- Mapbox has received $280 million in Series E funding from SoftBank.
- The latest investment will enhance AI-powered map and location services.
- Riding on the generative AI wave, Mapbox has launched AI voice assistant MapGPT.
Many industries are grappling with the question of how best to exploit the power of AI and, in particular, generative AI. The location industry is no different. Last year, we saw the introduction of HERE’s AI-powered Unimap, which kickstarted the race among digital mapmakers. SoftBank, which has been an active investor in AI ventures, has poured $280 million into a Series E funding round of Mapbox recently. Earlier, the map company had raised a total of $360 million since its inception in 2011, with $160 million coming from SoftBank in 2017 to help Mapbox’s expansion in new industries and regions. The recent investment from SoftBank can be seen as another AI-centric bet in its portfolio to support Mapbox’s efforts to develop its own AI-based map and related location services.
Let us have a look at Mapbox’s current offerings in the automotive industry and how its new AI-powered products are addressing the challenges posed by ADAS and electrification:
Mapbox Autopilot Map works by combining data from vehicle sensors with maps. Using AI, the system detects the changes in the road when enough vehicles have passed through a certain point and then integrates the changes into the Autopilot Map. This keeps the map fresh and improves the coverage as Mapbox customers drive billions of miles every week. The update process is very bandwidth efficient as the tile that includes the data gets updated instead of downloading the whole map. This also means the maps get updated at a very low cost.
Mapbox for EV is a service that addresses the flawed EV experience for customers, OEMs and the charging point operators (CPOs). Mapbox offers three EV products to address these three stakeholders. Mapbox EV Routing service takes account of the in-vehicle battery system and monitors energy consumption patterns to predict the range of the vehicle by considering the vehicle type, road elevation and weather conditions. Mapbox EV Charging service tracks the real-time availability of charging stations and offers secure integrated payments. Using its AI and ML technology, the company collects data on energy consumption patterns, availability, compatibility and performance of the charging station. This allows OEMs to get insights into the overall EV experience and provide a seamless driving experience by partnering with CPOs.
Jumping on the generative AI bandwagon, Mapbox has introduced MapGPT, a location-aware conversational AI service. MapGPT is based on live data and location services from Mapbox. It can have in-depth and richer conversations about real-time traffic, POIs, road networks and navigation. MapGPT uses Large Language Models (LLMs) from ChatGPT3.5 and ChatGPT 4.0 when connected to the internet, while the offline MapGPT uses a Tiny Large Language Model (LLM) from Mapbox. MapGPT comes with plugins which are also called “actions” with third-party integrations for music and restaurants. It is highly customizable as it allows custom actions like cabin temperature. OEMs can also integrate their own LLM into MapGPT. Customizations for voice, personality, animated avatar and wake word are also available. Mapbox has announced partnerships with a Japanese payment service called PayPay which allows drivers to purchase and pay for services.
Over the years, Mapbox has emerged as a developer-focused location platform with excellent graphics rendering capabilities. With its open-source and modular architecture, the company is posing a serious challenge to the incumbents. Mapbox secured the fourth position in Counterpoint’s Location Platform Effectiveness Scorecard 2022.
The recent funding is expected to help Mapbox in the following ways:
Expansion in automotive industry: The company has managed to pivot well into the automotive business by acquiring customers like BMW, Toyota and General Motors for its navigation services. However, carmakers are experiencing a paradigm shift led by technological innovations like electrification and autonomy. They are struggling to offer best-in-class solutions for EV driving or assisted driving as both these rely on highly accurate location-based data. Realizing this gap, Mapbox wants to bring AI into most aspects of mapping and navigation to offer intelligent maps leveraging the growing number of data points they are receiving. The company will be putting more resources into building maps (ISA, ADAS/AD) and services (EV routing, Charging Points and Payments) contextualized data to address the challenges posed by ongoing trends like autonomy and electrification. BMW Mini EV will be equipped with Mapbox technology including EV services.
First-mover advantage: Different players across the automotive value chain are looking at various use cases to integrate generative AI in the car, one of them being the personal assistant. Since the emergence of ChatGPT, only a handful of companies have explored the generative AI potential to elevate the cockpit experience. Hence, through MapGPT, the company is drawing the attention of automakers to create fully customized voice assistants capable of having natural and richer conversations. Mapbox is well-positioned to marry its location-based services like navigation to build customizable intelligent AI voice assistants for natural and rich conversations on the go. Having a first-mover advantage, the company can seek partnerships with automakers, Tier-1 players and even semiconductor companies like Qualcomm and Nvidia to customize its MapGPT based on their chips.
Growth and valuation: Mapbox has raised a more than $600 million over eight rounds of funding since its inception. In 2021, the company was in talks to go public through a SoftBank-backed SPAC at a valuation of $2 billion. However, Mapbox decided to remain private. Riding on the AI hypercycle, Mapbox’s valuation is expected to go higher. This increased valuation will likely attract further investments and partnerships.