The BYD Problem
What London Tech Week revealed about China's lead, robotaxis and the future of software
Introduction
Last week I wrote about some of the ideas I encountered at London Tech Week. What struck me most wasn’t the technology, but the gap between what investors are discussing and what founders are actually building.
Whether it’s China’s scale advantage, the rapid progress of autonomous vehicles or the way software companies are reorganising themselves around AI, the future seems to be arriving faster than most public market investors appreciate.
I went to London Tech Week for the same reason I went to Italian Tech Week: it’s the fastest way I know to see how founders, VCs and operators are thinking about AI before that thinking reaches public markets.
The question is if investors who dismiss AI valuations without understanding the underlying assumptions are flying blind.
Presentations
I discuss below for free subscribers three themes which emerged
China’s scale advantage is dangerous to ignore
Autonomous Vehicles are moving beyond science projects
AI is starting to reshape organisations
As ever, premium subscribers get the full writeups. Before we get to that, my podcast sponsor AlphaSense has come up with a special offer for my readers:
How many positions have you passed on simply because there wasn’t enough time to go deep? AlphaSense AI-Led Expert Calls run expert interviews on your behalf, so coverage gaps become a thing of the past.
From Tegus by AlphaSense
Stella Li, BYD
She walked through BYD which has 4 main industries - auto, rail, renewable energy and electronics. It was started with 20 people and $350k of capital. They provide a full end-to-end energy eco-system from solar power and battery storage to electric vehicles.
BYD’s scale is hard to comprehend. They have 120,000 engineers in their R&D departments, make 52 patent applications each day, have 42k patents granted and a further 71k have been applied for.
They want to make the car exciting and create an emotional connection to the consumer – there were videos of a car swimming and they make another which jumps over potholes. The Extreme version of the latter, the U9, with 1000 hp, is the fastest car on the planet at 496km/hr; it was shown driving full race around a track at night. Without a driver. Presumably we shall start to see self-driving car races.
They are launching their premium brand, Denza, in Europe. The TV ad has Daniel Craig, the James Bond actor, driving the car aggressively. It launches in the UK at Goodwood in July. I shall let you know my thoughts then.
They are targeting 1000km between charges and already have flash charging where the battery can go from 10% to 70% in 5 minutes and will get to full in 9 minutes, doing this through battery storage so there is no drain on the grid. They are targeting 20k chargers in China by year-end and have installed the first in London.
I wrote that they had $12.7bn invested in autonomous driving, and they have 3.15m vehicles collecting data, with 200m km of data collected each day.
They see the future as the car being like a super assistant – you will ask the car on your drive home “what did the stock market do today?” The car has different avatars which talk to you naturally and understand your emotions. You will use voice control to open the window – the car becomes more than a mode of transport.
This sounds like science fiction, but it’s already happening and China has an astonishing scale advantage – her throwaway line about the AI understanding your emotion from your voice is what the founder of ElevenLabs (the western leader in voice technology) was citing as their next objective.
Investment takeaway: I have been negative on the European OEM auto space for >10 years now, mainly because the EV transition exacerbates the over-capacity and further dilutes returns. It’s hard to see how even a company the size of Stellantis can compete unless Europe bans Chinese imports. It’s not just the cars, it’s the eco-system and the scale of R&D.
Stellantis trades on c.6x 2027 earnings estimates and 2.5x EV/EBITDA. That looks cheap. The problem is not valuation, but forecasting. After listening to BYD, I am less certain than ever what a European OEM earns in 2030 (and presumably the components sector also).
Perhaps the Japanese and US manufacturers will enjoy better protection, perhaps the EU will intervene. But how then do the European OEMs manage without access to Chinese markets? Ferrari used to sell under 10% of its cars in China, Mercedes and BMW about a quarter and these are high profit units. Not so much now, as we just saw in BMW’s latest results. This is too big a subject to cover here.
Alex Kendall, Wayve
AVs are coming and sooner than you think. The company was last valued at $8.6bn and was hoping to launch their trial last week. You can express interest through the Uber app (Account/Settings/Trip Preferences/Autonomous Vehicles). I can’t wait.
