China’s AI tech unicorn TuSimple got one step closer to become the world’s first public-listed Level-4 (L4) autonomous driving company with its filing of an IPO with the US Securities and Exchange Commission last month. If the IPO is given the go-ahead, the Nasdaq-listing would be a milestone for the nascent smart-trucking logistics industry.
To date, TuSimple has completed 10 funding rounds and raised over $800m from prominent backers Sina, Nvidia, CDH Investments, UPS, Werner Enterprises and Goodyear Ventures. Investors and logistics companies are betting on autonomous trucking technologies to revolutionize the global truck freight industry estimated to be worth $4tn.
However, like previous pre-IPO tech unicorns Uber and Lyft, TuSimple has yet to make a profit. Net losses jumped from $45m in 2018 to $178m in 2020. The $308m total net loss for three consecutive years from 2018 was mainly due to increased R&D personnel and expenditure to expand the company’s fleet of autonomous semi trucks, TuSimple said in its IPO prospectus.
A pragmatist "who believes [a business] is certain to make profit by creating value for the world,” CEO and co-founder Chen Mo told media that profitability could be achieved with an unmanned fleet of just 5,000 trucks generating $300m per year. According to him, there are 3.5m trucks currently operating in the US.
Founded in 2015, and with headquarters in both China and the US, 50 of the company’s 70 trucks are based in the US, each generating $100,000–$200,000 per month in revenue. Over 80% of its 839 full-time staff are R&D personnel based mainly at the company’s San Diego HQ, with the rest of the team stationed in Arizona and China. The US freight truck market is estimated to be worth $800bn, according to the company’s IPO prospectus.
US driver shortages
The US is an important market for TuSimple where scalable smart-trucking solutions are urgently needed to help traditional haulers tackle acute driver shortages and high operational costs. Labor accounts for over 43% of the per mile semi truck freight cost structure in the US.
The American Trucking Association forecast that driver shortages, which more than tripled from 2005 to 2018, is expected to rise 2.5 times to over 160,000 by 2028. With fully autonomous trucks, the US truck freight industry could cut overall operating costs by about 45%, saving $85bn–125bn every year, according to McKinsey.
Profitability could be achieved with an unmanned fleet of just 5,000 trucks
According to TuSimple CTO and co-founder Hou Xiaodi, the US truck freight industry has a higher labor cost [than China] and the policies are more business-friendly for self-driving tech companies. After several successful pilot projects, TuSimple has started various commercial projects in the US with global partners.
In 2020, a pilot program with UPS was expanded to 20 trips per week, including the Arizona–Texas and El Paso–Dallas routes. TuSimple also provides trucking services to over 12 corporate clients in the US, including household brands and Fortune 100 companies. A global partnership with automotive supplier ZF will also expedite expansion in North America as well as Europe and China.
The majority of TuSimple’s R&D and engineering facilities in the US are in San Diego. It has a truck depot in Arizona. The company works with truck manufacturers like Navistar and Peterbilt, powertrain specialist Cummins and tech software companies like Nvidia to design and build customized autonomous L4 trucks for B2B long-haul trucking clients in the US.
Freight version of Waymo One
TuSimple aims to be more than just a traditional hauler providing cargo transportation services using driverless trucks. The company is currently building the world’s first Autonomous Freight Network (AFN) in the US with strategic partners including UPS, Penske Truck Leasing, US Xpress and McLane Global Logistics.
Chen’s vision is to build the AFN into a freight version of Waymo One, a ride-hailing app for driverless passenger vehicles developed by Google spin-off Waymo. The AFN will connect hundreds of shippers, carriers, railroad companies, freight brokers, fleet owners and truck hardware suppliers across the US. Users will have access to a range of autonomous trucks, HD digital route maps, transport terminals and the TuSimple Connect autonomous operations monitoring system.
Expected to be fully operational by 2024, the AFN will provide autonomous L4 semi truck services in partnership with shippers, carriers and fleet operators strategically located on the routes. The nationwide transportation network will help logistics service providers reduce costs, boost efficiency and improve safety standards.
However, the AFN business model will only be viable through mass production and adoption of autonomous trucks by industry partners. The Navistar-TuSimple partnership, for example, has accepted over 5,700 (non-legally binding) reservations to make customized L4 autonomous semi trucks for about 10 potential customers, with mass production expected to start in 2024.
The company is building the world’s first Autonomous Freight Network in the US
“The manufacturer needs time to model the vehicles, select auto parts, validate the prototypes and set up mass production lines,” Chen said in 2020. “[Four years] is the minimum amount of time required,” he added. Customers will then be able to purchase the autonomous trucks through Navistar’s traditional sales channels in North America.
The company will also continue to grow its business in China and eventually expand into Europe by making the L4 semi trucks purpose-built specifically for both markets, in partnership with TRATON, a subsidiary of the Volkswagen Group, one of the world’s largest producer of commercial vehicles.
Easier R&D, profits
TuSImple was initially founded 2015 as Beijing TuSimple Interconnection Technology to provide image recognition services to internet giants in China. The startup managed to quickly secure RMB 50m Series A funding from social media giant Sina’s investment arm Wei Venture Capital. When the concept of autonomous driving gained momentum a year later, companies, including Tesla and Nvidia, started developing self-driving technologies.
But TuSimple decided not to jump on the robotaxi bandwagon and decided to create robotrucks instead because self-driving trucks offer an easier commercial market entry point. “If you want to make profit through robotaxis, $50bn–100bn investments are required,” Chen said. “According to our estimate, only about $1bn is needed to make a profit from self-driving trucks,” he added.
Long-haul autonomous trucks usually follow fixed routes, making R&D easier than for passenger vehicles that have to navigate complicated urban traffic. As a result, TuSimple’s image recognition technology was quickly pivoted to the development of self-driving trucks in 2016.
Since then, TuSimple has improved and validated its autonomous systems for freight trucks. The company also looked out for prospective investors with key roles in the autonomous driving startup ecosystem. Besides prominent investors like UPS, Navistar and TRATON, global chip giant Nvidia also took a 3% stake in 2017.
“Nvidia is capable of providing the computing capability required by deep learning algorithms that are widely used in autonomous driving,” Chen said. “It can also optimize and customize products according to our R&D needs.” By 2024, when TuSimple and Navistar begin mass production, the self-driving semi trucks will be AI-powered by NVIDIA DRIVE AGX with enhanced safety features.