text| Juny Edit| VickyXiao
"Recently, for the AI project, our boss's current instruction is to kill three thousand by mistake rather than let one go." Lucas, an investor, told Silicon Starman. In recent months, in order to get more exposure to cutting-edge technologies and excellent projects related to generative AI, he has frequently traveled to and from offices in China and the United States, and has been brushing up on AI-related knowledge in his spare time.
Technology companies roll up big models, and investment circles have to follow suit. Since the advent of ChatGPT, AI has been pulled into the fever zone from the cold winter of investment. In the past few years, the tide of capital flowing to Web3 and metaverse has rapidly turned around and flocked to artificial intelligence. After missing out on OpenAI, everyone is frantically looking for the next AI company that can change the world.
In addition to the beginning of all in AI by major technology companies, young developers and startups in the field of generative AI in Silicon Valley have begun to flock to the Hayes Valley neighborhood in downtown San Francisco, and a new AI-driven "Cerebral Valley" is emerging.
Hot money pouring in, AI startups "barbaric growth"
How hot is financing in the field of AI now?
According to research firm PitchBook, in 2018, when OpenAI first released the initial version of the GPT model, the total annual funding of the entire generative AI field was only a measly $408 million, and after the release of ChatGPT last year, this figure climbed to $4.5 billion in 2022. And this year, in the first quarter alone, startups in generative AI have announced about $11.6 billion in deal funding involving 46 deals.
Analysts predict that on current trends, global investment in generative AI will easily be several times higher this year than last year.
Image from Picthbook
AI projects have become popular across the board, from past projects chasing funds to now people chasing projects, and sometimes entrepreneurs even rely on an unformed idea to attract investors to rush to throw money, and the valuation of the project skyrockets.
For example, two former Google AI researchers, Niki Parmar and Ashish Vaswani, previously co-authored a paper on the architecture of Transformer's deep model in 2017, which caused a lot of repercussions in the industry. At the beginning of this year, when investment institutions heard that they were leaving Google and wanted to start an independent business, they all rushed to the door. Without a business plan, a clear product idea, or even a name for the company, they received several investments that were valued at $50 million.
For example, LangChain, an open-source library of software developer tools, raised $10 million at a valuation of $50 million without any revenue, and just a few weeks later, Sequoia Capital led the investment of about $20 million, making LangChain's valuation quadruple to about $200 million.
In the current investment winter, generative AI has become a sudden, contrarian financing sector, which has not only attracted AI researchers to the sea, but also entrepreneurs from other tracks have also turned.
One of the obvious shifts is that many former Web3 and metaverse believers have begun to convert to AI. The number of participants at last year's various Web3 conferences has plummeted this year, and the various discussion groups now seem to seamlessly become AI groups.
"I am going to do product entrepreneurship in AI next, and there are already investors who are willing to invest in us." An entrepreneur who previously started a metaverse and received millions of dollars in investment recently told Silicon Star. But when Silicon Star asked what kind of product to do, he said that he had not thought very clearly at present, but it was definitely the direction of AI.
As for this sudden turn, he said that on the one hand, it is too difficult to obtain financing in the metaverse, and the obstacles to the further development of the original product are too great, on the other hand, he is optimistic about the commercial application of AI technology in the next few years, and believes that the future development potential is greater.
"Now the expectation of AI is definitely stronger than Web3 and the metaverse, and as far as I know, even if they don't completely switch to AI, people are now trying to connect existing products with AI."
Since the fourth quarter of last year, startups in the field of AI have sprung up. According to Crunchbase, in the past six months, the proportion of seed and previous funding for AI startups worldwide has reached about 70%, and many of the companies that have received investment are newly established, and the technology and products have not yet been released or tested.
Data from Crunchbase, Silicon Star Man Mapping
Silicon Valley's new AI stronghold – Cerebral Valley
Investors are desperate to bet on the next big technology in AI, and entrepreneurs are hoping to be the next OpenAI. In Silicon Valley, this wave of AI investment and entrepreneurship is particularly fierce.
In recent months, meetups, events and conferences on generative AI have been held in Silicon Valley, and there is a wave of excitement in the air that feels like a return to the entrepreneurial boom of the early days of the mobile Internet. Hayes Valley, located in downtown San Francisco, has become the most concentrated area for young developers and entrepreneurs in the field of AI.
As more and more AI hacker houses appear in this area, more and more developers, entrepreneurs and investors, and various AI events and conferences are beginning to be placed in this area. So in San Francisco's young AI community, people recently nicknamed Hayes Valley — Cerebral Valley.
