview of some major findings:
1. Buyers typically hire professional negotiation teams, while sellers do not.
2. Due to long contract terms, suppliers cannot easily switch customers. This means that at contract renewal time, customers have the leverage to choose other suppliers, while suppliers do not have this advantage.
3. Even for private suppliers, negotiations almost always occur at the end of the fiscal year or fiscal quarter. (Buyers insist this is always the case, while some suppliers deny it.)
4. Excessive pressure from sales quotas can lead to significant discounts.
5. Customers are only willing to pay a small fraction for products from small suppliers and startups, even when purchasing products that are similar or even better than those from large suppliers.
6. Bundling practices can undermine monopoly profits and reduce the value of high-quality products.
7. A few large customers often contribute to the majority of a supplier’s revenue.
The above points mainly revolve around contract signing and negotiation, some involve organizational structure, and some involve practical operations, but none are profitable for EDA suppliers. Additionally, there are some positive factors:
1. Historically sticky users.
2. Little competitive pressure from startups.
3. Customer product search behavior is only triggered by pain points, never by price/cost.
For EDA, it seems like good news when system companies become new customers, but in reality, it may not be so. We often think that system companies have larger budgets and are more generous, thus improving EDA’s profit margins. However, system companies typically have shorter contract terms, which helps cultivate younger, more adaptable engineers who can switch tools. This enhances their bargaining power. Additionally, inexperienced users require more technical support, increasing suppliers’ costs.
In summary, industry structure, business models, and contract practices collectively drive value capture, which I will elaborate on in the next section.
My Insights
Which industry does Electronic Design Automation (EDA) belong to?
We can classify an industry from different perspectives. For outsiders, we can explain that EDA belongs to the software industry. For tech enthusiasts, we can say EDA belongs to the semiconductor industry. If I were to explain to business or economic researchers, I would say EDA provides process tools for chip design.
Why is the business domain of EDA so important? First, research shows that stakeholders evaluate products and companies from a categorical perspective, simply put, they categorize them to make it easier to understand and compare with similar products and competitors. If a company cannot be understood, meaning it cannot be categorized, it risks being underestimated. In this case, if investors and analysts do not know the appropriate reference group for EDA, they will not pay attention to it (or give buy recommendations), and the analysts’ “lack of attention” itself can negatively impact stock market valuations. At least existing research indicates this.
EDA (Electronic Design Automation) is indeed unique, and the market size is small. So, what industries can analysts refer to? What other stocks might the same analyst also focus on? CAD (Computer-Aided Design)? Yes, but the customer and downstream structures can vary significantly. Semiconductors? Yes, they are often covered by the same analyst, but the semiconductor industry itself is not a good reference unless their revenues are jointly influenced by downstream applications. Who would want to study this industry? Unless you have an electronic engineering background, no one would want to. So, how do stock analysts view EDA? A black box.
Willingness to pay is the second reason we should inquire about the nature of EDA business. Value is subjective, and we should truly understand how hardware designers perceive EDA.
The value of a business is typically divided into two categories: revenue generation and pain alleviation. EDA clearly does not belong to the former, while the latter is often driven by individual consumption seeking transient pleasure. Some say that companies profiting from the seven deadly sins are the most successful. Amen! Pride, envy, and lust—social media and cosmetics. Gluttony—fast food and alcohol. Sloth—takeout, gaming, watching sports, etc. Rest assured, EDA does not fit any of these criteria. EDA is not a consumable. EDA does not spoil. EDA does not create addiction.
Thus, EDA is more like a painkiller. Of course, if not careful, some engineers might even say it is a tool that “creates pain”: “Ugly!” “Frustrating!” “Outdated.” “Stupid.” You can feel their pain. But when I pointed out that perhaps the absence of tools would be more painful, there was little disagreement. The problem is that we lack understanding of the counterfactual situation, making it difficult to appreciate its value. You only understand the value when you lose it.
All of this indicates that even without competition, challenges in business models and practices (which we will explore in the next section), the pricing of EDA tools is inherently difficult.
