Featured Article

Valued at $1B, Kai-Fu Lee’s LLM startup unveils open source model

The venture capitalist and computer scientist wants to create an OpenAI equivalent for China

Comment

Image Credits: TechCrunch

Kai-Fu Lee, the computer scientist known in the West for his bestseller “AI Superpowers” and in China for his bets on artificial intelligence unicorns, has a new venture — and a great ambition.

In late March, Lee launched a company called 01.AI with the vision to develop a homegrown large language model for the Chinese market. The venture puts him in competition with other prominent Chinese tech leaders, including Sogou’s founder Wang Xiaochuan, who have been swiftly gathering talent and venture capital to establish China’s equivalents of OpenAI.

“I think necessity is the mother of innovation, and there’s clearly a huge necessity in China,” Lee, who’s 61 and leading 01.AI as CEO, told TechCrunch in an interview, explaining the motive behind starting the company. “Unlike the rest of the world, China doesn’t have access to OpenAI and Google because those two companies did not make their products available in China, so I think many doing LLM are trying to do their part in creating a solution for a market that really needs this.”

01.AI’s growth is a fitting reflection of the rapid development in the generative AI field. Seven months after its founding, the startup has released its first model, the open source Yi-34B. The decision to introduce an open LLM as its debut product is a way to “give back” to society, said Lee. For people who have felt LLaMA is a “godsend” to them, “we’ve provided a compelling alternative,” he added.

As of writing, Yi-34B, which is a bilingual (English and Chinese) base model trained with 34 billion parameters and significantly smaller than other open models like Falcon-180B and Meta LlaMa2-70B, came in first amongst pre-trained LLM models, according to a ranking by Hugging Face.

“We still believe that larger models, when trained well, on a large amount of high-quality data, will always outperform substantially smaller models of comparable quality and comparable technology, so I think [Yi-34B] outperforming much larger models is something that we don’t usually see,” said Lee. “We feel quite confident as we released models that are 100 billion to 400 billion over the next coming year, year and a half, these models will be dramatically better than today’s model that we announced.”

The startup’s ability to commence model training quickly is no doubt an outcome of its smooth fundraising, which is critical to securing top-tier talent and AI processors. While declining to disclose how much 01.AI has raised, Lee said it’s valued at $1 billion after receiving financing from Sinovation Ventures, Alibaba Cloud and other undisclosed investors.

01.AI has already grown to more than 100 employees, over half of whom are LLM experts from major multinational and Chinese tech firms. Its vice president of technology, for instance, is an early member of Google’s Bard, and its chief architect was a founding member of TensorFlow and worked alongside renowned researchers like Jeff Dean and Samy Bengio at Google Brain. The key figures behind Yi-34B are Wenhao Huang, a Microsoft Research Asia veteran, and Ethan Dai, who held senior AI positions at Huawei and Alibaba.

Having backed over 10 unicorns and venture-built seven companies through Sinovation Ventures, Lee is possibly one of the most well-connected investors and entrepreneurs in China.

“It’s been, you know, over 25 years since the founding of Microsoft Research Asia, and everything I’ve done has been about getting super great talent,” said Lee, who launched Microsoft Research Asia, the U.S. giant’s biggest research center abroad, before spearheading Google China. Over the years, Microsoft Research Asia has earned the reputation as the “West Point” for nurturing China’s AI entrepreneurs.

“Now, of course, you want to pay people fairly, and you need to be competitive in pay, but I really think that it’s also about people believing they can make a difference and believing the company can succeed,” said Lee, appearing in a video call at 9:30 p.m. Beijing time. His staff were just as self-motivating. One of the startup’s infrastructure experts was working well into the wee hours that day, still messaging Lee at 2:15 a.m. to express his excitement about being part of 01.AI’s mission.

It’s no secret that building LLMs is a costly undertaking. To sustain its cash-intensive operations, 01.AI has plans for monetization right from the start. While the company will continue to open source some of its models, its objective is to build a state-of-the-art proprietary model that serves as a foundation for a diverse range of commercial products.

“We can’t open source everything,” said Lee. “We were quite cognizant of the fact that these large language models require a lot of compute, and therefore, are very expensive. When we raise a lot of money, most of it will be spent on the GPU. Given that, we needed to first acquire as much GPU as we could, which we did.”

Like other LLM players in China, 01.AI has proactively stockpiled GPUs in anticipation of U.S. sanctions; it borrowed money to buy processors even before it landed funding. Over the past year, the Biden administration has heightened restrictions on China’s access to high-end AI chips, prompting Chinese firms to pay inflated prices for chips. The foresight was rewarded — 01.AI now has a supply that will suffice for at least the next 12-18 months.

Aside from causing headaches for Chinese firms, U.S. sanctions have been a catalyst for innovation by encouraging them to optimize the use of computing power. “With a very high-quality infrastructure team, for every 1,000 GPUs, we might be able to squeeze 2,000 GPUs workload out of them,” said Lee.

01.AI’s path to monetization hinges largely on its ability to find product-market fit for its expensive AI models. While top-notch LLM scientists are scarce, there’s no shortage of product talent in China.

“China’s not ahead of the U.S. in LLM, but there’s no doubt China can build better applications than American developers mostly because of the phenomenal mobile internet ecosystem that was built over the last 12 years or so,” argued Lee.

While the founder gave no details on the services in the pipeline, he hinted that the company is experimenting with concepts in the productivity and social directions, and he’d be “disappointed” if 01.AI didn’t release an app within this calendar year.

