Sequoia Capital USA: GenAI is a 10x productivity revolution

Compiled by: Lavida

In September 2022, Sequoia Capital USA released Generative AI：A Creative New World 研究引发了第一波关于 Generative AI 的讨论，而紧接着 ChatGPT、GPT-4 的问世加速了 GenAI 领域的发展。在红杉美国的 AI Ascent 大会上，几位合伙人就对过去这一年半中 GenAI 的发展进行了相当系统的回顾，GenAI 领域的进步远比人们预想当中要更迅猛。

Unlike previous rounds of AI, GenAI has achieved amazing results in the past year: the total revenue generated by GenAI in the first year after its emergence was about 3 billion US dollars, which does not include the revenue indirectly generated by technology giants and cloud vendors through AI. It took the SaaS industry nearly 10 years to reach this level. In terms of specific implementation, GenAI has already created real profits in industries or scenarios such as customer service, law, and writing.

Although the explosion of the application layer is not as optimistic as the market predicted a year ago, several partners of Sequoia Capital in the United States also pointed out that with the emergence of more intelligent foundation models, such as Sora, Claude-3 and other new models have been launched recently. The PMF cycle of AI products will definitely accelerate in the future. In addition, it takes a process for new technologies to mature from their emergence. The emergence of revolutionary applications also takes time. In the mobile Internet era, representative applications such as Instagram and Doordash appeared several years after the launch of iPhone and App Store.

The following is the table of contents of this article. It is recommended to read it in detail based on the key points.

01 Why Now: From cloud computing to AI

02 Today: AI is Everywhere

03 Future: Everything is Generated

The market has experienced a complete AI Hype Cycle: There was over-hype during the bubble period, and there was disappointment and doubt during the trough period, but now the market is climbing back to the Plateau of Productivity. People are gradually realizing that LLM and AI really work through the three links of creation, reasoning, and interaction, and these capabilities have also been integrated into applications in various fields for our use.

Three capabilities of AI: creation, reasoning, and interaction

AI already hascreateandreasoningFor example, GenAI can generate text, images, audio and video, chatbot can answer our questions, or help us with multi-step task planning like Agent, which is something that no previous software could do. It also means that software can handle creative tasks of the right brain and logical tasks of the left brain at the same time - this is the first time in history that software canInteract with humans in a human-like way, which is of great significance to the business model.

Why Now: From cloud computing to AI

Sequoia partner Pat Grady answered the question "Why has AI technology exploded recently" by reviewing the development of the cloud industry over the past 20 years.

Pat believes that cloud computing is a major shift in the technology field, which has overturned the past technology landscape and brought about new business models, applications, and ways of human-computer interaction.In 2010, when the cloud industry was still in its early stages, the total market value of global software was about $350 billion, of which cloud software accounted for only about $6 billion. But by last year, the total size of the software market had grown from $350 billion to $650 billion, and cloud software revenue reached $400 billion. This means that in 15 years, the CAGR of cloud software has remained at 40%, achieving amazing growth.

The cloud is a good analogy for AI. The cloud can replace traditional software because it has more interactive capabilities similar to humans; similarly, today's AI technology has reached new heights in creativity, logical reasoning, and human-computer interaction.In the future, one of the great opportunities for AI will be software replacing services. If this change can be achieved, the market potential of AI will not be hundreds of billions of dollars, but tens of trillions of dollars. It can be said that we are standing at the greatest and most unlimited value creation potential in history.

Technological changes since the 1960s and representative companies

As for why he believes that now is an important time to participate in AI, Pat Grady mentioned that Sequoia has witnessed and benefited from several technological changes in history since its establishment. In this process, the team also has a clear understanding of how different technological waves influence each other and drive the world forward:

• 1960s: Sequoia founder Don Valentine was in charge of marketing at Fairchild Semiconductor. The origin of the name "Silicon Valley" is also directly related to Fairchild Semiconductor's silicon-based transistors;

• 1970s: People built computer systems based on chips;

• 1980s: Network technology connects PCs together, and the software industry is born;

• 1990s: The birth of the Internet changed the way people communicate and consume;

• 2000s: The Internet has gradually matured and started to support complex applications, and cloud computing emerged;

• 2010s: With the popularity of mobile devices, the mobile Internet era has arrived, which has once again changed the way we work.