But perhaps we should start with some positive aspects. Several factors favor the large EDA vendors that occupy 90% of the market share:
Strong user stickiness. Integrated circuit designers always face time pressure. Changing tools means adjustment, and adjustment means wasting time. Unless absolutely necessary, no one wants to change established processes, which means as long as EDA vendors do not perform terribly, they can barely maintain their position.
Minimal competition from startups and small businesses. In the past, many startups surpassed established companies and gained market share. But that era is long gone. A lack of investment, increasing complexity of issues, and the full exertion of oligopolistic power have led to the decline of EDA startups. A few small or emerging companies can only pick up scraps: they provide solutions to marginal issues, generating only a small fraction of the revenue of large vendors, and when large vendors bundle competitive tools with other products, they gain nothing.
Customers only search for alternative products when they encounter pain points. They do not seek alternatives simply because existing products are expensive. This means that once a customer is secured, the pressure of price competition is minimal. But this also means that the pressure is primarily on initial pricing. Once customers are locked in, as long as the pain points are manageable, they will remain.
To my knowledge, EDA suppliers have effectively leveraged the above factors, especially the second point, which squeezes the survival space of small businesses (not without cost). However, there are more factors negatively impacting the industry’s value capture, among which I believe incentive mechanisms and pricing are the main culprits.
1. Quotas
The story goes like this:
It was September 3, 2010, one month before Company X’s fiscal year ended on September 30. The management team set a revenue target of $300 million for the following year, meaning every frontline employee bore a sales quota. John was a salesperson whose job was to sell various tools to new and existing customers. John’s annual sales quota was $2 million. So far, he had completed $1.4 million in orders and had the opportunity to secure several new customers, totaling $850,000 in sales. However, he had not met his quota last quarter, and if he failed again, he would be fired. John had a monthly mortgage of $4,000 and two children, one in third grade and the other just starting school. John’s wife was unhappy with his recent work hours and frequent travel.
Ten days later, on September 13, John closed a deal that reduced his account balance from $600,000 to $300,000. John initially expected the new customer, Elppa, to purchase 20 licenses for between $450,000 and $550,000. Negotiations lasted a week, and the customer insisted on a price of $350,000, claiming it was their budget limit. By September 21, John felt immense pressure and asked his manager if it was possible to lower the price to $350,000. The manager nodded in agreement to lower the price to $400,000, as he was also trying to meet his performance targets. Ultimately, both parties settled at $400,000 and promised to provide additional support to the new customer. John breathed a sigh of relief. The customer was also satisfied. Management expressed satisfaction with the quarter’s performance. The license price for Elppa actually dropped from the target of $550,000 to $400,000, a reduction of 27%. Hopefully, the gap can be narrowed in the future.
This is a completely fabricated story. However, such situations have indeed occurred: if employees fail to meet performance targets for two consecutive quarters, they will be fired; customers always negotiate under the pressure of performance targets at the end of each quarter; and it is indeed difficult to raise prices for existing customers. Also, mortgages can be as high as $4,000 per month. It is said that to attract customers, companies sometimes give away some tools for free, resulting in those tools’ prices becoming unsustainable. It is said that during very tough times, companies even offered discounts of up to 98% (compared to the usual discount rate of 60% to 80%). Although this performance metric system (i.e., discontinuous incentive mechanism) used by all suppliers can sometimes boost sales, it can also misdirect focus, especially when performance metrics are always based on total amounts.
Given the current business model, this story describes some key issues regarding EDA’s value capture problem.
2. Negotiation
Most customers sign multi-year contracts. Such contracts benefit both parties. Customers can focus on their business and establish daily operations, while suppliers can be confident in maintaining operations in the next quarter. However, this also means that customers’ budgets are mostly locked in by existing contracts. In addition to limited customer mobility, EDA suppliers also find it challenging to acquire new customers. This puts suppliers at a disadvantage in contract negotiations. Since potential new customers are locked in by existing contracts, suppliers have almost no alternatives during negotiations and cannot compare with existing customers. In contrast, while it is not easy for customers to switch suppliers, they can do so at any time. This leads to an imbalance in negotiating power. This imbalance is particularly evident in the following situations: (1) the customer is already using other competitive tools; (2) the customer base is young and adaptable; (3) the customer is large in scale.
