Asking hot and cold AI: Where are we now if the bubble is the ultimate destination?

Source: "Science and Technology Innovation Board Daily"

Editor Zheng Yuanfang

Image source: Generated by Unbounded AI‌

The generative AI boom has spread for more than half a year, and the global stock market fluctuates due to AI technology.

Previously, Rothschild bluntly stated "AI is overvalued", and he was "less and less sure" about adding positions in AI technology; later, Stability AI released a bold statement, thinking that AI "will is one of the biggest investment themes for the next few years," and the biggest bubble ever, but it hasn't even started yet.

Both long and short sides have their own reasons, but if you look closely at the opinions of all parties, it is hard to say that it is not "the ass determines the head"-the former has just reduced its holdings of Nvidia stocks at high prices, and the latter has developed a hot AI tool Stable Diffusion, and raised more than $100 million from investors including Coatue, Lightspeed Venture Partners, and others.

On the one hand, the traffic of AI robots such as ChatGPT may peak, and on the other hand, many U.S. stocks such as Nvidia have continued to rise and even hit new highs.

What is the nature of this AI technology revolution, and how will the secondary market affect it? If the bubble is the end of this AI boom, where are we now?

Primary and secondary market upsurge

Nvidia, as one of the most representative stocks of this AI craze, has set a record stock price a few days ago. On July 14, local time in the US stock market, Nvidia’s intraday stock price reached 480.88 US dollars per share. 211.22%**.

The U.S. stock prices of companies such as Advanced Micro Computer and Broadcom have also continued to rise in recent days. The former also reached a record high last Friday**, with an increase of nearly 260%** within the year.

Of course, in the primary market, the AI fever has not subsided.

In China, Zhou Bowen, the former head of AI at JD.com, Li Yan, the core figure of Kuaishou AI, and Li Kaifu, chairman and CEO of Sinovation Ventures, have started a group business. Overseas, many star projects such as Cohere, Runway, and Replit have emerged. According to data from CB Insights, an American research institute, 13 AIGC companies have become unicorns (startups with a valuation of more than US$1 billion), and the average time it takes them to start a business and become a unicorn has been cut in half.

** Chatbots losing their freshness? **

However, it must be admitted that doubts about “the ebb of AI fever” have emerged in the market.

The initial doubts originated from a news that "ChatGPT traffic peaked", but the recent decline in its website traffic has become more predicament-ChatGPT visits fell by nearly 10% month-on-month in June, and data shows that its user loss rate has soared to 20%.

Figure | ChatGPT traffic and participation in June (Source: SimilarWeb)

SimilarWeb warns that People seem to have lost their novelty with generative artificial intelligence chatbots, and artificial intelligence must now prove its value more than sit back and enjoy its success.

At the same time, it has also been pointed out that AI is not actually intelligent.

Gary Marcus, professor emeritus of psychology and neuroscience at New York University and founder of the machine learning company Geometric Intelligence, believes that current artificial intelligence systems are not very intelligent.

He said that the often touted neural networks in current artificial intelligence function completely differently from the neural networks of the human brain; while artificial intelligence machines can perform reflexive statistical analysis, they have little deliberate reasoning ability. While AIs are capable of learning, this learning revolves around word statistics and correct responses to cues; they don't learn abstract concepts, and unlike humans, AIs don't have internal models to understand the world around them.

** "A Real Revolution" or "Still Not Smart AI"? **

More investors are beating their hearts: Is this AI boom a hype bubble? Goldman Sachs released a 20,000-word report this month, inviting a number of analysts and industry insiders to express their views.

Professor Marcus mentioned above is a representative of the bearish faction. His answer to the question "Is the artificial intelligence hype too much?" is yes.

According to Marcus, the intelligence of AI systems has been greatly exaggerated. “Anyone who thinks AGI is just around the corner is almost certainly wrong.”** “AIGA may eventually happen, but today we’re nowhere near it, and no amount of investment is likely to change that. "**

However, More respondents are optimistic, seeing the development of AI technology as "a real change".

For example, Kash Rangan, chief software and Internet analyst at Goldman Sachs, pointed out that AIGC technology can create new content in the form of text, image, video, audio and code, and realize it through natural language rather than programming language, which is its key Transformative Features**.

Sarah Guo, founder of artificial intelligence venture capital firm Conviction, believes that previous iterations of artificial intelligence technology required humans to write deterministic code to perform specific tasks (“software 1.0”), or collect training data to train neural networks to complete specific tasks (“software 2.0”). ”), and now the basic model has natural language ability, reasoning ability, and general knowledge of the world, which reduces the burden on enterprises to collect training data, creates the era of “Software 3.0”, and helps enterprises improve efficiency.

Goldman Sachs Global Senior Economist Joseph Briggs estimates that the widespread adoption of AIGC technology in the United States and other developed countries can increase annual labor productivity growth by about 1.5 percentage points within 10 years, and eventually increase global annual GDP by 7%.

**Does the AI bubble appear? **

As for the secondary market, whether AI hype is overdone has also become a focus of debate.

After Rothschild reduced its holdings of Nvidia last month, its global chief investment officer, Benjamin Melman, once said frankly, "Will we increase our position in AI technology? Given that the valuation of AI is too high, we are less and less sure... If (valuations) continue to grow, we will even become more cautious."

Dominic Wilson and Vickie Chang, market strategists at Goldman Sachs, also reminded that in the past innovation-led productivity booms, such as electricity (1919-1929), personal computers and the Internet (1996-2005) ** after the widespread spread, stock prices and valuations The sharp rise in value turns into a bubble that eventually bursts**.

However, most analysts currently believe that the AI bubble has not yet appeared.

Goldman Sachs US Internet analyst Eric Sheridan pointed out that in the past few months, the vast majority of artificial intelligence-themed companies have outperformed the broader market, and their price-to-earnings ratios are still at relatively reasonable multiples**. Bubbles typically refer to enterprise value versus eyeballs/clicks, addressable market dynamics, or sheer euphoria as drivers of valuation, not P/E. So this is very different from previous tech bubbles.

Rangan also said that AI may not have entered the hype cycle, unlike other large technology cycles, such as the shift from distributed systems to cloud computing, where opposition from established players has slowed technology adoption, And this time the AI boom is being driven by the global tech giants — *“When technology providers agree that technological change is happening, it’s real. When customers start getting interested, it’s not hype.” *

Players in the industry also gave their own "bubble expectations" - Emad Mostaque, CEO of Stability AI, an open source artificial intelligence company, believes that AI will indeed develop into a bubble, and it will be the biggest bubble in history, "I call it '.AI (dot AI)' bubble", but this bubble "hasn't even started".

Mostaque said that AI "will be one of the biggest investment themes in the next few years" and that the total required investment could reach $1 trillion "because it is more important than 5G as an infrastructure of knowledge". However, he added that AI is still in the "early stages" of development and "not quite ready" for large-scale deployment in large industries such as financial services, "but we can see the value in it."

If the bubble is destined to appear but has not yet appeared, what AI investment opportunities are worth paying attention to at this point in time?

Rangan and Sheridan are optimistic about large technology companies that develop basic AI models, as well as "shovel-selling" companies such as semiconductor companies, cloud computing service providers, and infrastructure companies**.

In addition, Guo added that she is bullish on the opportunities in the entire AI field and is excited about the application prospects of AI. It is worth mentioning that, as an early investor, she pointed out that within a certain range, she did not pay so much attention to corporate valuation, but instead turned more attention to markets, products and entrepreneurs with broad upside potential.