China’s SenseTime amasses firepower for global AI push

This fundraiser follows hot on the heels of a $600 million capital round in April and values the AI unicorn at $4.5 billion: plenty of dry powder to pursue China’s global leadership ambitions in the field.

Chinese artificial intelligence unicorn SenseTime has raised $620 million to help finance its buildout across facial recognition, autonomous driving and medical imaging.

The series C+ funding round was led by Fidelity International, Hopu Capital, Silver Lake and Tiger Global. Qualcomm Ventures and other investment firms and strategic investors also participated.

This latest fundraiser gives it a valuation of over $4.5 billion, maintaining its position as the world’s largest and most valuable AI unicorn. Alibaba led the series C round of funding in April which collected $600 million.

SenseTime is not alone in raising capital to build out AI capabilities; Baidu, Tencent, and Alibaba are all expanding their AI offerings, recruiting talent and investing in AI startups globally,

Chinese AI startups accounted for 48% of AI funding globally last year, surpassing the United States for the first time, according to data from CB Insights.

Having achieved profitability in 2017, SenseTime has rapidly deployed its computer vision and deep learning technologies in smart cities, smartphones, internet entertainment, automobiles, finance, retail, and other industries.

Such industrial applications dovetail with China’s initiative to upgrade its manufacturing capability by embracing innovative technologies. The US has said the policy, dubbed “Made in China 2025”,  distorts the market in favour of supported companies. Next generation information technologies is one of ten targeted sectors.

SenseTime is competing head-to-head with the likes of US Big Tech companies Google and Facebook in the field of deep-learning.

Deep pockets will certainly help in the battle for supremacy. Salaries are rising as AI companies compete in a shallow pool for talent. SenseTime is also snapping up talent from the financial sector, luring Bocom International ECM head Esther Wong in March.

Elsewhere, Baidu USA offered a base salary of $130,000 to $175,000 per annum for a machine learning engineer last year, according to a petition filed with the US Labor Department for non-immigrant workers, reported by CBInsights. Google offered around $110,000 for an ML engineer.

SenseTime is looking into opening offices in the US, a spokeswoman told FinanceAsia. It already has offices in Hong Kong, Beijing, Shenzhen, Shanghai, Chengdu, Hangzhou, Kyoto, Tokyo and Singapore. 

For the last three consecutive years, SenseTime has upheld a year-on-year growth rate of 400%. This year up until May, its business contract revenue has increased by more than 10-fold so far.

SenseTime has more than 700 partners and customers including MIT, CUHK, Qualcomm, NVIDIA, Honda, Alibaba, Suning, China Mobile, UnionPay, Wanda, HNA, Huawei, Xiaomi, OPPO, vivo, Weibo and iFLYTEK.

Recently, the company launched the SenseDrive driver monitoring system. It also launched SenseAR, China’s first locally developed augmented reality platform.

The public and private sectors in China are working together with the government to make it a global leader in artificial intelligence in the next decade.

Earlier this year, SenseTime signed an agreement with Shanghai Shentong Metro Group to deploy AI solutions for metro traffic monitoring. SenseTime also signed an agreement with the city of Chengdu.

Last year, FinanceAsia named SenseTime as one of its  top 10 disruptive start-ups in the region. In November, co-founder Xu Bing told FinanceAsia of his vision of establishing SenseTime as the equivalent in artifiical intelligence to the State Grid

Since its inception in 2014, SenseTime has raised capital from Temasek, Qualcomm, IDG Captial and CDH Investments. The company counts China Mobile, Wanda, HNA Group and Huawei among its customers and partners.

SenseTime’s high-performance deep learning supercomputing platform is one of China’s largest, boasting more than 8,000 GPUs (graphic processing units). It is able to support building models with hundreds of billions of parameters, categorising hundreds of millions of pieces of data and training billions of images.

¬ Haymarket Media Limited. All rights reserved.