Each technological wave is superimposed and evolved on the basis of the previous one. Although the concept of AI appeared as early as the 1940s, it was not until recent years that AI turned from an idea and a dream into reality, began to be commercialized, and solved practical problems in people's daily lives. The prerequisites for achieving this breakthrough include:

• Low-cost and sufficient computing power;

• Fast, efficient and reliable network;

• The global penetration of smartphones;

• the trend towards onlineness accelerated by Covid;

• All of the above processes bring a large amount of data to AI.

Pat Grady believes thatAI will become the theme in the next 10-20 years, Sequoia has a strong belief in this, although this hypothesis remains to be confirmed.

Representative companies from the Cloud and Mobile to the AI era

Regarding the future industry landscape of AI, Pat Grady first summarized the emerging trends in the Cloud and Mobile eras.Companies with revenue exceeding $1 billion (as shown on the left side of the above picture). Although the rightmost side represented by AI is almost blank now, it also symbolizes the huge potential value and opportunities in the current market.Pat Grady predicts that in the next 10-15 years, the space on the right will be filled by 40-50 new company logos, which is exactly the opportunity that makes them excited.

Today: AI is Everywhere

Sequoia partner Sonya Huang first reviewed the development of AI in the past year in areas such as customer service, law, programming, and video generation.

AI applications in various fields

2023 is a pretty big year in the history of AI. A year and a half after ChatGPT was released, the industry has been undergoing drastic changes. Last year, everyone was talking about how AI would revolutionize different fields and provide amazing productivity gains, but now AI has become the focus of attention.

Klarna CEO Sebastian Siemiatkowski X Tweets

In the field of customer service,KlarnaCEO Sebastain once publicly stated that Klarna is now using OpenAI to handle 2/3 of customer service inquiries, and AI has replaced the equivalent of 700 full-time customer service staff. There are currently tens of millions of call center agents in the world. Against this background, Sonya believes that AI has found PMF in the customer service market.

legal serviceAn industry that was considered the least willing to embrace technology and take risks a year ago has now emerged as HarveySuch companies can automate many of lawyers' tasks, from daily paperwork to advanced analysis.

For example, in the field of programming, after one year, we have developed rapidly from using AI to write code a year ago to having independent AI software engineers. There are also AI video generation companies like HeyGen, which can help people generate Avatars to participate in Zoom meetings.

Sequoia's Pat Grady uses HeyGen to generate an avatar for a Zoom meeting

GenAI’s 10x Growth

Comparison of AI and SaaS revenue growth rates

According to estimates,GenAI generated a total revenue of about $3 billion within a year of its emergence, not including the revenue generated by technology giants and cloud service providers through AI. In contrast, it took SaaS nearly 10 years to reach this level.It is precisely because of this speed and scale that everyone is more confident that GenAI will continue to exist.

Actual user scale of major GenAI products

As can be seen from the above figure, customers’ demand for AI is not limited to one or two applications, but is all-encompassing. Many people know how many users ChatGPT has, but when observing the revenue and usage data of many AI applications, it is found that now, whether it is to B or to C, startups or existing technology companies, many AI products have found suitable PMF in various industries, and the application scenarios have become very diverse.

Funding share of foundation model and application layer

From the perspective of investment distribution, the imbalance of capital allocation is a significant problem. If GenAI is compared to a cake, the bottom layer of the cake is the foundation model, and the middle is the developer tools and infra, the top layer is the application. A year ago, people expected that a large number of new companies would emerge in the application layer due to the progress of the foundation model layer. But the actual situation is the opposite.More and more foundation model companies are emerging and have raised a lot of money, but the application layer seems to be just getting started.

AI's $200 Billion Question

Last year, David, a partner of Sequoia Capital in the United States, published a discussion on AI's $200 Billion Question. If we look at the current investment in GPUs, last year only Nvidia chipsThe AI industry has spent about $50 billion on AI, but the confirmed revenue is only $3 billion. These data show that the AI industry is still in its early stages, with a low input-output ratio, and there are still many practical problems that need to be solved.

MAU, DAU and next month retention rate of AI products and mobile applications

Although the number of users and revenue of AI products seem impressive, they are still far lower than mobile applications in terms of DAU, MAU and next month retention rate.Many users mentioned in user surveys that there is a gap between expectations and experience of AI applications. Some product demos look cool, but they are not very practical in practice, which also leads to users not using them for a long time.

