Author | ZeR0
Edit | Desert Shadow
Zhidong reported on May 17 that overnight, a number of American generative AI startups were exposed to a shortage of funds:
Replit, an AI programming unicorn in San Francisco, announced a 20% layoff of 30 employees in the early hours of this morning.
Reka AI, a large language model startup, was revealed to be likely to be acquired by data storage and analytics company Snowflake for $1 billion.
Stability AI, an AI unicorn company that was in danger of selling itself or even going bankrupt, is fighting for a "life-saving money", with investors including Sean Parker, the first president of Facebook, and Prem Akkaraju, the former CEO of visual effects company Weta Digital.
As tech giants launch and upgrade a variety of generative AI services, both free and paid, AI startups face fierce competition for customers.
Due to the abundance of free AI products on the market, enterprise customers have generally slowed down or reduced their spending on business software. This has put AI startups that do not have an advantage in capital reserves to face the pressure of winning new orders and competing against competitors.
1. Is the generative AI industry shrinking? A number of start-ups cut 20% of their workforce
In the early hours of this morning, Beijing time, Amjad Masad, CEO of Replit, an American AI programming startup, posted an email on the social platform X, announcing that the company has laid off 30 employees, accounting for about 20% of its total workforce.
Founded in 2016, Replit has raised more than $220 million and is currently valued at more than $1 billion, making it a new generative AI unicorn.
Masad said in an interview that the company wants to serve more enterprise customers than individual developers; Replit layoffs are predominantly in non-technical roles, such as marketing, recruitment, and support.
He revealed that Replit still has a lot of cash and has not yet used the $100 million raised in the last funding round.
Not only Replit, but a number of generative AI startups have started layoffs in recent months, including Tome, Jasper AI, Deepgram, etc., and they have also shifted their business focus to enterprise users.
In July last year, one of the early winners of the generative AI boom, American AI unicorn Jasper AI, laid off an unspecified number of employees nine months after it raised $125 million at a $1.5 billion valuation. Its CEO, Dave Rogenmoser, said the layoffs were to "focus on gathering resources and becoming the best AI copilot for the marketing team."
Then in November, Deepgram, an American AI speech recognition and transcription startup, announced that it had laid off about 20% of its workforce, or about 20 people, including data scientists, researchers and engineers, which is already its second layoff in 2023.
Deepgram was founded in 2015 to sell speech recognition services to enterprise customers. The company's chief executive, Scott Stephenson, blamed the layoffs on rising interest rates, which led to a reduction in the company's capital, saying that "we have to be conservative and suppress the mindset of growth at all costs and focus instead on efficiency."
According to PitchBook, last fall, Deepgram announced that it had raised $47 million, for a total of $86 million in 2021. It is valued at $267 million. In an email to employees about the new round of layoffs, Deepgram executives cited tougher funding conditions for startups, macroeconomic challenges and the company's poor performance last year.
In mid-April, generative AI startup Tome rebuilt its team, laying off 20% of its 59 employees, including a consumer market team and product developers who are more focused on users of Tome's products for free.
Keith Peiris, co-founder and CEO of Tome, told foreign media that these consumer-centric product developers will be replaced by enterprise salespeople to target potential new customers, as well as developers who focus on B2B software products.
The founders of the AI startup, from Meta, initially positioned their products as a tool to attract professionals and consumers, but recently they have realized that the sales industry is more willing to pay for additional features such as sales pitches. According to Peiris, the new focus is on software that is "more information-oriented than aesthetic".
For example, Tome is developing an AI tool that can extract valuable information from SEC filings that could be useful for public companies' sales pitches. It can also connect with Salesforce data and can mine inside sales call records from past customers.
2. Stability AI lost 30 million in the first quarter and is negotiating new financing with the first president of Facebook
AI unicorn Stability AI has been in talks with at least one buyer in recent weeks about selling due to a cash shortage, The Information reported on Wednesday. Headquartered in London, UK, Stability AI employs around 180 people.
Stability AI privately revealed that the company's revenue last year was $8 million, up from about $1.5 million in 2022, according to people familiar with the matter. The first quarter of 2024 had revenue of less than $5 million and a loss of more than $30 million. But the numbers don't match the information made public by its founder, Emad Mostaque, in the fall of 2022, when he said that Stability AI was on track to generate tens of millions of dollars in revenue that year.
