Why Semiconductor Manufacturing Is Hard to Scale

Semiconductor manufacturing is difficult to scale even when demand is high because chip fabrication requires extreme precision, massive capital investment, and a tightly controlled supply chain that cannot be quickly expanded. The core challenge is that building a new fab takes years and billions of dollars, and even then, yields and performance depend on thousands of interdependent process steps that are notoriously hard to replicate at higher volume.

Formula

The calculator uses two simple formulas: pace = finish time / distance and speed = distance / (finish time / 60). Here, t is the finish time in minutes, d is the distance in kilometers, pace is in minutes per kilometer, and speed is in kilometers per hour. The first output shows your average pace, and the second shows your average speed.

Worked Example

Using the default values: finish time = 50 minutes, distance = 10 km. First, calculate pace: 50 / 10 = 5.0 min/km. Then calculate speed: 10 / (50 / 60) = 10 / 0.8333 = 12.0 km/h. So, you ran at an average pace of 5:00 per kilometer and an average speed of 12.0 km/h.

Background: Why Scaling Is a Unique Problem

Semiconductor manufacturing is unlike most industries. You cannot simply add more workers or run machines faster to meet demand. A modern chip factory (fab) costs between $10 billion and $20 billion to build and equip, and takes three to five years from groundbreaking to volume production. Even if a company starts building today, the chips won’t ship until years later. Moreover, the technology is so advanced that only a few companies in the world can make the most cutting-edge chips, and they are already operating at full capacity.

How a Chip Is Made: The Basics

To understand why scaling is hard, you need to know the basic process. Chips are built on silicon wafers through a series of steps:

• Design: Engineers create a blueprint with billions of transistors, often using electronic design automation (EDA) tools.
• Photolithography: A light-sensitive material is applied, then exposed to light through a mask to pattern the circuit. This is repeated dozens of times.
• Etching and Deposition: Material is removed or added to build the layers of the transistor.
• Doping: Impurities are introduced to change the electrical properties of silicon.
• Metallization: Tiny wires connect the transistors.
• Testing and Packaging: Each chip is tested, then cut from the wafer and packaged.

Each step must be performed with atomic-level precision. A single dust particle, a slight temperature variation, or a tiny vibration can ruin an entire batch of wafers.

Why Scaling Is So Difficult

Several factors combine to make scaling a nightmare:

• Extreme precision: Modern transistors are only a few nanometers wide—smaller than a virus. The equipment must place atoms almost exactly where they should be. This requires cleanrooms with less than one particle per cubic meter of air, and the machinery itself is incredibly complex.
• Yield is not linear: When you increase production, you don’t automatically get more good chips. Yield—the percentage of working chips on a wafer—depends on defect density and process maturity. A new fab might start with 50% yield, and improving it to 90% takes months of tweaking. If you rush, you end up with more wafers but not more chips.
• Equipment lead times: The machines that make chips, such as EUV lithography systems, are made by a few suppliers and take over a year to deliver. There is no off-the-shelf inventory; each machine is custom-built.
• Skilled workforce shortage: You need highly trained engineers and technicians who understand the process. These people are scarce, and training them takes years.
• Supply chain fragility: A chip fab relies on thousands of chemicals, gases, and materials. If any one supplier has an issue, production stops. The 2021 global chip shortage was partly caused by a fire at a Japanese chemical plant and a drought in Taiwan that affected water supply for fabs.
• Economic risk: Building a new fab is a huge bet. If demand drops or technology shifts, you could lose billions. This makes companies hesitant to invest unless they see long-term demand certainty.

Common Misconceptions: “Just Build More Fabs”

People often think that if demand is high, companies should just build more fabs. But it’s not that simple. First, there is a limit to how many fabs can be built because of the cost and the availability of specialized equipment. Second, even if you build a new fab, it will produce chips using the same process technology, which may already be outdated by the time it ramps. Third, increasing capacity for one type of chip (e.g., logic) doesn’t help with another (e.g., memory), because the processes are different. Finally, the industry operates on a business cycle; overbuilding can lead to a glut and price crashes, which is why companies are cautious.