Customers only need to negotiate with a few EDA suppliers; whereas suppliers must negotiate with hundreds of customers. In these repeated negotiations, large customers often hire professional negotiators whose sole responsibility is to negotiate contracts; while suppliers rely on salespeople who are experienced and sociable engineers. No matter how excellent these salespeople are, it is hard to deny the skill gap between them, not to mention the pressure from performance metrics. However, it is precisely at these moments that the value created by EDA can truly be converted into actual revenue.
3. Bundling
Bundling is often considered an efficient price discrimination tool that can maximize value capture by meeting each customer’s willingness to pay (WTP) for specific products. The principle is that each customer has a different willingness to pay for specific products, but when a series of products are bundled, this difference is largely offset. The price of bundled products is usually fixed.
However, despite the same name, bundling in EDA is entirely different from the tools used in typical pricing strategies. In fact, using the term “bundling” is meaningless; a more accurate description would be “there is no fixed price, you can buy any product you want, we hope you try our other products, the more you buy, the bigger the discount, we just charge for all products together.”
So how does this practice harm economic development institutions?
For example, when a customer wants to purchase a competing product of A (which may be superior), the salesperson might say, “If you buy our B and C, we can give you A for free!”
This kind of “bundling” is essentially a form of poor price competition that squeezes the survival space of small business owners, while large suppliers, due to contract lock-in, have to endure low prices in the long term, as we discussed earlier.
There is also no mutual restraint among large suppliers to avoid competition on the same product features. Business school classrooms like to use the example of Coca-Cola and Pepsi to explain that these two companies are so profitable largely because they do not compete on price but rather win consumer recognition through brand differentiation. Companies can leverage their strengths rather than pursue complete uniformity.
This kind of “bundling” also obscures the value of each tool. Companies no longer need to compete A1 with A2 individually, but can bundle A1, B1, and C1 together to compete against any competitive tool. When you offer A1 at a low price, you actually undermine the profits of B1 or C1, even if one of the tools may have a monopoly position. As for how much value is lost, only these suppliers can provide answers based on their data.
Industry Structure
One of the decisive factors is the structure of EDA and its downstream industries. The three major EDA companies alone account for 90% of the market share. Such a high market concentration should bring strong market power. However, if you observe any large supplier’s customers—semiconductor companies—you will find that two or three companies can contribute 70% of their revenue. It seems that suppliers’ market power is not as strong as it appears.
Business model?
Many believe the problem lies in the business model. I am not so sure. The common view is that although EDA provides critical tools supporting downstream chip design and applications, the current business model does not allow it to capture a fixed share of the value created at the end. Others argue that unless EDA can charge per production unit like Silicon IP, this business model will be unsustainable.
Let’s examine the logic behind these arguments. It is akin to saying that universities should charge a fixed percentage of students’ future income; otherwise, it is unfair because a university degree helps them find jobs. Or that power tool manufacturers should charge customers based on the value of the houses they build. Furthermore, the coffee you bought this morning energized you and helped you close a $2 million contract, so you should pay the coffee shop a 1% commission. This makes no sense.
But this is how people think, and the pricing logic of EDA suppliers follows the same pattern: we estimate customers’ revenues and price discriminate based on that; we believe that for customers with simple designs or low application value, lower fees should be charged, while as design complexity and subsequent revenues increase, fees should be raised. Whether intentionally or unintentionally, many suppliers adhere to the same pricing logic.
This logic contains two parts. One explanation is that once a tool is provided, it can be used for 100 hours or 1,000 hours, and thus higher fees should be charged for users who use it longer. This part seems reasonable because suppliers are indeed providing more tools for heavy users, and suppliers have little or no additional costs. One solution is cloud services and usage monitoring, which can be gradually implemented over time.
The other viewpoint is that some customers use the tool to produce 1 million chips, while others only produce 10,000. Given that the tool helps the former achieve higher profits, should higher licensing fees be charged to the former? I believe not. The tool’s fees should be based on the added value it brings to customers—in this case, the cost savings compared to alternatives, which also determines the customers’ maximum willingness to pay—and at least cover its production costs (in this case, its maintenance costs). As for the exact position of the price within this range, it depends on competition and negotiation, which has been discussed in the previous sections on industry structure and negotiation.
