Zhi Dongxi reported on August 15, The Information reported that on August 3, Google signed a strategic agreement with the large model startup Character.AI, reaching a disguised acquisition deal of "model licensing agreement + core team poaching." This is the third transaction between technology giants and AI startups in the past six months. This move has inspired a thought: Who will be the next AI startup to be acquired?

Seven startups that have emerged in the generative AI field have raised billions of dollars in funds to develop cutting-edge AI models. Although these startups have made significant progress in financing, they still face many difficulties when competing with large technology companies with strong computing resources and a large customer base.

Advertisement

However, startups have a large number of top technical talents, which makes them potential acquisition targets for companies like Apple that have not yet entered this field. Relevant investors revealed that Apple executives have contacted at least three AI startups and discussed the possibility of acquisition transactions.

01. Technology giants successively acquire startups for AI layout

Before Google reached a deal with the large model startup Character.AI, technology giants have started to accelerate the layout of the generative AI field. Microsoft and Amazon have reached similar acquisitions with Inflection AI and Adept AI respectively. Investors and industry executives believe that other large technology companies, such as cloud data platform Salesforce, cloud database platform Oracle, chip manufacturing company Nvidia, and even emerging AI companies such as xAI or OpenAI, may also take similar actions in the next few months.

Behind the rush of technology giants to acquire AI startups, it reflects the increasing market demand for advanced AI technology. Companies with strong computing resources and a large customer base are supplementing their AI technical strength through acquisitions. For example, Snowflake has been discussing cooperation with the natural language processing model company Cohere recently, and this cooperation may further deepen and eventually reach a comprehensive acquisition.

02. 7 startups expected to be acquired: involving multimodal AI such as text, images, videos, etc.

After communicating with investors and corporate executives, The Information has compiled a list of 7 startups that are expected to become targets for similar transactions. Each company on this list has researchers who are training AI models, which is the internal development capability that large technology companies urgently need.

1. Reka: Focusing on the development of multimodal AI models, raising $57.5 million in fundsReka is a multimodal AI model development company founded in 2022 by a team of former Google researchers. Leveraging the investment boom following the launch of ChatGPT, the company successfully raised $57.5 million in funding. Reka is now valued at approximately $300 million, with major investors including Snowflake, Radical Ventures, and Nat Friedman.

Currently, Reka has developed multimodal AI models capable of processing text, audio, images, and videos. Although these models have not yet been applied on mainstream platforms such as Amazon or Google, Reka has reached agreements with its main supporters, Snowflake and Oracle, to provide services on these companies' platforms.

According to Bloomberg, Reka had previously engaged in acquisition negotiations with cloud data giant Snowflake, but ultimately failed to reach an agreement. Therefore, other potential buyers still have the opportunity to utilize Reka's research team to enhance their own AI products.

2. Imbue: Focused on AI Inference and Coding, Raised $232 Million in Funding

Imbue is an AI inference and coding company established in 2021 with the support of major investors such as Lightspeed Venture Partners and Nvidia. The company has successfully raised $232 million in funding and is now valued at over $1 billion.

The Information mentioned that Imbue, formerly known as General Intelligent, is one of the most well-funded AI model development companies, but has not yet started promoting its models. The company is seeking to build intelligent agents capable of performing complex tasks for humans.

Imbue spokesperson Tom Freeman stated that the company has always regarded its research work as a long-term project, and currently, no competitors have successfully built usable intelligent agents.

Imbue's last round of financing was a year ago, and the company currently has more than 30 employees. These factors suggest that it may be about to enter the buying and selling market.3. Cohere: Committed to Natural Language Processing Models, Raised $980 Million in Funding

Cohere is a natural language processing company founded in 2019 by several researchers and engineers who previously worked at Google Brain, a deep learning AI research team under Google AI. The company has successfully raised $980 million in funding and is now valued at $5.5 billion, with its main investors including Oracle, Nvidia, Salesforce, and Tiger Global Management.