Wayve has taken a different approach from other AV systems. It’s not based on high definition maps but on a model of the world with its own onboard intelligence. They are launching in London which has complex environments (20x the road works of San Francisco). By year-end, they will have launched in Tokyo and in 10 cities in 2027.
The car hardware costs just $6k!
It’s a much cheaper approach and it can go into rural areas and navigate situations which it hasn’t seen before. He highlighted the potential savings, not just of lives but 1.5% of GDP lost through accidents.
They have partnered with Nissan (3m cars pa) and Stellantis (5m) and you will be able to buy a car with their tech installed. Nissan has 50 different models and wanted a generalist AI solution. They have also licensed the tech to Uber. Nvidia, Qualcomm, AMD and Arm are all investors. They have partnered with UK online supermarket Ocado and with Asda to mount cameras on their delivery vans to collect data.
The CEO and founder wants to go beyond AVs into robotics, with B2B applications like logistics first then consumer robotics later.
He was asked if an 11 year old today would have to sit a driving test – he failed his first test. He said there would be robotaxis in all major cities in 6 years and you will be able to buy a car with Wayve AI which can monitor your driving and alert you to brake. In future, every vehicle will be autonomous.
Investment takeaway: If Wayve is successful, Tesla and Alphabet’s Waymo will face a serious competitor. It might be good news for Uber and even Stellantis. I don’t know how you can evaluate the probabilities at this stage.
But what is clear is that AVs are coming and sooner than you think. And that’s huge – just one example: what does it mean for UPS vs FedEx vs Amazon. UPS has the dearest drivers but will have to pay them off; FedEx may make lower savings but will have a much lower upfront cost. Amazon is probably the winner. And what happens to car insurance premiums?
Ivan Zhao, Notion
The Co-Founder and CEO of Notion (dressed in a 3-piece suit) was in conversation with Tim Bradshaw of the FT. They got early access to ChatGPT and decided to pivot the company and launched the Notion AI product. Notion used to be a database tool which required very little coding and can be used to generate simple web pages and help with workflows – for example, one content creator used it to map different content topics for posting in newsletters and on social media. I use it to organise my Kindle highlights. It’s exactly the type of SaaS application which was most at risk from AI.
Last Christmas, he and his wife went to Mexico on vacation and used AI to build video games. They were great and they then realised that we were at a tipping point and that knowledge work would develop in a different way. The capability of agents is almost there and he had to think about how to reshape the organisation to adopt the technology.
But this is not straightforward - it’s not like building a bridge. For an LLM to be effective, it’s more like brewing beer - you have to create the right environment and let the yeast do the work. Output is 2-3x higher on coding, less on customer relevant tools.
AI changes how they think about talent. Capability becomes more substitutable by AI, and agency becomes more important – can you bring people along? They now look for people with higher agency and better taste. They are hiring more younger people, straight out of school – they have higher agency, more energy and greater optimism; they then pair them with older people who have more developed taste. You learn faster today than in previous generations – they just hired a 16-year-old coder.
More capable, smarter and more cost-effective models are coming and you need to look at your company and how to reorganise to take advantage.
Investor Takeaway: The CEO of a software company whose core product was vulnerable to AI was leaning into the disruption, rather than defending the old model. Zhao made me think that the debate around SaaS is probably too simplistic. AI may displace some software categories, but it is unlikely to eliminate the systems that own the data. The key for investors may be to scrutinise closely management teams’ actions to embrace AI.
Takeaways
London Tech Week felt different from tech conferences even a year ago. The discussion is less about what AI might do in the future. Companies are already redesigning products, workflows and even organisational structures around it.
The three themes above were my main high-level takeaways from London Tech Week. But some of the most interesting discussions came from private AI companies growing at extraordinary rates and from founders thinking about what happens when AI becomes embedded in every industry.
Below are my notes from some of the fastest-growing private technology companies in the world, including Harvey, Oura and Reflection AI, together with the implications for public market investors.