news

Big companies are blocking AI startups and exposing a "death list", and almost no former Google boss is spared

2024-08-27

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

New Intelligence Report

Editor: Editorial Department

【New Wisdom Introduction】AI startups founded by former Google employees have all been acquired by big tech companies! What is this evil wind blowing in Silicon Valley? Foreign media summarized the reasons why it is difficult for Google AI researchers to start their own businesses, and even gave a "heartbreaking" prediction list, naming the companies that are about to be acquired by big companies.

The much-anticipated AI startup has lost its glory in just four months?

Just now it was reported that the French AI startup H (formerly known as Holistic), which had just received $220 million in seed round financing, was revealed to have left the company!

Three of H's five co-founders have left, reportedly due to "operational and business differences."

These former Google employees be like: Why didn’t anyone tell me that starting a business was so difficult?

Former Google employee: Life is tough out there

H's dramatic change came at a time when a bunch of former Google employees were starting companies and things were changing.

A careful inventory shows that every one of these companies, Inflection, Adept, and Character, were acquired within less than three years of their founding.

Another similar AI development company, Reka, was also looking good and had already started acquisition talks with Snowflake this year. The company’s valuation was expected to reach $1 billion, but no one expected the result — the negotiations broke down in May.

In March this year, Reka founder Yi Tay wrote a long article, reviewing the difficult process of starting a business in the past year and sharing his 24-hour schedule.

There is also Ideogram, an AI image tool that was highly praised by netizens after they tried it out a while ago. It is also a company founded by four former Google employees.

Although it has achieved quite impressive results, the company is said to be discussing acquisition matters with at least one technology startup, but the negotiations have not made any progress yet.

The CEO of Cohere is one of the eight Transformer developers who wrote the groundbreaking paper, along with Noam Shazeer, the former CEO of Character AI.

Cohere also faces a lot of questions, such as the company has a lot of funds on its books, but its employees have very low incomes.

Foreign media The Information simply asked such a soul-searching question:

Did someone put a curse on the startups of former Google employees?

However, upon closer examination, it can be found that the motivations behind these entrepreneurial failures are actually quite typical.

Why is it difficult to start a business?

All along, Google departments have been more like "academic research institutions" that do not urgently need to generate cash flow, and focusing on research is the core.

The most important thing for researchers working at Google is to focus on doing good research. The rest of the commercialization strategies are the responsibility of the decision makers.

Once you start your own business, the skills required in the two environments are completely different.

For researchers in academia, being good at research does not mean being good at managing/operating a company.

There are two exceptions: Sergey Brin and Larry Page, the two founders of Google, dropped out of their studies to start the company while pursuing a master's degree in CS at Stanford.

The founders of the AI ​​star company Databricks are seven scholars who developed the company's core technology while working as researchers at UC Berkeley.

And they are just examples in the vast world.

After starting the company, I bridged the skills gap by hiring people who understood the business of technology.

Secondly, personnel is also a big problem.

Even the best startups often face internal turmoil, and it’s not uncommon for founders to disagree.

In addition, founders with academic backgrounds may not be adapted to the company's management structure, increasing the possibility of conflict.

In addition, the high cost of AI development and fierce competition from existing companies make it difficult for these startups from the outset.

Moreover, even the best AI startups will find it difficult to survive independently without the financial support and cloud services provided by large companies such as Google, Amazon, and Microsoft.

Investors are more cautious

So, for investors, they must bet on companies with potential GenAI technology.

Many appear to have failed to conduct adequate due diligence and investment decisions may have been made too hastily and without caution.

Last year, venture capitalists seemed ready to throw millions of dollars at any researcher coming out of Google, Meta, or other well-known AI labs.

Even their demo was as vague and clichéd as Company H. One of the PPT contents is shown below.

However, investors are becoming more skeptical about their expectations for returns from AI and are becoming more cautious.

Of course, it’s not all bad news for founders.

Character AI, founded by founder Noam Shazeer, could become a billionaire through its acquisition by Google.

His investors, including Andreessen Horowitz, received at least a 2.5-fold return on their investment through the deal, although this is considered a relatively low return in the venture capital industry.

Many AI startups founded by Google employees have been swallowed up

The following list is a collection of five companies founded/operated by former Google employees in the past few years.

They involve different fields such as chatbots, AI security applications, warehouse robots, and email assistants.

These companies have not announced financing since the third quarter of 2022, which may mean that they may be acquired by others.

These people who left to start their own businesses once held different roles in Google departments.

For example, Yaron Singer of Robust did two years of postdoctoral research in the field of AI.

AI Startups vs. Tech Giants

Just recently, we reported the story of "The CEO of a 2.5 billion unicorn took the lead in running away and packed up 30 employees and sold them to Google overnight."

This is the third sale of an AI startup to a major tech giant in just five months.

Naom Shazeer ran away with the president

People are asking the same question: Who will be the next startup?

