Home / Insights

Strategy frameworks

Process power: what rivals can see at Zara and TSMC, and still can't copy

The last of Hamilton Helmer's 7 Powers, and the rarest: a way of working that took decades to build and that competitors can see but can't copy quickly.

By Dan Lee · November 2026

In July 2020 Intel told investors its next manufacturing process was running about twelve months behind its own yield target (Intel). Intel had the money, the fabs and decades of experience. Its foundry business lost $13.3 billion in 2024 and $10.3 billion in 2025, and its 10-K names TSMC as its primary competitor in leading-edge manufacturing (Intel 10-K). Intel now has TSMC make some of its own chips.

Clothing has its own version. Rivals have studied how Zara designs, makes and restocks clothes for more than twenty years. Its owner still earns more than twice H&M's operating margin.

That gap is process power, the last of Hamilton Helmer's 7 Powers, and the rarest.

What the Power is

Helmer defines process power as routines built into an organization that lower cost or improve the product, which competitors can match only through a long, sustained effort. The benefit side is operational excellence, which is common. Every lean program and every continuous-improvement effort produces one.

What makes it Power is the barrier, which Helmer calls hysteresis: the improvement takes so long to build that competitors can't shortcut it, even when they can see it. It comes from two things. Complexity: the know-how touches many functions and suppliers at once. And opacity: much of it is tacit, held in people and habits rather than written down. Helmer's own example took fifteen years to transfer to the company's own suppliers. Improvements a rival can copy in a year are worth having, but they aren't Power.

The consumer case: Inditex (Zara)

Zara's owner, Inditex, has run a different clothing business since the first Zara opened in 1975. It designs in house, with more than 700 designers, produces in short runs through clusters of nearby suppliers, and ships everything through central hubs in Spain, so it can make more of what is selling and less of what isn't (Inditex). Harvard Business Review laid the model out in detail in 2004 (HBR). Rivals have studied it ever since.

They haven't matched the results. In fiscal 2025 Inditex reported a 58.3% gross margin and, by our arithmetic, a 20.1% operating margin on €39.9 billion of sales (Inditex results). H&M earned about 8% (H&M), and Gap about 7% (Gap). Inditex also held inventory equal to about 8% of sales, against about 15% at H&M and 14% at Gap. Selling what's in demand and holding less stock is what the process produces. A competitor can copy any one piece. The whole system, built over fifty years, has proved much harder to copy.

Inditex's limit is a newer model. Shein brings clothes to market in five to seven days with on-demand production (Reuters via Yahoo), and Inditex now buys from 50 countries, not mainly from near Spain.

The B2B case: TSMC

TSMC started in 1987 as a pure foundry: it makes chips that other companies design and sells none of its own. Over almost four decades it built a way of starting each new manufacturing process and raising yields quickly, one generation after the next: 7 nanometers in 2018, 5 in 2020, 3 in 2022, and 2 at the end of 2025. TSMC says it competes on "faster yield improvement and shorter cycle time" (2025 annual report), and that customers' value for "timely delivery of high yield products... is reflected in our pricing."

The learning curve shows in its margins. Each new generation starts out diluting gross margin by a point or two, then lifts it as yields rise. Gross margin was 46% in 2019, 60% in 2025 and 66.4% in the third quarter of 2026 (third-quarter results). In 2025 TSMC ran 305 process technologies for 534 customers. Each one adds learning that a rival with few outside customers doesn't get.

The rivals tried. Intel slipped on yield and is now a TSMC customer. Samsung struggled to reach target yields on its 3-nanometer process in 2022, and customers moved work to TSMC (Korea JoongAng Daily). In the second quarter of 2026, TrendForce estimated TSMC's share of foundry revenue at 72.5% and Samsung's at 5.9% (TrendForce).

The know-how is valuable enough to steal. In 2025 Taiwan prosecutors charged three people with taking TSMC's 2-nanometer trade secrets (The Register).

Where it breaks

Process power and scale are easy to confuse, and TSMC now has both. With 72% of the market and a capital budget of $60 to $64 billion for 2026, some of its margin is scale. Some is the AI boom. Don't read 66% as what process power alone is worth: TSMC's margin was 46% in a weaker year.

It also may not travel. TSMC's first Arizona plant slipped a year for lack of skilled installers (Engadget), and the company expects its overseas plants to dilute gross margin by 2 points in 2026, widening to 3 to 4 points. If the know-how depends on people and suppliers in Taiwan, moving it costs money. Customer concentration is rising too: TSMC's two largest customers were 36% of 2025 revenue.

What it looks like in your business

In a consumer business, process power is a retailer's supply chain or a restaurant chain's operating system that rivals have studied for years and still can't match on cost, speed or consistency.

In a B2B or industrial business, it is a cost or quality gap in manufacturing that competitors have tried, and failed, to close: faster new-product ramps, lower scrap, tighter tolerances held for decades. A family manufacturer with forty years of a particular way of working may have it. A platform assembled from four acquisitions in five years almost certainly doesn't.

Two warnings. If "the system" lives in one plant manager or the founder, it isn't process power. It is a person, and it may leave at closing. And everyone's costs fall over time. Only a gap between you and your competitors at the same point in time is evidence.

Questions to ask

• Where do we beat competitors on cost, yield, speed or quality, and have they tried to close the gap?
• How long would it take a well-funded competitor to match it, even if they hired our people?
• How much of our margin is this capability, and how much is utilization or a strong market?
• Is the know-how written down and spread across teams, or held by a few people?

What to do on Monday

Pick your last three product launches or new lines. Measure how long each took to reach target yield or cost, and compare that with what you know about competitors. If you ramp faster every time, that's worth protecting and pricing for: write down the routines, pair your experts with junior staff, and treat departures as a trade-secret risk. If you don't, call your operational work what it is: valuable, worth doing, and copyable.

Inditex's and TSMC's full cases are in our free download, 7 Powers in practice. Previous: cornered resource. Back to the start: the 7 Powers explained.

Common questions