Sequoia Capital: The Second Chapter of Generative AI

The rise of generative AI sparked a frenzy of excitement and unleashed an unprecedented wave of innovation and enthusiasm in the technology and investment communities.

The pinnacle of generative AI has been brewing for decades:

Sixty years of Moore's Law have provided the computational power to process trillions of data points. Forty years of the internet (accelerated during the COVID-19 pandemic) have generated training data worth trillions of tokens. Twenty years of mobile internet and cloud computing have put a supercomputer in the hands of every individual. In other words, decades of technological progress have created the necessary conditions for the takeoff of generative AI.

The emergence of ChatGPT has ignited a spark, unleashing an innovation density and passion that has not been seen in years, perhaps since the early days of the internet. In Silicon Valley, this overwhelming excitement is particularly intense, where AI researchers have achieved rock star status. Every weekend, hacker spaces are filled with new autonomous agents and conversational chatbots. AI researchers have transformed from the well-known "garage hackers" to a special forces unit wielding billions of dollars worth of computing power.

However, the market's excitement for AI quickly turned into near hysteria. Suddenly, every company became an "AI co-pilot." Investors found themselves caught in an unsustainable funding and talent war, as well as a frenzy of GPU purchases.

As expected, cracks soon began to appear: artists, writers, and musicians claimed that generative AI works infringed on copyright. Debates about ethics, regulation, and the imminent arrival of superintelligence plagued governments worldwide. Perhaps most concerning was the rumor circulating within Silicon Valley that generative AI was not living up to its promise. These products fell far short of expectations, as evidenced by poor user retention rates, and user demand for many applications began to plateau.

Such noise is not uncommon in the tech industry. In 1998, when the internet was just beginning to gain popularity, a renowned economist declared, "By 2005, it will become clear that the internet's impact on the economy is no greater than that of the fax machine."

Undoubtedly, despite the noise, hysteria, uncertainty, and dissatisfaction, generative AI has had a more successful start than SaaS. Not only has funding skyrocketed, but a range of well-known applications has also emerged.

However, these early signs of success do not change the reality that many AI companies lack product-market fit or sustainable competitive advantages, and the overall prosperity of the AI ecosystem is unsustainable.

What will the second chapter of generative AI look like?

According to Sequoia Capital, the first year of generative AI represents the "first chapter" of this technology, characterized by technological elimination. The market discovered a new tool, large-scale language models, and launched a wave of novel applications that served as lightweight demonstrations of the new technology.

Sequoia Capital believes that the market is now entering the "second chapter," which will be driven by customer support. In this second chapter, the focus will shift more towards solving the problems faced by humanity. These applications are fundamentally different from the first batch of applications, as they tend to use the underlying model as part of a more comprehensive solution rather than the entire solution. The new applications will introduce new editing interfaces to make workflows more sticky and improve output.

Sequoia Capital believes that the market has begun to transition from the "first chapter" to the "second chapter". Companies entering the "second chapter" include Harvey, which is developing custom law master's programs for elite law firms; Glean, which is indexing workspaces to make generative AI more relevant in the workplace; and Character and Ava, which are creating digital companions.

Sequoia Capital's updated generative AI market map is as follows. Unlike last year's map, Sequoia Capital has chosen to organize this map by use case rather than model pattern. This reflects two important drivers of the market: generative AI evolving from a technical tool to practical use cases and value, and the increasingly multimodal nature of generative AI applications.

Where does generative AI stand now?

Sequoia Capital believes that generative AI is not lacking in use cases or customer demand. Users desire AI to make their work easier and improve their work products, which is why they are so enthusiastic about these applications.

But will users stick around? Not necessarily.

User engagement tends to decline after the initial excitement. Some of the best e-commerce websites have monthly active users and daily active users accounting for 60-65% of total users; WhatsApp's ratio is 85%. In comparison, the median active users of generative AI applications is 14%. This means that users have not yet discovered enough value in generative AI products to use them daily.

In short, the biggest problem with generative AI is not finding use cases or demand, but proving its value. In other words, how can these applications change people's work and lives?

Sequoia Capital believes that currently, generative AI is still in its "awkward teenage years". Even if the products are not meeting expectations, the flaws are often fixable.