How Companies Try to Scale: Strategies and Limits

Despite the difficulty, companies do try to scale. They use several strategies:

• Ramping existing fabs: They can increase output from current fabs by improving yield and reducing downtime. This is the fastest and cheapest way, but it only yields a 10-20% increase.
• Building new fabs: This is a multi-year, multi-billion-dollar effort. Companies like TSMC, Samsung, and Intel have announced new fabs in the US, Japan, and Europe, but they won’t produce chips until 2024-2025 at the earliest.
• Expanding capacity with partners: Some companies use “fab-lite” models, outsourcing to foundries. But foundries like TSMC are also limited.
• Improving process technology: Moving to a smaller node (e.g., from 7nm to 5nm) can increase the number of chips per wafer, but it is extremely difficult and expensive. It also requires new equipment and processes.

The fundamental limit is that scaling takes time and money, and there is no shortcut. Even with unlimited funding, you cannot compress the learning curve.

Costs and Economic Realities

The cost of scaling is staggering. A leading-edge fab costs $10-20 billion. The R&D for a new process node costs billions more. This is why only a few companies can afford to play. The cost per wafer is high, so chipmakers must sell a huge volume to break even. This creates a barrier to entry and limits the number of players. Additionally, the cost of a mistake is huge: a single error in the process can ruin a batch of wafers worth hundreds of thousands of dollars.

Safety and Environmental Concerns

Semiconductor fabs use hazardous chemicals, including acids, solvents, and toxic gases. They also consume enormous amounts of water and electricity. Scaling up production means more environmental impact, which requires permits and compliance with regulations. Water scarcity is a particular issue in places like Taiwan, where fabs use millions of gallons per day. This adds another layer of difficulty to scaling.

Alternatives: Not a Simple Solution

Some people suggest alternatives like using older technology or chiplets. Using older nodes (e.g., 28nm) is easier to scale because the equipment is more mature and cheaper, but it can’t meet the performance needs of cutting-edge applications like AI or smartphones. Chiplets—breaking a chip into smaller pieces—can improve yields and allow mixing different nodes, but they introduce new design and packaging challenges. These are not quick fixes.

Frequently Asked Questions

Why can’t we just build more fabs quickly?

Building a fab takes 3-5 years. You need to design the building, install ultra-clean air systems, and order specialized equipment that itself has a 1-2 year lead time. Even after construction, it takes months to qualify the process and achieve acceptable yields. So “quickly” is not possible.

Why don’t companies stockpile chips when demand is low?

Chips are perishable in a business sense. They become obsolete quickly because technology advances. A chip that is in demand today may be worthless in two years. Also, holding inventory ties up capital and warehouse space. So companies prefer to run just-in-time.

Why does the US rely on Taiwan for chips?

TSMC, based in Taiwan, has the most advanced manufacturing capabilities. They invested heavily in R&D and built a lead that others haven’t caught up to. The US and other countries are trying to bring manufacturing back, but it will take years and billions of dollars.

What is yield and why does it matter?

Yield is the percentage of chips on a wafer that work. If you have a 70% yield, you get 70 good chips out of 100. Improving yield is the most effective way to increase output without building new fabs. But yield is affected by defects, which are hard to eliminate.

Why don’t they just use bigger wafers?

Larger wafers (e.g., 300mm vs. 200mm) allow more chips per wafer, but they require completely new equipment and processes. The industry is slowly moving to 300mm, but it took years to transition. Moving to 450mm was considered but abandoned due to cost.

Related

• Logitech Mouse Lagging: Fixes That Work
• Virtual Host: Definition, Types & How It Works
• How to Clear a Stuck Print Job in Queue

in t | Finish time | min | 50 | 0
in d | Distance | km | 10 | 0
out pace | Average pace | min/km | t / d | 2
out speed | Average speed | km/h | d / (t / 60) | 2