How US manufacturing was gutted with a smile

Few management theories have been as influential as the Smile Curve. And few have been as destructive. The influence of the Smile Curve has been profound, embedding itself in the strategies of global corporations for over three decades. While the curve may have “worked”, it was always measuring the wrong thing, misleading America into its current de-industrialized and de-skilled predicament.  

The Smile Curve was proposed by Stanley Shih, founder of Taiwan’s Acer Inc, in 1992. The company began in 1976 with the name Multitech as a computer parts distributor and consultant on the use of early microprocessors. By the 1980s, the company had become a manufacturer of IBM-compatible PC clones, selling computers under its Acer brand as well as contract manufacturing for third parties.

Through its various business lines, Shih discovered that most of the value was captured in R&D/branding, which differentiated products, and marketing/service, which drove revenue. The least value was captured by the company’s manufacturing arm, where competition compressed margins and capex requirements diluted returns.

With globalization – especially after China joined the WTO in 2001 – few business models were uncorrupted by the Smile Curve. Western companies (Japan and Korea included) crawled all over each other to divest manufacturing operations, become asset light and “move up the value curve.” Research, branding and design were sexy. Marketing and sales were rock and roll. Manufacturing was for hopeless bores with paunches and comb-overs.

Apple famously does not manufacture any of its products. Manufacturing had long ago been outsourced to original equipment manufacturers (OEMs) like Foxconn.

The sweaty browed labor of manipulating a pair of dexterous hands, managing thousands of factory workers, troubleshooting finicky machinery, programming industrial robots and orchestrating just-in-time delivery from dozens of suppliers were all chores not only on the bottom of the Smile Curve, but also done in noisy, smelly, sometimes dangerous and always poorly decorated factories.

The “real work” was done in Apple’s swanky Cupertino offices, over an iced matcha latte macchiato, where MIT grads wrote code for whiz-bang user “experiences”, designers rounded the edges of rectangles and Harvard MBAs planned the next FOMO maximizing marketing campaign.

Of course, Cupertino is where the real work is done. Have you seen the cost breakdown of an iPhone? Have you seen an iPhone’s gross margins? A 256GB iPhone 16 Pro Max sold for $1,199 with a 59.5% gross margin. Its total build of materials was $485 – $428 in parts and $57 in assembly costs.

Cupertino is where it’s at with mid-level software engineers commanding compensation north of $200,000 per year and marketing managers on poverty wages of $170,000 per year.

Is it not obvious which ends of the Smile Curve are adding the most value? Just look at the proles in Foxconn’s Zhengzhou plant in China – 180,000 workers assembling Apple products for $3.89 per hour plus dormitory housing.

What is that? $10,000-14,000 per year? And what do manufacturing process engineers make? According to Doubao, mid-level process engineers at Foxconn are paid $21,000 per year, with senior engineers making $55,000 per year.

But now, all of a sudden, everyone is telling us that the US must re-industrialize. America has grown dangerously dependent on Chinese manufacturing, they say. The US has lost the necessary skills for an advanced industrial economy, they say. The loss of manufacturing capacity is a national security risk, they say.

Fair enough. Message heard. Loud and clear. The economic efficiency promised by the Smile Curve may need to be sacrificed a bit, just to have a little more slack in the system. Just so China can’t turn the screws or, god forbid, think industrial capacity is advantageous in a kinetic Taiwan scenario.

So how come it is so hard? If manufacturing resides at the bottom of the Smile Curve, where the least value is added, should it not be the easiest part of the value chain to replicate?

The higher-skilled workers are, of course, the software engineers, designers and marketing managers in Cupertino. Manufacturing is low-value add. It says so right there, in the cost breakdown of an iPhone, in the compensation of a Foxconn Zhengzhou factory worker and process engineer.

In multiple interviews, Elon Musk explained that manufacturing was two to three orders of magnitude more difficult than design – that’s 100x to 1,000x more challenging. All the effort expended on design is ultimately a rounding error of the engineering required to manufacture a product.

Old-school engineers understand this intuitively. Design engineers and manufacturing engineers exist in professional tension. The manufacturing guys made the final call because they were downwards compatible, able to do the work of the design engineers, but not vice versa.

The value of manufacturing only appears low on the Smile Curve because the sell-off of American assets has distorted value capture. The globalization trade, turbocharged by China’s 2001 WTO accession, has been to exchange American assets for Chinese goods.

Software engineer, designer and marketing manager compensation in Cupertino is being priced to the value of US assets sold and is not, in all likelihood, doing higher-level work, as many surely believe.

All of this is an expression of the Dutch disease and Baumol’s law. Dutch disease refers to the Netherlands’ economy in the 1950s after natural gas was discovered in the North Sea. The resulting boom strengthened the currency and redirected capital to the energy industry, resulting in deindustrialization and loss of manufacturing competitiveness.

Rather than selling one natural resource like natural gas, the American economy has been living beyond its means by financializing its economy and selling off all kinds of assets – stocks, bonds, treasuries, agencies, real estate, derivatives and now cryptocurrencies.

Baumol’s law is typically used to explain why wages and prices in labor-intensive, low-productivity service sectors (e.g., the arts, healthcare, education) rise with the overall economy, even without productivity gains, making them ever more expensive.

Under Baumol’s law, unproductive sector wages must be benchmarked to booming sector wages or risk losing workers. After all, Silicon Valley needs school teachers and nurses to remain school teachers and nurses and not all go to coding boot camp.  

Globalization and Dutch disease have corrupted Baumol’s law in that it has benchmarked the value and compensation of not especially difficult jobs – software engineering, design, marketing – to the value of America’s asset sales.

Outsourced manufacturing was not the easy, low-end part of the value chain. It was, in fact, the hardest, highest-skilled and most difficult to replicate part of the value chain. In a Dutch disease environment, that is precisely why it was outsourced.

By living off asset sales, America was able to keep the fun, cushy and “creative” e-mail jobs whose value was inflated by Dutch disease and Baumol’s law while abandoning the difficult, dirty, intricate, frustrating and labor-intensive jobs to China.

The proof of all of this is downward compatibility. As it turns out, Xiaomi and Huawei phones are every bit as whizbang as the iPhone. There are, in fact, over a dozen Chinese mobile phone producers who figured out how to do R&D, design, branding and marketing.

Similarly, China incubated over 100 electric vehicle (EV) makers whose savage competition is revolutionizing the global car industry. They have raised the bar so high that Western markets have thrown up massive tariffs in a panic to protect legacy car makers.

Moving from manufacturing to R&D, design, branding and marketing is going downhill–downwards compatibility. Apple and legacy car companies struggle to recreate China’s manufacturing capacity elsewhere because moving from R&D, design and marketing to manufacturing is going uphill–downwards compatibility goes the other way.

It is not so much that the Smile Curve was erroneous. It did (and still does) help companies determine where to focus to maximize shareholder value. The Smile Curve is not, however, measuring skills, difficulty or replicability. It is, in fact, revealing the effects of Dutch disease and Baumol’s law in a globalized world.

US re-industrialization will not be a trivial endeavor. America has let the most difficult skills on the value chain wither. Curing Dutch disease would likely require the US to stop the assets-for-goods trade, which, in the process, will reprice tech bro and Wall Street compensation closer to Zhengzhou process engineer levels.

This may be a difficult pill to swallow for those who, for decades, have convinced themselves of their value based on the misleading Smile Curve.