Scouting for new AI use cases & top Generative AI Autonomous driving providers

?? We are at the crossroads of a technology revolution, digital transformation and energy transition.

I started this newsletter to share my thoughts on how to prepare for the disruption and how to monetise emerging opportunities.

My OS (Operating System) = ??Innovate ??Monetise ??Automate

Join me on the journey towards automation and continuous innovation.

• ?? Discover innovation opportunities, solution providers and process automation to manage disruption and monetise trends across industries
• ?? Scout the Future of Automotive & Mobility: breakthrough vehicle technology, changing passenger UX and new business models
• ??Find digital tools to boost productivity and process excellence.

This newsletter in a nutshell

??I analysed funding in +1000 AI start-ups (+$35.5 Billion) to find use cases with high market potential & tech readiness

??3 innovative players in Generative AI & LLMs for Autonomous Driving: Haomo.ai DriveGPT, 百度 Level 4 ASD and Waabi Level 4 Autonomous Trucks ????

?? Top Level 2-4 Autonomous Features & Product Launches from my Trackers

I'm doing a new Workshop on AI-Autonomous Driving on October 23.

?? Check the dates and program here.

??I analysed funding +1000 AI start-ups to find use cases with high market potential & technology readiness

My scouting methodology

I'm analysing start-ups because they have identified industry pain points and present innovative ways to solve them, e.g. AI Employees

Of course, you can get signals from industry investments as well, such as Microsoft , BlackRock and 英伟达 investing +$30 Billion to build data centres. But, I'll leave this for Part 2.

I'm examining funding data from various proprietary sources because of how the size of the deal ($ Billion) and who is investing signal gaps, opportunities and investment strategies (e.g. Oil&Gas majors investing in EV charging infrastructure).

My search extends beyond just Automotive (Autonomous vehicles, Electrification, etc.) to reduce any bias from industry-specific needs, representative data and use cases.

This will be helpful as I explore non-automotive use cases relevant to automotive (e.g. via tech transfer).

My assessment criteria are:

• Market Potential: Total Addressable Market (TAM), TAM Growth, Value-add, Differentiation
• Technology Readiness: Innovation (e.g. Patents), Tech Readiness Level (e.g. Prototype)
• Business Readiness: Founder / Team, Competition, Customer voice (e.g. NPS)

?? Here are 3 things I have found from Start-up funding (Part-1)

?? Signals from Start-up funding data vs Industry Adoption converge to AI use cases in manufacturing, health and information showing potential

AI startups are raising big bucks, across funding rounds. Funding is booming led by OpenAI competitor xAI, AI hyperscaler CoreWeave and Autonomous Driving start-up Wayve.

+1000 AI start-ups raised $35.5 Billion in Q1-Q2 2024, led by Musk's xAI

To give some perspective, investment in start-ups working on Autonomous Driving in Q1-Q2 2024 amounted to $3.9 Billion.

The start-ups operate across many industries. To name a few:

• ?? Artificial Intelligence | Elon Musk's xAI raising $6 Billion to race OpenAI ;
• ?? Supercomputers | CoreWeave , a AI hyperscaler, delivering a cloud platform that manages complexity through automation, making supercomputing accessible, raised $1.1B;
• ?? Automotive | Wayve working on EmbodiedAI for autonomous driving raised $1.05 Billion;
• Other industries include Data management for AI applications, AI-Health / Medical, AI Sales;

For insights into the actual adoption of AI in businesses, read a working paper by the National Bureau of Economic Research about the early adoption of AI in the U.S. The researchers defined AI adoption as using AI for production.

What's more, AI is being adopted unevenly in the U.S., with use clustered in large companies (+5,000 employees) and industries such as manufacturing and health care.

?? AI start-up Funding in Automotive focused on Embodied AI, compute & data

?? Embodied AI for Autonomous Driving (Wayve), AI computing platforms (Recogni raised $102M for its AI inference system), AI accelerators, data platforms (Pateo), predictive analytics and Autonomous Trucks (Gatik) are some of the key domains based on funding.

The majority of the funding comes from those working on Level 4-Autonomy because of Wayve's funding round. Wayve, a UK company working on EmbodiedAI for autonomous driving, raised $1.05 billion Series C.

Wayve believes that Embodied AI is the next step beyond Generative AI and LLMs to teach vehicles how to drive autonomously

?? Scouting for use cases to Act, Prepare or Watch

Signals from Start-up data vs Industry Adoption converge to AI use cases in manufacturing, health and information as those to Act on, e.g. invest, buy or build

Looking at the state of AI, I expect demand for power to train AI models and process data will skyrocket!

Here are some new use cases I find interesting based on their market potential and tech readiness. To make it more actionable, I'm stating examples in vehicle applications.