Yes, after raising billions of dollars in funding, when it comes to developing their own AI models, these startups find that competing with these large technology companies with huge computing resources and large customer lists is like an ant trying to shake a tree.

However, these startups, which have top-notch technical talent, have become popular in the eyes of many large technology companies that are waiting and watching.

One of them is Apple.

According to insiders, Apple executives have discussed acquisitions with at least three AI startups.

Now, a visible siege has been launched within Silicon Valley - on one side are startups struggling to survive, and on the other side are the salivating technology giants.

There are Google and Character AI, Microsoft and Inflection AI, Amazon and Adept AI, and Salesforce, Oracle, Nvidia, or new players like xAI and OpenAI are also gearing up and may take action within a few months.

The Information also listed seven startups that investors and executives believe are very likely to become deal targets.

They all hired researchers who are familiar with training AI models, which is exactly the capability that large technology companies most desire to develop in-house.

Some well-known AI startups were not included in this list for the following reasons.

Mistral AI has a deep bench of research talent, but it is a pillar of France’s national AI strategy, so it is unlikely to be involved in complex deals with American tech companies.

You.com is an AI assistant similar to ChatGPT that primarily uses models built by other companies, but may still be attractive to some buyers.

Salesforce, for example, acquired the previous startup of the founder of You.com; its CEO Marc Benioff is an investor in You.com.

Reka

Founded: 2022

Funding raised: $57.5 million

Valuation: $300 million

Major investors: Snowflake, Radical Ventures, Nat Friedman

Reka, founded by former Google researchers, raised money amid a wave of investor excitement surrounding ChatGPT.

Today, Reka has built models that can process text, audio, images, and video.

Reka has already struck deals with Snowflake, one of its main backers, as well as Oracle, to offer products on their platforms, though the models have yet to make their way out on more widely used services like Amazon or Google.

Bloomberg reported that Reka had negotiated an acquisition with Snowflake, but the talks fell through.

Therefore, there may be other potential buyers who could use Reka's research team to improve their own AI products.

Imbue

Founded: 2021

Capital raised: $232 million

Valuation: Over $1 billion

Major investors: Lightspeed Venture Partners, Nvidia, Astera Institute, Eric Schmidt, Amazon Alexa Fund, Dell, Social Capital

Imbue, formerly known as Generally Intelligent, is one of the best-funded developers of AI models but has yet to start selling its own models.

These models focus on reasoning and writing code, and the company's goal is to build AI agents that can perform complex tasks for people.

Imbue last raised funding a year ago and has more than 30 employees, according to its LinkedIn information.

All of these factors mean it could soon become a buyers’ market.

Imbue spokesman Tom Freeman said the company has been working on this project for years and no competitor has yet successfully built a usable agent.

Cohere

Founded: 2019

Capital raised: $980 million

Valuation: $5.5 billion

Major investors: Oracle, Nvidia, Salesforce, Tiger Global Management, Index Ventures, Radical Ventures

Cohere just completed a large round of financing earlier this summer, easing the pressure on it to seek a sale in the near term.

The Toronto-based company is working to make the Command R family of models highly accurate in commercial applications such as search and summarization, and in multiple languages.

That is likely to attract enterprise-focused companies like Salesforce and Oracle, which are also among investors.

Ideogram

Founded: 2022

Capital raised: $96.5 million

Valuation: $580 million

Major investors: Andreessen Horowitz, Index Ventures, SV Angel

Ideogram launched last August and has become a hot new model startup known for its high-quality image generation.

However, in the field of AI raw images, competition is very fierce. Companies including Google, Adobe, Midjourney, etc. have launched similar products and open-sourced many free models.

The company last raised $80 million in February.

AI21 Labs

Founded: 2017

Capital raised: $336 million

Valuation: $1.4 billion

Major investors: Coatue Management, Nvidia, Google, Samsung, Intel, Pitango

Founded in 2017 and headquartered in Israel, AI21 Labs is a company that develops text generation models for enterprises.

At present, it has built a real business with tens of millions of revenue. The team behind it is very attractive to Oracle and Salesforce, both of which launched their own products in the past year.

As previously reported, AI21 completed its latest round of financing in August last year, with investors paying more than 20 times its expected revenue for that year.

Ori Goshen, co-founder and co-CEO of AI21, said, "We have never considered selling. Our goal is to create some substantial technology."

Essential AI

Founded: 2023

Funding raised: $65 million

Valuation: Currently unknown

Major investors: March Capital, Nvidia, Google, Thrive Capital

Essential AI was co-founded by Transformer duo Niki Parmar and Ashish Vaswani.

They are developing an AI model for enterprises, which they call "Enterprise Brain." Before the establishment of this company, the two also founded the startup Adept.

Essential AI currently has about 20 employees and raised $60 million in its Series A funding round in December last year.