Improvement of basic model capabilities

Although these are objective problems, they are also opportunities. Last year, companies invested heavily in GPUs, which led to a smarter foundation model. Sora, Claude-3, and Grok all show that the baseline intelligence level of AI is improving, so the PMF of AI products will accelerate in the future.

The Evolution of the iPhone and the App Store

It takes time for new technologies to mature, and it also takes time for groundbreaking applications to emerge. Take the iPhone as an example. Many applications in the early days of the App Store were very primitive, just showing new technologies, but not really solving problems or creating value. Small games like flashlights or beer drinking later became built-in system applications or dispensable gadgets.Truly influential apps like Instagram and Doordash didn’t appear until several years after the iPhone and the App Store were launched.

AI technology isXiaobai NavigationMany AI applications on the market are still in the demo or early exploration stage, just like the early applications in the App Store, but perhaps the next generation of legendary companies has already emerged.

AI has been widely used in various scenarios, including AI customer support,AI FriendshipCustomer service is one of the first AI application scenarios to truly achieve product PMF in enterprises. Klarna is not an exception, but a general trend. AI friendship is one of the most surprising application scenarios of AI. Its user numbers and usage indicators show that users have a strong love for it. In addition, cross-departmental and cross-functional enterprise internal knowledge sharing (Horizontal enterprise knowledge) applications also have great potential.

Future: Everything is Generated

4 predictions for AI in 2024

Based on the above analysis, several partners of Sequoia also made predictions on the development of AI in 2024.predict.

Prediction 1: Copilot will gradually transform into an AI Agent.

In 2024, AI will transform from a copilot that assists humans to a real replacement for some human jobs. Agent. AI will become more like a colleague rather than just a tool, which has already begun to emerge in industries such as software engineering and customer service.

Prediction 2: Models will have stronger planning and reasoning capabilities.

Many people criticize LLM for just repeating statistical patterns in previous data instead of really doing in-depth thinking and logical reasoning. This situation will be improved through new research directions. Some studies are trying to make the model better at reasoning and game-style value iteration. These methods allow the model to have a certain amount of "thinking time" before making a decision. These attempts are expected to make AI more capable of performing higher-level cognitive tasks, such as planning and reasoning, next year.

Gameplay-style value iterationIt is a concept borrowed from the field of reinforcement learning, which means that the model can evaluate the long-term value of different actions and plan future actions based on these values, similar to strategic thinking when playing chess or games.

Prediction 3: LLM will become more accurate and will gradually expand from being mainly used in To-C entertainment applications to enterprise-level applications.

In To-C application scenarios, users don’t particularly care about AI errors because people mainly use AI for entertainment, but when AI is used in enterprise applications, especially in high-risk fields such as medical care and defense, accuracy and reliability become critical. Researchers are developing various tools and technologies such as RLHF, Prompt Training, and vector databases to help LLM achieve "five nines" (99.999% uptime) high accuracy and reliability.

Prediction 4: A large number of AI prototypes and experimental projects will be put into use.

Many AI prototypes and experimental projects are expected to enter the market in 2024. Unlike the experimental stage, when the product is actually used by users, a series of factors such as latency, cost, model ownership, and data ownership management need to be considered, which also means that the focus of computing is expected to shift from pre-training to inference. Therefore, 2024 is a critical year. People have high expectations for these products and must ensure that this transition process is correct.

The long-term impact of AI

Judgment 1: AI is a large-scale, cost-driven productivity revolution.

There are many types of technological revolutions, including the communications revolution brought about by the telephone, the transportation revolution brought about by trains, and the productivity revolution brought about by agricultural mechanization.AI obviously brings about a productivity revolution.

The productivity revolutions in history all have similar patterns: people first use tools, then people cooperate with machines, and finally humans collaborate with collaborative, networked tools. This shows that the development of AI will go through a process of gradual evolution from a single point to a highly integrated network, which will greatly change the way we work and produce.

Historical Change from Scythe to Combine Harvester

In the field of farming, humans have been using sickles for more than 10,000 years, and then the mechanical harvester was invented in 1831. Now we have a complex network of combine harvesters composed of tens of thousands of machine systems, and a single machine in the system is an agent.