1. Portable generative AI solutions for Customized Driving Experiences: use generative AI to create personalized in-car experiences and interfaces based on driver preferences, e.g directly to users' devices.
2. Platform to manage Vehicle Data for AI development: Implementing AI-driven data management platforms to handle the vast amounts of data generated by autonomous and connected vehicles.
3. AI Accelerators for In-Vehicle Processing: Vision processors for edge devices that enhance the performance of in-car AI systems, including ADAS and infotainment.
4. Embedded AI for Vehicle Systems: Embedded systems to develop software-centric platforms for embedding AI with Machine learning in critical vehicle systems like braking and steering.
5. Multi-lingual AI agents for use cases such as AI Sales or AI employees (Artisan AI BDR).

?? Stay tuned for Part 2 where I'll share more data and my quantitative methodology.

?? 3 players to watch ???? in Generative AI & LLMs for Autonomous Driving

Generative AI has strong potential to solve challenges in traditional approaches to achieving Level 4 Autonomy.

I'm looking for innovative suppliers or OEMs who are developing new ground-breaking solutions, and their technology readiness is high to hit the market strongly.

Three Chinese players show how they have developed solutions for Autonomous Driving using large-scale models and often a vision-only approach.

• Haomo.ai with DriveGPT an autonomous driving generative large-scale model aiming at end-to-end self-driving cognition.
• 百度 with their Level 4 Apollo Self Drive (ASD) and end-to-end large-scale model
• Waabi in Autonomous Trucks

?? Haomo.ai launched DriveGPT in WEY as L2+ AD ( or is it L4?)

GWM’s brand WEY launched the Mocca (Coffee in Germany) with NOH (Navigation on HPilot) powered by Haomo.ai’s DriveGPT, 高通 ’s SnapDragon Ride and RoboSense(速腾聚创) ’s M1 LiDAR.

Haomo.ai has trained DriveGPT with the same GPT algorithm used by OpenAI and data from over 40 million km of driving.

Their goal is to create an autonomous driving generative large-scale model aiming at end-to-end self-driving cognition.

In April 2024, Haomo.ai and Qualcomm announced the Haomo HP370.

?? Baidu's Level 4 Apollo Self Drive features in JIDU Auto's Robocar 07

In September 2024, JIDU announced the upgrade from PPA intelligent driving to ASD (Apollo Self Driving),?the Level 4-AD large-scale model.

JIDU Auto, a Joint-Venture between Baidu and Geely, is the first brand to adopt Baidu Apollo ADS, which combines pure vision with an end-to-end large-scale model to support fully unmanned Level 4-Autonomous Driving tailored to Chinese roads.

The company claims this is the only system in China and one of only two globally vision-based Level 4-capable systems, together with Tesla.

However, the company believes its system is more powerful in China because of training in domestic roads and its 2-stage, end-to-end large model approach vs. Tesla FSD’s 1-stage.

?? Waabi's Copilot4D: A Foundation Model for Autonomous Trucks

In March 2024, Waabi introduced their Copilot4D: A Foundation Model for Self-Driving.

The integration of generative AI into Copilot4D enables it to learn from vast amounts of sensor data without requiring extensive human supervision. This is pivotal for scaling autonomous driving technologies effectively and safely.

It allows them to predict future scenarios based on real-time data from LiDAR sensors.

The model enhances decision-making processes by allowing vehicles to foresee how other road users will behave, thus making proactive rather than reactive decisions

Copilot4D, the first foundation model that explicitly reasons in 3D space and the fourth dimension, time, learning remarkable capabilities for interacting and acting in a dynamic world, whether in simulation, like Waabi World, or in the physical world we live in.

?? Which other innovative companies have you spotted?

??Top ADAS Product Launches & L2-3 Features from my Partially-Automated Trackers

?? ADAS update: Level 2 & 3 Feature launches YTD 2024

?? I analysed +50 product launches from Audi, BMW, Ford, Mercedes, Tesla and mapped the most interesting by SAE Automation Level, Cruising / Parking and Timeline.

?? The launch of Level 3 features from BMW Group and Mercedes-Benz AG-Benz standout, but they are not activated yet. ZEEKR极氪 obtained a license for Level 3 automated driving tests in China

? Level 2 features are proliferating across carlines, see here Stellantis and VW.

?? Shall I do a separate one for Autonomous Trucks?

?? I analysed +100 ADAS Features, Product & Service launches in 2024 to find innovators

?? Comment below to get access.

Key findings:

?? GenerativeAI: Helm.ai, DS Automobiles, Volkswagen Group, Peugeot

???? Indian Institute of Technology, Bombay-Hyderabad develops autonomous vehicles for Indian roads ??

? Level 2 launches: Xiaomi Technology, GAC Motors, AUDI AG, Jeep, Mercedes-Benz AG, XPENG

?? Autonomous Trucks: Scania Group, FERNRIDE

That's all for this week!

?? Did you find it interesting?

?? What should I cover next?