Innovation is slowing down—and Big Tech is to blame

In 2005, years before Apple’s Siri and Amazon’s Alexa came on the scene, two startups—ScanSoft and Nuance Communications—merged to pursue a burgeoning opportunity in speech recognition. The new company developed powerful speech-processing software and grew rapidly for almost a decade—an average of 27% per year in sales. Then suddenly, around 2014, it stopped growing. Revenues in 2019 were roughly the same as revenues in 2013. Nuance had run into strong headwinds, as large computer firms that were once its partners became its competitors. 

Nuance’s story is far from unique. In all major industries and technology domains, startups are facing unprecedented obstacles. New companies are still springing up to exploit innovative opportunities. And these companies can now tap into an extraordinary flood of venture capital. Yet all is not healthy in the startup economy. Innovative startups are growing much more slowly than comparable companies did in the past. 

To support MIT Technology Review’s journalism, please consider becoming a subscriber.

Surprisingly, a major culprit is technology—specifically, proprietary information technology in the hands of large firms that dominate their industries. We’re accustomed to thinking of technology as creating disruption, in which innovations introduced by smaller, newer companies enable them to grow and ultimately replace older, less productive ones. But these proprietary technologies are now suppressing industrial turnover, which has declined sharply over the last two decades. This loss of dynamism has broad negative implications for the US economy. It has slowed the growth of innovative firms. And researchers have tied that slower growth to substantially slackened productivity growth, which affects the entire economy, all the way down to personal incomes.


Nuance began in 1994 as a spinoff from SRI, a Stanford laboratory that had developed speech-­recognition technology for the US government. ScanSoft was a Xerox spinoff. Before the two merged in 2005, speech recognition was constrained by computer processing power. Systems recognized only limited vocabularies, though they nevertheless proved useful in narrow commercial applications such as telephone customer support centers and transcription of medical records.

By the late 2000s, things had changed. As computers became more powerful, Nuance was able to develop a major innovation: “large vocabulary continuous speech recognition.” Now you could say anything about any topic, and the technology could accurately transcribe it in real time. Nuance used this technology in an app called Dragon Dictation, which Apple featured when it introduced the iPhone 3GS at its 2009 Worldwide Developers Conference. Once Apple validated the product, Samsung and all the other phone manufacturers wanted it. So did Google, Amazon, and Microsoft. Nuance grew rapidly, both by signing up these major customers and also through millions of individual consumers who purchased the iPhone app, which became the number-one business productivity application in the iTunes store. In 2011, Apple introduced Siri, which was based on Nuance technology. Nuance’s revenues grew to $1.7 billion in 2013.

But this growth was short-lived. Nuance wasn’t the only one to realize that voice was poised to become a prime channel for human interaction with computers and cloud services. Voice recognition was no longer just about dictating text but about shopping, searching for information, selecting music and video entertainment, controlling appliances, and much more. It was fast, hands-free, and—compared with the keyboard and mouse—a much more natural way for humans to communicate.

Big Tech started plowing big money and talent into this opportunity. Apple invested in developing its own systems, Amazon pursued its Alexa voice assistant, and Google followed quickly with its Home Assistant. And those companies successfully raided Nuance’s talent pool, bringing top people into their folds. Amazon now has over 10,000 engineers working on Alexa products, more than 10 times the number of core R&D employees Nuance had at its peak.

In addition to their financial resources, the big companies also had the advantage of large customer bases, complementary products, and vast amounts of data at their disposal, enabling them to continually improve their voice-recognition systems. Today there are 300 million Alexa devices installed; Google handles 5.6 billion searches each day on average, and half its users report using voice for search. Amazon has a thriving ecosystem where third-party developers add new “skills” to Alexa—over 100,000 of them, ranging from playing specific radio stations to telling jokes. In addition, Amazon has licensed the Alexa far-field technology to appliance manufacturers, which use it to control dishwashers, clothes washers and dryers, and vacuum cleaners. 

Nuance could not compete on this battlefield. It retreated to focus on market niches such as health care before being acquired by Microsoft in 2021.


What happened to Nuance is not just a retelling of the old story of large firms out-investing startups. Across a wide range of industries, dominant firms are employing large-scale information systems to outflank their competitors, including innovative startups. They are using proprietary software to better manage complexity and thus differentiate themselves from rival firms. And this has allowed them to increase their market dominance and avoid being overtaken by rivals.

In retail, Walmart’s inventory management and logistics software allows it to stock its stores with a far greater selection of products at lower cost, tailor each store to local needs, and respond quickly as demand changes and hot products emerge. Using large data systems, leading financial companies tailor credit cards and home equity loans to individual consumers on a massive scale and then target the marketing of these products. Even the top waste-management companies and health insurers are making large investments in proprietary software to beat their competition. In aggregate, firms (excluding those whose product is software) now invest over $240 billion in their internal software each year, up from $19 billion in 1985. Large firms account for most of that change. The top four companies in each industry, ranked by sales, have increased their investment in their own software eightfold since 2000, far more than even second-tier firms.

And these investments have paid off. Since the 1980s, the top four firms in each industry have increased their market share by 4% to 5% in most sectors. My research shows that investments in proprietary software caused most of this increase. 

This greater industry dominance by top firms is accompanied by a corresponding decline in the risk that they will be disrupted, a prospect that has obsessed corporate managers ever since Clayton Christensen’s The Innovator’s Dilemma came out in 1997. At the time Christensen wrote his book, disruption was on the rise. But since about 2000—when top firms started their investment spree in proprietary systems—this trend has declined sharply. In a given industry, the chance that a high-ranking firm (as measured by sales) will drop out of one of the top four spots within four years has fallen from over 20% to around 10%. Here, too, investments by dominant firms in their internal systems largely account for the change. While some new technologies disrupt entire industries—think of what the internet did to newspapers or DVDs—others are now suppressing the disruption of dominant firms.

How does this happen, and why does it apparently affect so much of the economy? It is because these business systems address a major shortcoming of modern capitalism. Beginning in the late 19th century, innovative firms found that they could often achieve dramatic cost savings by producing at a large scale. The shift dramatically reduced consumer prices, but there was a trade-off: in order for companies to achieve those large volumes, products and services needed to be standardized. Henry Ford famously declared that car buyers could have “any color so long as it is black.” Retail chains achieved their efficiencies by providing a limited set of products to their thousands of stores. Finance companies offered standard mortgages and loans. As a result, products had limited feature sets; stores had limited selection and were slow to respond to changing demand; and many consumers could not get credit or obtained it only on terms that were costly and not suited for their needs.