This summer, Cohere, headquartered in Toronto, just completed a large round of financing, which has alleviated the pressure to seek acquisitions in the short term. Cohere is committed to building a series of models called Command R, which can be used for commercial applications such as search and summarization, and have high accuracy in multiple languages.

The Information believes that Cohere may attract companies focused on serving enterprises, such as cloud platform Salesforce and database platform Oracle, and the former has long been an investor in Cohere.

4. Pika Labs: Focusing on Developing Video Generation AI, Raised $141 Million in Funding

Pika Labs is an AI video generation startup founded in 2023 by Chinese entrepreneur Guo Wenjing with the support of major investors such as Spark Capital and Lightspeed Venture Partners. The company has successfully raised $141 million in funding and is now valued at $470 million. Its main investors include Spark Capital, Lightspeed Venture Partners, and Sequoia Capital.

Pika Labs AI is committed to building AI that can generate videos, which is a technology that its larger competitors have not yet made available to the public. The company allows each user to generate at least 25 three-second videos for free, while paid subscription users can use editing tools and generate more clips.

A deal with a larger technology company could provide Pika with more high-quality data to train its AI models. Apple could use Pika's technology for new iPhone features or even create new shows through its ownership of the copyrights to a large number of professional videos on Apple TV. Alternatively, OpenAI could enhance its video generation capabilities by leveraging Pika's technical team, which could make it more powerful than its competitor Google.5. Ideogram: Focused on Developing Image Generation AI, Raised $96.5 Million in Funding

Ideogram, a company known for its high-quality image generation models, was founded in 2022 with the support of major investors such as Andreessen Horowitz and Index Ventures. The company has successfully raised $96.5 million in funding, now valuing at $580 million. However, the field of image generation quickly became crowded, with companies like Google, Adobe, and Midjourney launching their own AI models, and many free open-source models available on the market.

The Information believes that, considering the similarity between image and video generation AI, Ideogram, like Pika, is of great value to both OpenAI and Apple.

6. AI21 Labs: Committed to Developing Text Generation AI, Raised $336 Million in Funding

AI21 Labs is a text generation AI company founded in 2017 by artificial intelligence pioneer and technology expert Amnon Shashua. The company has successfully raised $336 million in funding, now valuing at $1.4 billion. Its main investors include companies such as Coatue Management, NVIDIA, Google, Samsung, and Intel.

This Israeli-based startup has developed text generation models for businesses. AI21 Labs has been in existence longer than most independent AI model development companies and has established an actual business, generating tens of millions of dollars in revenue. The technical team behind it would be a valuable asset to companies like Oracle or Salesforce, although the latter has also launched its own AI models in the past year.

AI21's most recent funding round was in August 2023, when investors put in more than 20 times the company's projected annual revenue.

7. Essential AI: Specializing in Generative AI Models, Raised $65 Million in FundingEssential AI is a generative AI model development startup founded in 2023 by two former Google researchers, Niki Parmar and Ashish Vaswani. Both founders are among the inventors of the Transformer model, which is the foundation of current generative AI applications like ChatGPT. Its main investors include companies such as March Capital, Nvidia, and Google. Essential AI has about 20 employees, and such a top-tier technical talent startup team is highly attractive to large technology companies.

Nevertheless, this startup may be relatively inexpensive. Because in its Series A funding round in December last year, it only raised $65 million in capital and currently does not have a clear valuation.

03. Conclusion: 7 AI companies may be acquired, and the market will see more transactions

When competing with resource-rich tech giants, AI startups are facing increasingly severe challenges. Although these 7 startups have collectively raised billions of dollars for the development of AI models, they still struggle to compete with large companies in terms of funding, computing resources, and data acquisition. At the same time, tech giants have already established strong brands and user bases in the market, which also invisibly increases the difficulty for startups to gain market share to support their business development. However, the technical talent and innovation capabilities of these companies also make them ideal acquisition targets for large companies. In the coming months, the market may see more similar transactions.