Image taken from Google Maps
Hacker houses have been in Silicon Valley for more than a few decades, and to some extent, they are a continuation of Silicon Valley's classic garage culture. Entrepreneurs gather in the "House of Hackers" to live, work, exchange technology, inspire and find like-minded partners, which in turn has spawned many innovative technology companies.
Today, in the various "hacker houses" in Hayes Valley, people's activities revolve around AI, sharing sessions, networking activities, hackathons are held in this small community, attracting more and more attention from industry practitioners, and investors are actively looking for potential stocks in Hayes Valley.
For example, NFX, a venture capital firm that has invested in several generative AI companies, recently equipped a secret bar with a red wall in Hayes Valley and transformed the venue's garage into an event space with a dance floor, inviting AI developers to meet irregularly. Some investors see AI as a very strong area of connection, and have publicly said that if you start an AI startup in San Francisco, especially in the Hayes Valley, it will be a great advantage.
Image from Bloomberg Beta investor Amber Yang Twitter
In March, an invitation-only AI summit was held in Hayes Valley for the first time under the name Cerebral Valley. On the list of participants, we can see some of the hottest entrepreneurs in the field of AI recently, such as Hugging face CEO, Runway CEO, Stability AI CEO, Anthropic chairman, etc., as well as a group of top Silicon Valley venture capitalists such as Sequoia and Founders Fund. More and more AI events in the field are starting to use "brain valley" as a venue.
|Technology giants stand on the platform, new AI unicorns begin to "grab territory"
Driven by both technology and capital, entrepreneurs are pouring in, and startups in the field of AI are also growing rapidly.
According to CB insights, by the end of last year, there were only 6 unicorns in the global generative AI field, including OpenAI, Hugging Face, Lightricks, Jasper, Glean and Stability AI, but by the beginning of May this year, the number of generative AI unicorn companies had reached 13. In other words, in less than half a year, the number of generative AI unicorn companies has doubled.
However, from the current geographical distribution of these 13 generative AI unicorn companies, although the global characteristics are strong, Silicon Valley is still the most concentrated area. In addition to Cohere's headquarters in Toronto, Hugging face, Runway and Glean in New York, Lightricks in Israel, Jasper in Austin, Texas, and Stability AI in London, the remaining six are all from the Silicon Valley area.
And behind these unicorns, many have giants. Among them, the well-known king combination of Microsoft and OpenAI, in addition, Google, Amazon, Apple and other technology giants are also taking rapid action to expand their AI territory in the form of self-research + investment.
List of unicorns in generative AI as of May 8, 2023
For example, in addition to accelerating the development of its own PaLM model and chatbot Bard, Google has also continuously injected capital into OpenAI's biggest competitor Anthropic this year. Having invested $300 million in February, it participated in Anthropic's $450 million new round last week. According to investor presentations seen by the Wall Street Journal, Anthropic expects to raise up to $5 billion over the next two years to advance its big-language model, Claude, in competition with ChatGPT.
Amazon is not idle, while busy building a generative AI team and launching its own big language system Amazon Bedrock based on AWS, on the other hand, Amazon also joined hands with Hugging Face to provide comprehensive support for Hugging Face to build a product BLOOM that competes with ChatGPT, Bard and other language models on AWS.
Although Apple has not been high-profile in AI investment recently, it is also quietly acquiring some AI startups. For example, in February this year, it bought WaveOne, a California-based AI algorithm company, and its products are already working online in multiple of Apple's machine learning businesses. According to statistics, in the past few years, the number of AI startups acquired by Apple ranks first among technology giants.
In addition to technology giants, large investment funds such as Red Shirt, A16Z, and SoftBank are also betting on these top AI startups to become their behind-the-scenes "financial masters" and talkers.
For these rising AI "upstarts", they have to face the fact that the cost of training large-scale AI models is too high. According to a recent report by research firm EpochAI, the average cost of training various machine learning models running generative AI could grow to $500 million by 2030 due to the ever-increasing amount of data on the internet.
But so far, most AI startups have not found a clear path to profitability. Compared to the huge operating costs, the revenue brought by existing products is only a drop in the bucket. Therefore, these AI companies have to rely on a lot of capital and application support if they want to grow, and from this point of view, the AI war may eventually be a game between giants in the future.
There is no denying that we are experiencing an important technological inflection point. Generative AI is like a gold rush, fueled by capital, with some great products and companies poised to take off.
But at the same time, like a lot of concept hype in the past, a large number of startups and funds are also violently disrupting the pool of AI commercial applications. Only after the big waves will we really witness the birth of some AI giants.
Article source: Silicon Star Man