Back to Media Center

To solve the Autonomous Vehicle investment problem, the industry needs to adopt a horizontal integration model

To solve the Autonomous Vehicle investment problem, the industry needs to adopt a horizontal integration model

By Manish Singh, VP of Marketing

Autonomous vehicles will make the world’s roads safer. The economic and societal harm from motor vehicle crashes amounts to a whopping $871 Billion in a single year, according to a study released by the National Highway Traffic Safety Administration (NHTSA). Road traffic crashes cost most countries ~3% of their gross domestic product according to the WHO, which is >$2T annually. Human error is involved in 95% of all motor vehicle crashes, according to the NHTSA, creating a problem that self-driving cars can address.

However, the cost of developing Autonomous Vehicles (AVs) is astronomical, and only a select few companies can afford it. Last week, Alphabet’s self-driving car company, Waymo, raised a whopping $2.25 billion in its first external funding round. Clearly, investment is needed to build transformative products.

As an industry, we should consider that the vertical integration business model may not be the right strategy to address such a large-scale issue. The burgeoning AV industry needs to transition from this approach to a horizontal one in order to reduce the cost and complexity of development. By making this switch, we can unleash the intellectual prowess of smaller companies’ best-of-breed expertise by allowing them to participate in the marketplace.

For the horizontal model to operate successfully, players in the ecosystem will have to be aligned. We need the development of standards for hardware, software, the interoperability between the two, and the compliance to certify the alignment of everyone involved.

Enter our company: Recogni. We specialize in vision cognition, which is an instrumental part of the AV solution. As a specialist, in order to solve such an important problem, we need to interoperate and interface with the other components of an AV system. Specifically, an industry-wide perception API will aid the performance and execution of perception, which is integral to the autonomy puzzle.

Vertical vs. Horizontal Integration

Today, emerging automotive companies, such as Waymo and Tesla, build the majority of products in a vertically integrated model. As we have mentioned above, the rest of the industry needs to react by adopting a horizontally integrated model and sourcing most of the hardware and software components from highly specialized vendors.

A company that has successfully executed vertical integration is Apple. The tech giant creates premium products by controlling their supply chain and setting strict requirements for their vendor partners. They build many components themselves, and the hardware and software operate seamlessly because they were designed specifically for each other.

On the other end of the spectrum is horizontal integration, with product examples such as Android phones or PC laptops. These products may have a different look or feel from Apple products, but they are much less expensive. The components are more standardized, and they typically come from a large number of vendors. There are significant differences in the products, target customers, and corresponding requirements with regards to either vertically or horizontally integrated products.

Each strategy is different when it comes to building products:

Manufacturing: The difference between vertical and horizontal integration depends on the scale of the effort. A vertically integrated company builds most components internally. This increases technological and execution risks and makes it difficult to scale, since a company’s production volumes will be lower than its horizontally integrated counterparts.

Tesla has struggled to mass-produce its vehicles and compete with less-specialized auto manufacturers from a scaling standpoint. Avoiding the design and manufacturing process of many key components within the product is an advantage for the horizontal model. Essentially, a horizontally integrated company may have less control over vendor quality, but it has more options regarding the supply.

With horizontal integration, a company may develop some of its own secret sauce, which allows it to focus only on what it excels in.

Customization: A vertically integrated product often uses custom components and interfaces in order to build a highly optimized system. This includes time and money when it comes to investing to develop these custom interfaces and components. In contrast, a horizontally integrated product will use standardized interfaces which enable vendor diversity, minimize risk, and promote scaling.

Investment: In order to minimize the risks and associated costs with supply, a horizontally integrated company will invest in operations, procurement, and supply chain management, which may be more important than the R&D processes for the product itself.

Costs: The costs required to develop and build custom components are prohibitively high, so vertical integration generally leads to a more expensive product with fewer margin-stacking issues. Highly focused product development gives the vertically integrated company an advantage to develop its target markets.

Scale: In the case of vertically integrated products, the potential for scaling is limited because they are restricted by how much the internal vendors can produce. With horizontally integrated products, the company and its competitors can scale very quickly.

Ecosystem: In the case of horizontally integrated products, there are multiple vendors selling products, thus the overall ecosystem is large. In the case of vertically integrated products, there is a limit to the number of products a company can sell, so the ecosystem is much smaller. Effectively, horizontal integration enables access to the best-of-breed technology and yields superior products for the consumer.

The absence of standards may stifle the rate of innovation in horizontally integrated markets, especially with regards to software. Hardware products may be compliant to broad market standards due to industry adjacencies, but software is less likely to be so.

The automotive industry must also be concerned with standards that affect their ability to sell their products in specific markets. Examples of this include meeting procurement specifications in domestic markets or complying with certification requirements for foreign exports. Moreover, as many products now become components of larger systems, part of meeting specifications is demonstrating compatibility with the performance attributes of other products.


You May Also Like

3rd Party Media
3rd Party Media
Articles
Converting a network for the fastest inference chip on the market

Recogni Core-Tech Team *Conversion: Making a full-precision network compatible with our chip Introduction Deep Neural Networks (DNNs), in particular convolutional ones, are the go-to approach for visual perception tasks that…

Subscribe to Our Updates

Join our newsletter to stay up to date with Recogni!

Enabling ADAS to Full Autonomy