There is a similar pattern in the field of knowledge work and writing. The initial tools for knowledge work were only pen and paper, which later developed into programming, and now computers and IDEs can assist software development on a large scale. Software development will no longer be an isolated process, but a series of machine networks that collaborate to build complex engineering systems, with multiple agents jointly completing code generation.

Writing used to be purely manual, but later humans collaborated with machine assistants, and now many tools can be used for collaboration. For example, the AI assistants that people use now are not only GPT-4, but also Mistral-Large and Claude-3, and they can verify each other to get better answers.

AI brings about a general reduction in costs in various industries

The impact of the productivity revolution on society is universal and far-reaching. From an economic perspective, this means that costs can be significantly reduced. The above figure shows The number of employees required per $1 million in revenue for S&P 500 companies is falling rapidly, and this change means we will be able to get work done faster and with fewer people. This does not mean we are doing less, but we are doing more in the same amount of time.

Technological progress in all fields throughout history has brought deflation. Take computer software as an example. Due to continuous technological innovation, the price of software is constantly falling.In the areas that matter most to society, such as education, healthcare, and housing, prices are rising much faster than inflation, and AI can help reduce costs in these areas.

Therefore, the first key judgment about the long-term impact of AI is:AI will be a massive, cost-driven productivity revolution, helping us do more with less in key areas of society.

Judgment 2: Everything can be generated

The second judgment mainly discusses what AI can do.

A year ago, Jensen Huang made a prediction that in the future, images will no longer be rendered, but generated. This means that we are moving from storing information as a matrix of pixels to representing it as a multi-dimensional concept.Take the letter "a" as an example. In the past, "a" was stored as the original data of ASCII code 97, but now computers no longer only focus on pixel representation, but understand the conceptual connotation of the letter as an English letter in a specific context.

What’s even more powerful is that the computer can not only understand this multidimensional representation and render it as an image, but also contextualize it, understanding the meaning of “a” as an English letter in a specific context, rather than just an isolated symbol. For example, when seeing the word “multidimensional”, the computer will not focus on the letter “a” itself, but understand the entire context and the meaning of the word.

This process is the core feature of human thinking. Just like when we learn the letter "a", we are not memorizing a pixel matrix, but mastering an abstract concept. This way of thinking can be traced back to Plato's theory of ideas 2,500 years ago. Plato believed that there is an eternal and unchanging world of ideas behind everything, and things in the real world are a reflection of the perfect form of the world of ideas. This is similar to the current AI learning process.

This has a huge impact on businesses. Companies have already begun to integrate AI into specific processes and KPIs. For example, Klarna, mentioned earlier, has used AI to improve customer support performance and create a high-quality customer experience by establishing an AI retrieval information system. This change is also accompanied by the emergence of new user interfaces, which may be completely different from the support communication methods we have used in the past.

This trend is significant because it means that businesses may eventually operate like neural networks, with various parts connected and working together, learning and adapting to each other in a self-optimizing way and constantly improving efficiency.

Taking the customer support process as an example, the above figure is a simple customer service process diagram. The customer service department has a series of KPIs, which are affected by factors such as cultural speech, language generation, and customer personalization. These factors form sub-patterns or sub-trees in the optimization item tree diagram, and finally form a hierarchical and interconnected system diagram, in which the feedback of language generation will directly affect the final KPI of the customer service. With this abstract method, the entire customer service process will be managed, optimized and improved by the neural network.

Consider the situation of enterprise customer acquisition. AI technologies such as language generation, growth engines, and ad customization and optimization can help enterprises better meet the needs of each customer. The interaction between these technologies can drive enterprises to self-learn and adapt like neural networks. Individuals will be able to complete more work, which will also give rise to more one-person companies.

The article comes from the Internet:Sequoia Capital USA: GenAI is a 10x productivity revolution

Related recommendation: EZswap Protocol: Gamified NFT DEX model, or the biggest beneficiary of ERC404

The business development model of EZswap Protocol is to use NFT as the underlying asset and build a decentralized trading platform that can handle all kinds of new assets. Author: Jonas Researcher Market Status Whether it is a bull market or a bear market, new asset types such as inscriptions areBlockchainIt is more favored by capital, and narratives are more likely to be recognized by the market. …