In addition to the cold winter of global capital markets, investors are becoming wary of their fortunes, more importantly because self-driving companies are facing technical barriers, coupled with high investment, long return cycles and commercialisation, where the number of start-ups has not exploded, while those who have survived previous booms have opted to go dormant, even if they are not in a hurry to expand blindly, and some have even begun to seek alternative products to save the curve.


Recently, Starsky Robotics announced it would downsize its fleet. The company, which has truly achieved pilot services for driverless trucking operations, has been underwhelming, with even news that \"Starsky Robotics is in talks with potential buyers, including even some competitors, to stay in business.\"


Even if StarskyRobitcs end up being forced to sell itself, it won't come as much as a surprise, after all, that a star company valued at $200 million ended up being bought, laid off and shut down on a small scale. At the beginning of the year, a chinese start-up made headlines, not because of its impressive results, but because of the collapse of its founding team, its unreasonable ownership and governance, and its low prices in capital markets.


This is just the beginning. With the industry's attitude to self-driving moving from \"blindly pursuing\" to \"returning to reason \", future capital flows to the emerging industry will only become more cautious.


According to data from Crunchbase, there are only a handful of self-driving start-ups with big funding this year. On the one hand, investors have spent their money in previous years of expansion, and when the real technology is ripe, good companies are in the embarrassment of having no money to invest in; on the other hand, self-driving, which is also a branch of the AI field, is not able to fall to the ground as quickly as other products in commercialization, with long returns and high investment objective factors that make today's capital markets daunting.


Therefore, the field of artificial intelligence may advance into the contradiction between lack of money and need of money. Especially self-driving, commercialization needs a lot of financial support behind, technology competition has even evolved into capital competition. A prominent phenomenon this year is that car companies have replaced wind as the \"gold owner\" of many start-ups. Volkswagen, for example, has become an equal shareholder with Ford; the Renault-Nissan alliance has not only supported Waymo's expansion, but has also become a round-A investor in the company. It is with the ability to keep financing burning money that companies such as Nuro and Aurora have avoided the same fate.


Some start-ups have even come to terms with investors to tackle the lack of financing by making quick money. And because of the long liquidity cycle, the capital winter unicorns also live trembling. It has previously been reported that the company has been so tight in revenue and projects this year that there has been a backlash from time to time that "it shouldn't take immediate benefits too seriously and ignore the long-term value of the company.” 」


But reality is always reality. The value of artificial intelligence lies in its subversive change in human life. But the technology investment will certainly face the landing difficulty, the cycle is long and so on question. But in this competition extremely intense self-driving breakthrough competition, the strong technical strength is indispensable, the stable capital blessing is very necessary.


And if the impact of capital is removed, the whole autopilot industry will develop smoothly in 2019, and after the small elimination competition, whether it is the mainframe factory, the supplier or the technology company, it is almost in constant exploration.


Because there is no way to set clear goals, most car companies choose a more secure \"two-legged walk\" approach: first, to explore the robo-taxi service operation mode; and second, to gradually iteration of the auto-driving capacity based on the idea of mass production. Although there is a degree of overlap, but generally two relatively independent direction.


But most mainframe factories lack confidence in the robo-taxi issue, because the realization of l2 autonomous driving capacity is not too difficult for the mainframe factory, after all, people always control the driving power. But robo-taxi is totally different, the main body of security responsibility will be completely transferred from the human to the machine, so the need for vehicle hardware and software system solid. But there is a general consensus that full-stack self-driving technology is very difficult to achieve, and few companies now have the technical power to match.


Around mid-november 2019, daimler's ceo said in public that it would conduct \"internal checks\" on the robotaxi business and acknowledged that making them safe was far more difficult than initially thought when future earnings potential was questioned. Daimler will also adjust its spending on robotaxi and self-driving technology, with the potential to first apply it to commercial vehicles at freight companies. There is a similar consensus among the group.


The industry as a whole has gone from its first blinds to its gradual return to rationality. Traditional car companies that hold on to tech companies and start-ups are also groping to find their niche. For now, until 2030, it is unlikely that there will be a certain amount of market for full self-driving, and in this doomed long-term tug-of-war, mainframe factories are faced with such daunting tasks as recruiting software talent, building partnerships, and persuading consumers to pay the bills.


Clinging to each other reduces the financial pressure on partners and gives start-ups access to the huge OEM's large customer resources. BMW, Mobileye, Intel have provided precedents, and Volkswagen Ford Argo AI has followed. Such alliances will only grow in number.


The \"limited scenarios\" here include mining areas, ports, parks, etc., following the established commercial drop line, covering scenes including parking, logistics distribution, street cleaning, etc. For example, Nuro, a Silicon Valley start-up that has received $100 million in B-cycle funding this year, aims to develop and deliver unmanned vehicles to solve the \"last kilometer\" distribution problem. In addition, there are already a number of self-driving companies in the country have started to run in this field, in order to get more early bird dividends. Such as Yu potential technology, smart walker, Xijing technology, rookie, Yingcher technology, Zhijia technology, Tucson future and so on.


Compared with the whole scene driverless, the closed area under the single line, the algorithm requirements are relatively low, more conducive to mass production landing. But although this field has succeeded in creating a number of new companies, and the capital is also coming in, but the problem is that landing does not mean that there is value, and there is market demand to release the corresponding product value, the current application of self-driving to limit the scene is also in the exploration, there is no absolute rigid demand.


Looking back on 2019, the calm and restraint of the capital markets allowed the entire self-driving industry to enter a slow-growing cycle, and each appeared to be dormant for the winter in an attempt to conserve energy to return at the right opportunity. Some companies are even being forced to look for opportunities in other product directions in the hope of boosting their competitiveness through \"a side business \". For the mainframe plant,2020 could face pressure to deliver on its \"mass-produced L3 self-driving car\" promise. But in the light of the current situation, this point may continue to be delayed. Because even if the technology is mature, the corresponding laws and regulations will affect the landing of mass production vehicles.


One thing is certain: the autopilot industry has gone through a period of passion and fanaticism. Almost all the companies in the field have grown rational and patient in the protracted war. Usually people no longer blindly pursue the ability to achieve full self-driving, but take a few steps back to innovate in service mode or commercialization.


geek park (id: geekpark) believes that the large-scale application of high-level self-driving technology should be the robo-taxi net car service. The direction of phased iteration through ADAS may undergo a revolutionary process. There is now a general consensus on the point at which \"full self-driving\" will be achieved, at least until 2030.