The person also revealed that Stability AI currently owes nearly $100 million in outstanding bills to cloud providers and other companies. That could mean the company doesn't have much cash left.
In the space of two years, venture capitalists have gone from flocking to rapidly shifting away from Stability AI.
Researchers at Stability AI are likely to be the main targets of the deal as well. Stability AI executives told potential buyers that the company has 70 to 90 researchers, according to people familiar with the deal negotiations. That number is likely to decline rapidly, and in recent months, Stability has lost a lot of key technical talent, including three founders.
However, its funding shortage may face some turnarounds.
A group of investors, including Sean Parker, the first president of Facebook, and Prem Akkaraju, the former CEO of visual effects company Weta Digital, have discussed investing in Stability AI, according to The Information reported today, a person directly involved in the deal talks revealed.
Ben Ullmann, a spokesperson for Stability AI, said in a statement: "Stability AI is working exclusively with a consortium of world-renowned technology investors. Stability AI leadership is working closely with the consortium to make a significant equity investment in the company. He called the consortium "completely aligned" with the company's leadership.
The funding could be a "life-saving money" for Stability AI and is expected to allow Stability AI to forge closer ties with media and entertainment companies that may buy its AI image and video generation software, allowing it to go head-to-head with larger competitors.
According to foreign media reports, OpenAI has also been competing for Hollywood studios, media executives and talent agencies in recent weeks. This week, Google launched Veo, a large AI video model that can generate videos of more than 60 seconds. Some animation, film and television practitioners have begun to explore the use of AI tools to participate in creation.
3. The generative AI industry has set off a wave of acquisitions, with Snowflake negotiating a $1 billion acquisition of Reka AI
As generative AI competition intensifies, some AI startups have been forced to exit or be swallowed up by tech giants.
In March, AI chatbot startup Inflection AI struck a deal with Microsoft, which hired the majority of Inflection AI's employees and paid the startup $650 million as part of a deal to license its models, in an alternative way to "buy" Inflection AI in a disguised way.
According to a report by Bloomberg today, people familiar with the matter said that Snowflake is in talks to buy startup Reka AI for more than $1 billion, although these talks may not reach an agreement.
Founded in 2022 by researchers from Google and Meta, Reka AI focuses on the development of large language models. It is valued at about $300 million in a 2023 investment round, including the Snowflake venture capital arm, according to Reuters.
Snowflake, which develops tools for data organization and analysis in cloud computing and sees generative AI as an accelerator for its business, acquired AI search engine Neeva last year.
In April of this year, Snowflake released Arctic, a large language model, and enables its customers to use AI models from third parties such as Reka on data in Snowflake.
Not only Snowflake, but also its rival Databricks has released its own open-source large language model, and last year acquired AI startup MosaicML for $1.3 billion.
Databricks was valued at $43 billion in September last year, while Snowflake is currently worth more than $55 billion.
Conclusion: Under the AI boom, startups are facing the test of funding
Replit lays off 20% of its workforce, Stability AI sells or finances to save itself, Reka AI sells itself...... These microcosms of the generative AI boom highlight the pressures AI companies face in developing AI models, products, and exploring commercial monetization.
Mistral, an open-source AI developer in Paris, France, has only recently turned a profit and is in talks with investors about a new round of funding that could reach a valuation of $5 billion.
In the long run, Mistral hopes to make money from its larger business model and yet-to-be-released application APIs. But an employee of a Mistral partner that resells Mistral's software said the vast majority of the partner's customers are using the smaller free model, and less than 10% of customers are using the largest paid model.
Today, the best business model for generative AI applications is murky. Even the world's top AI models have so far failed to address the high cost of running.
Due to the widespread penetration of AI products in the "volume" market, many AI companies offer their products for free to attract more users, and even if these companies offer attractive additional features, it is difficult for most users to be willing to pay for the products.
Generative AI startups are eagerly seeking more funding transfusions, and tech giants are scrambling to partner with generative AI startups or initiate acquisitions.
AI startups that can't find a source of revenue may put their own development path on the brink of danger.
来源：The Information，Semafor，彭博社