So, what possible remedies can improve EDA’s value capture ability?
Part Three: Remedies
Three remedies are proposed:
1. Improved incentive mechanisms
2. Small accounts (but more numerous)
3. Stay in different lanes
The first remedy focuses on incentive issues in negotiations, the second on negotiation capabilities, and the third on willingness to pay and negotiation capabilities resulting from market structure.
1. Improved Incentive Mechanisms
The only opportunity for EDA to demonstrate its value is at the moment a deal is made. Who makes the deal and how it is made determines the revenue for the next two years. I am not a contract expert, but even without being an expert, it is clear that a total amount-oriented quota system distorts incentive mechanisms.
Here are some things large suppliers can do without significantly changing their existing business models:
Hire one or two sales incentive design experts. Ideally, they should have a PhD in economics or finance. They can conduct data analysis and some simple formal modeling. They have experience in experimental economics and behavioral economics, or at least some understanding of it. They could be recent graduates or currently working at companies like Amazon, responsible for pricing models that have never truly been applied. Currently, large suppliers employ hundreds of engineers with PhDs but only a few with bachelor’s or master’s degrees working on performance analysis and pricing. No, let an economics PhD work; their value will exceed that of four bachelor’s degrees.
It is best to hire them directly rather than through consulting firms, as sales performance data needs to be reviewed continuously, and incentive mechanisms may need adjustments. However, it is also reasonable to first assess actual demand through economic consulting firms.
I think EDA companies could also hire similar personnel for pricing.
In summary, any incentive scheme should consider quotas, not just based on total amounts. The actual price of each license also needs to be included in the incentive formula.
2. Small Accounts
This suggestion is not aimed at acquiring new small contracts but rather to change the current situation where 70% of EDA suppliers’ revenue comes from two or three customers, leading to limited bargaining power in each contract. Clearly, EDA cannot change the structure of the semiconductor industry, but it can change the concentration of contracts by breaking large contracts into multiple smaller ones, thereby reducing the importance of each contract. Essentially, it is treating one customer as twenty customers. This also aligns with EDA’s pricing strategy based on project complexity and total value. Different projects from the same customer may have varying complexities and production scales.
For customers, this will increase their contract signing and negotiation costs, not to mention the vacancy time for each license (although we can expect that final pricing will be based on actual running time), so what are the benefits of doing this? The budget allocation for different products should be clearly separated. Responsibilities at the project or business unit level should also be clearly defined.
3. Stay in Different Lanes
Each should develop in different areas or at least leverage their strengths. Instead of struggling to make up for their shortcomings, it is better to invest resources in innovations around their unique advantages. This also applies to suppliers’ recent diversification into other fields such as IoT and automotive. With new innovation areas continuously emerging, now may be a good opportunity to reshape the competitive landscape.
The above suggestions are intended to be concise and not a comprehensive solution, but rather to inspire practical thinking about how EDA can achieve value capture. Continuous innovation in EDA is certainly crucial, but maintaining the development of this field and attracting talent also requires serious attention to the issue of value capture.
Source | Semiconductor Industry ObservationRecommended Reading——EDA Giant Synopsys Announces Restructuring!
“China’s ASML” New Kai Lai Releases Two Domestic Self-developed EDA Software!
Trump Announces Restrictions on Important Software Exports to China: EDA Again Becomes a Hard-Hit Area
2.17 billion yuan, Domestic EDA Leader Again Makes Major Acquisition!
EDA Giant Reportedly to Lay Off 10% of Workforce!
By 2027, China’s Domestic Chip Self-sufficiency Rate Reaches 30%, EDA Graphics Storage Surprises
Domestic EDA Achieves Major Breakthrough
First Domestic EDA Listed Company Major Shareholder Significantly Reduces Holdings!
US EDA Giant Cadence Pleads Guilty and Pays Over $140 Million in Fines!
☞ Business Cooperation: ☏ Please call 010-82306118 / ✐ Or email [email protected]