Interview with Gao Tianyao about AI Commercialization
COVID-19 and China
Gao Tianyao (高天垚), partner at Legend Star

EqualOcean interviewed with Gao Tianyao (高天垚), partner at Legend Star(联想之星) on July 1, 2019. 

Legend Holdings subsidiary Legend Star focuses on early-stage investments in artificial intelligence, TMT and healthcare area and in-depth incubation offering professional training to entrepreneurs. 

Mr. Gao joined the firm in 2016. With 10-year experience in technology investment, he once worked in a Wall Street boutique investment bank, JD Capital, and strategic investment department of listed companies, covering early to late stages. Mr. Gao holds a bachelor's degree from Beijing University of Technology and a master's degree in materials engineering from University of Florida and a master's degree in operations research from Columbia University.

Here are the key takeaways from the interview: 

Early technology investment: efficiency and public awareness

Gao Tianyao: We focus on early-stage investments with the idea of 'technologies transforming traditional industries'. From a financial investment perspective, a project with explosive power and fast growth potential is the target of early-stage investors who focuses on angel and series A round of financing. TMT and internet companies generally have the characteristic.

Based on our experience, some technology-oriented projects, such as early new material projects, are not suitable for early venture investments because it might take a very long time to make a breakthrough. 

Therefore, there are 'musts' for early technology investment – the technology must greatly improve efficiency and public must be well aware of it, which makes AI a good fit.

We invested in Megvii (旷视科技) and AISpeech (思必驰) as early as 2011 and 2012 respectively. In 2015, we began to invest in AI systematically. In recent years, AI commercialization is speeding up and starts to generate returns. In the beginning, all of the companies devoted to underlying technologies. Growing larger, they started to find directions to put into practice. As of now, we still invest in some underlying tech startups with self-developed, innovative, cutting-edge technology.

AI commercialization: seeking a big market

Gao Tianyao: Considering landing AI, we think more of 'industry + AI' rather than 'AI + industry', targeting those large markets.  

Taking security area as an example, the most valued and the fastest-growing Chinese AI companies, such as SenseTime (商汤科技), Megvii (旷视科技) and Yitu (依图科技) are all in this area. But we also need to consider huge incumbents Hikvision (海康威视) and Dahua (大华), whose growth rate are still as high as almost 50%. There are still an increasing number of scenarios to use surveillance cameras, indicating the industry is booming.

Besides, existing stock in this industry is huge. All the existing surveillance cameras are to be replaced by AI cameras in the future. As the penetration rate of AI increases, there is an explosion point. Therefore, security industry is the one with large new and existing markets.

Smart lock market is another large one. Smart lock suppliers target both "2B"(to-business) and "2C" (to-consumer) markets. For apartment rental service providers, smart lock, which could be easily reset each time, is definitely more convenient than traditional locks. At the same time, smart locks are sold to consumers directly. Households that upgrade locks to the intelligent ones are not likely to change back to traditional ones anymore.  

It seems the more the incumbents use AI, the more opportunities are there for AI startups in an industry. For instance, some AI startups target medical services and meanwhile traditional medical service providers are also exploring AI applications, despite slow development. In the automobile industry, Bosch (博世) has not yet developed a full-stack AI solution, and small startups are working on auto-related AI products such as Lidars.

Besides, in the process of commercialization, technology companies are transitioning from pure technology providers to solution providers. They used to sell technologies to downstream OEMs when entering the market, and now more and more AI companies start to expand their business along the value chain. The moves help startups to improve bargaining power, scale up and increase revenue.

Data is also important. Face recognition works because it is relatively viable to capture facial data comprehensively. Facial data are certain. But things are different for autonomous driving because the environment is so unpredictable and a minor failure could cause huge trouble.

Another aspect is data integration. To generate better analytical results, it must eliminate the barriers between different types of data about one subject or related subjects. Sensitive data are especially hard to aggregate, a situation limiting the analytical power of using machine learning. This is an obstacle in AI development in finance and healthcare.

It's a lengthy process for AI to gather feedback and iterate, and it also takes time for other technology development to complement to AI for further applications. 

The China-US difference in landing AI 

Gao Tianyao: We invested in two similar company in China and the United States. Beijing-based Reinlight (驭光科技), focusing on the design, manufacture and applications of advanced optics, grows fast in China. After landing deals with Huawei, Reinlight successfully found its position and built the supply chain in the mobile industry. But its US counterpart didn't survive, because US OEMs such as Apple have built strong connections with their suppliers for years and startups couldn't find their way to break into the market.

There are also projects suit for US but not China. For example, the US startup AMP Robotics who develops robotic systems to sort recyclable material that reducing the cost for recycling facilities. We don't see this track yet because it's still a labor-intensive industry in China. In Agriculture, California based Abundant Robotics creates apple-picking robots. However, in China, the cost of robotic apple picker is high than labor cost. 

AI commercialization situations by industries

Gao Tianyao: Consumer internet players could achieve revenues beyond CNY 1 billion. Internet modes make it easy to scale in a way that traditional businesses could not. Security and smart home companies could generate revenues of over CNY 500 million. All the others might not be able to achieve revenues beyond CNY 100 million.

Healthcare and financial services are the next biggest markets, though AI companies are required to build some deep connection within these industries. Education, just like healthcare, is also a promising market but it takes time.

Auto companies are targeting a huge existing market. Although autonomous driving technology is not ready yet, companies get revenue from ADAS and intelligent cockpit solutions. 

Latest Updates:

See Also

Communicate Directly with the Author!

Ask the author questions about the copied text

Research Reports
Editor's Picks