Autonomy for the Developing World

Autonomous vehicles are now an everyday reality in many cities in the United States and China. Waymo and Tesla have led the way with passenger AVs in the US. Pony, WeRide, and Baidu's Apollo Go are the leaders in China. Soon, we'll see autonomous platforms enter the mainstream in trucking (Aurora, Waabi), construction (Bedrock), public transit (Glydways), and maritime (Blue Water, Splash, Saronic).

Yet the biggest gains for billions of people—safer roads, less congestion, time saved, and 24/7 goods movement—lie in developing-world megacities.

Think of megacities such as Delhi, Jakarta, Bangkok, Sao Paulo, and Manila. Over 100 million live in these 5 cities alone.

1// Regulation: Developing markets lack a national framework and safety standards for AVs. Companies deploying AVs (pure AV players like Waymo or mobility companies like Uber) will need to work with a patchwork of city-level rules. This is a major blocker to scale as city-by-city negotiations are complex and costly. Scale is important especially in developing markets to drive down trip costs. China and Singapore have solved this with national regulations. As a result, Goldman Sachs now sees a path for robotaxis alone (excluding trucking & robobuses) to be a $47 billion market in China by 2035.

2// Localization: Operational design domains (ODDs)—the conditions in which an AV can operate—are chaotic and adversarial in megacities like Manila and Jakarta. Motorcycles dominate traffic, there are no standardized lanes or street signs, and driving culture is aggressive. Extreme weather and flooding will test the most advanced systems. AV driver software for both LiDAR (Waymo) and vision-based systems (Tesla, Wayve) were mostly trained in Western cities.

3// Unit economics: though hardware and software costs for AVs are falling, users in these markets need very low fares. Hence, getting to a large scale (fleet size, passenger volume) and high utilization (passenger miles, occupancy) while maximizing government incentives will be crucial. To be capital efficient, AV fleets need novel business models, such as carving out asset ownership from operations.

A novel solution combines public-private partnerships (PPPs) with an operating system and orchestration layer for AV market participants - regulators, OEMs, AV players, asset owners, capital providers, and transportation companies.

The north star is to help key players in the value chain to de-risk AV deployment using software, capital markets, and enlightened policy.

This platform will turn a capital intensive and asset-heavy venture into a low cost autonomy-as-a-service network. We do this by driving high utilization and separating ownership from operations.

As an example, this is a brief overview of how to deploy robotaxis and robobuses in Mega Manila and replicated in other dense cities in the developing world.

First, a NewCo initiates a Public-Private Partnership with the Office of the President and the Department of Transportation. PPPs are legal frameworks for infrastructure projects funded by the private sector. The Philippines passed a new PPP Law in 2023. Transport infrastructure like airports, bridgeways and tollways are the most common PPP projects.

The PPP deal is: in exchange for the government granting NewCo a master franchise to operate robobuses and robotrucks in fixed routes, NewCo will build and open up its software platform (fleet management, vehicle financing) and infrastructure (charging, teleoperations, incident reporting) to fleet owners who want to launch robotaxis for point-to-point service.

This will be positioned as a Manhattan Project for transport and economic development - the Philippine Autonomous Network (PAN).

PAN gives both regulators and private companies a common orchestration layer to finance, safely deploy and manage AVs at scale.

We'll build agentic AI tools to speed up the process (think PermitFlow for infrastructure deals). NewCo will simultaneously 1) start signing LOIs with AV & OEM partners; 2) acquire training data of local driving conditions (like Waymo's Open Dataset, or NuScenes)*; and 3) run a global competition for researchers to test their AV stack in a developing country. We can also start signing LOIs with AV & OEM partners.

Doing this helps create excitement for the market, while increasing the confidence of regulators, and educating the broader public on the benefits of autonomy.

We can gather local driving data two ways: (1) instrument an existing fleet; (2) crowd-source dash-cam footage and reward contributors with tokens (e.g., Poseidon on Story Protocol).

A somewhat useful precedent is Palantir working closely with the government in the early days to build deep product capabilities, before opening its platform to private companies. One can also think of NewCo as a railroad holding company—but instead of steel tracks it holds an exclusive AV network franchise.

Second, establish vehicle partnerships with global AV companies. For AV and OEM partners, NewCo offers large volume orders. This can take the form of buying / leasing the vehicles (ie; initially 50 Tesla, Pony, or Baidu robotaxis or 10 WeRide robobuses) or licensing the AV driver in partnership with an OEM (ie; Wayve's generalizable AV2.0 stack).

Third, launch AV deployments in defined ODDs with rigorous safety benchmarks. Start with roboshuttles because they run fixed, low-speed CBD routes, cut sensor costs, tap 24/7 BPO passenger traffic, and solve an urgent passenger problem (lack of predictable last mile connections from train stations). At the same, partner with the largest driver unions on reskilling programs to generate goodwill and preempt any labor backlash.

Fourth, open the network to third parties to own AVs as an asset class. After owning the initial fleet and proving the unit economics, third party owners can access an open platform to acquire / lease fleets from AV partners, access financing partners, and enroll their vehicles in the network.

For asset owners, NewCo helps facilitate the end customer relationship and software platform to orchestrate this network (whether deploying robobuses on fixed routes or robotaxis in a transport network like Grab). NewCo can also charge fees for shared services such as teleoperations, charging & maintenance, and fleet management. Financing partners get access to cash flow generating assets.

In sum, the learning loop here is: acquire local training data & simulations → generate regulator trust & confidence → initial deployments in one ODD to prove safety → grow public trust and passenger demand → attract fleet partners → generate more hours / kmh driven