The startup’s ultimate goal, according to Lee, is to become an ecosystem where outside developers can build applications easily. “The duty is not just to push out good research models, but even more importantly to make application development easy so that there can be compelling applications,” he said. “At the end of the day. It is an ecosystem play.” Time will tell if Lee’s AI endeavor will pay off.

China’s search engine pioneer unveils open source large language model to rival OpenAI

Meituan buys founder’s months-old ‘OpenAI for China’ for $234M

More TechCrunch

Intuitive Machines made history when it became the first private company to land a spacecraft on the moon, so it makes sense to adapt that tech for Mars.

Intuitive Machines wants to help NASA return samples from Mars

As Google revamps itself for the AI era, offering AI overviews within its search results, the company is introducing a new way to filter for just text-based links. With the…

Google adds ‘Web’ search filter for showing old-school text links as AI rolls out

Ilya Sutskever, OpenAI’s longtime chief scientist and one of its co-founders, has left the company. OpenAI CEO Sam Altman announced the news in a post on X Tuesday evening. pic.twitter.com/qyPMIcvcsY…

Ilya Sutskever, OpenAI co-founder and longtime chief scientist, departs

Blue Origin’s New Shepard rocket will take a crew to suborbital space for the first time in nearly two years later this month, the company announced on Tuesday.  The NS-25…

Blue Origin to resume crewed New Shepard launches on May 19

This will enable developers to use the on-device model to power their own AI features.

Google is building its Gemini Nano AI model into Chrome on the desktop

It ran 110 minutes, but Google managed to reference AI a whopping 121 times during Google I/O 2024 (by its own count). CEO Sundar Pichai referenced the figure to wrap…

Google mentioned ‘AI’ 120+ times during its I/O keynote

Firebase Genkit is an open source framework that enables developers to quickly build AI into new and existing applications.

Google launches Firebase Genkit, a new open source framework for building AI-powered apps

In the coming months, Google says it will open up the Gemini Nano model to more developers.

Patreon and Grammarly are already experimenting with Gemini Nano, says Google

As part of the update, Reddit also launched a dedicated AMA tab within the web post composer.

Reddit introduces new tools for ‘Ask Me Anything,’ its Q&A feature

Here are quick hits of the biggest news from the keynote as they are announced.

Google I/O 2024: Here’s everything Google just announced

LearnLM is already powering features across Google products, including in YouTube, Google’s Gemini apps, Google Search and Google Classroom.

LearnLM is Google’s new family of AI models for education

The official launch comes almost a year after YouTube began experimenting with AI-generated quizzes on its mobile app. 

Google is bringing AI-generated quizzes to academic videos on YouTube

Around 550 employees across autonomous vehicle company Motional have been laid off, according to information taken from WARN notice filings and sources at the company.  Earlier this week, TechCrunch reported…

Motional cut about 550 employees, around 40%, in recent restructuring, sources say

The keynote kicks off at 10 a.m. PT on Tuesday and will offer glimpses into the latest versions of Android, Wear OS and Android TV.

Google I/O 2024: Watch all of the AI, Android reveals

Google Play has a new discovery feature for apps, new ways to acquire users, updates to Play Points, and other enhancements to developer-facing tools.

Google Play preps a new full-screen app discovery feature and adds more developer tools

Soon, Android users will be able to drag and drop AI-generated images directly into their Gmail, Google Messages and other apps.

Gemini on Android becomes more capable and works with Gmail, Messages, YouTube and more

Veo can capture different visual and cinematic styles, including shots of landscapes and timelapses, and make edits and adjustments to already-generated footage.

Google Veo, a serious swing at AI-generated video, debuts at Google I/O 2024

In addition to the body of the emails themselves, the feature will also be able to analyze attachments, like PDFs.

Gemini comes to Gmail to summarize, draft emails, and more

The summaries are created based on Gemini’s analysis of insights from Google Maps’ community of more than 300 million contributors.

Google is bringing Gemini capabilities to Google Maps Platform

Google says that over 100,000 developers already tried the service.

Project IDX, Google’s next-gen IDE, is now in open beta

The system effectively listens for “conversation patterns commonly associated with scams” in-real time. 

Google will use Gemini to detect scams during calls

The standard Gemma models were only available in 2 billion and 7 billion parameter versions, making this quite a step up.

Google announces Gemma 2, a 27B-parameter version of its open model, launching in June

This is a great example of a company using generative AI to open its software to more users.

Google TalkBack will use Gemini to describe images for blind people

Google’s Circle to Search feature will now be able to solve more complex problems across psychics and math word problems. 

Circle to Search is now a better homework helper

People can now search using a video they upload combined with a text query to get an AI overview of the answers they need.

Google experiments with using video to search, thanks to Gemini AI

A search results page based on generative AI as its ranking mechanism will have wide-reaching consequences for online publishers.

Google will soon start using GenAI to organize some search results pages

Google has built a custom Gemini model for search to combine real-time information, Google’s ranking, long context and multimodal features.

Google is adding more AI to its search results

At its Google I/O developer conference, Google on Tuesday announced the next generation of its Tensor Processing Units (TPU) AI chips.

Google’s next-gen TPUs promise a 4.7x performance boost

Google is upgrading Gemini, its AI-powered chatbot, with features aimed at making the experience more ambient and contextually useful.

Google’s Gemini updates: How Project Astra is powering some of I/O’s big reveals

Veo can generate few-seconds-long 1080p video clips given a text prompt.

Google’s image-generating AI gets an upgrade