CES 2025 Day 2 – Autonomous head fake

Busy floors, apocryphal autonomous driving

Floor report – Orderly business

• The floor was even busier on Day 2 than it was on Day 1 but the chaos that usually accompanies this has not materialised.
• Meetings and catch-ups are not getting moved or shuffled around and everyone is turning up pretty much on time for everything.
• This is an unusual turn of affairs but suggests that attendees have come to CES to conduct business as opposed to killing time in pointless meetings to justify an evening spent in the casino.
• I expect that Day 3 will be quieter with Day 4 being catatonic by comparison.
• This is setting the industry up for a pretty good year from a consumer electronics perspective outside of the automotive industry where volume shipments remain under pressure from both a weak consumer and real competition from China.

Autonomous driving – The Turn?

• After several years of despondency, the word on the floor is that autonomous driving has turned a corner and will soon be showing the kind of improvements that will allow autonomous driving to become a reality within a reasonable time frame.
• A key part of this is the idea of using generative AI to create synthetic sensor inputs for unusual situations on the road that can be used to train the machine algorithms and reduce the rate at which they make mistakes.
• I am generally cautious about the use of synthetic data because reality is always stranger than fiction meaning that a computer will never be able to think of or anticipate most of the odd events or situations that occur on the road.
• This leads me to think that synthetic data will not solve the autonomous driving problem and that humans are going to be driving vehicles for a while yet.
• I continue to think that it is not until the machines start to really understand the road for what it is rather than as a series of statistical connections that the problem will be solved but this will be easier said than done.
• This is particularly the case because investments in generative AI and compute infrastructure are sucking all of the investment dollars available meaning that there is not much left to go elsewhere
• My 2028 target for the true commercialisation of autonomy (level 4 and up) continues to look increasingly optimistic.