How are you thinking about the fact that Tesla is not purchasing both tactile and torque from VPG?
I feel like this is a massive signal that they have already in-housed the hand sensors that are "good enough" for their applications.
Furthermore, if Figure only needs tactile sensors but no torque sensors that would also signal to me that either 1. it is possible to in house the torque sensors too or 2. there is another viable torque sensor supplier.
Both of those facts make me extremely weary to put long term faith into the revenue opportunity at Tesla.
I think it's a legitimate point to make and something I actually plan to cover with the IR team when I meet with them next. They are not the only sensor supplier in the marketplace, but they are the premium sensor supplier. Also, these design phases with Tesla and Figure have been many years, which signals that once they are embedded into the supply chain, it may not be as simple to take them out. Also, VPG has scale production capacity to support a humanoid hockey stick ramp and their pricing goes down with volume. I don't think either company is making their sensors in house, but could be using another supplier for torque/tactical. One thing that is a big positive is that VPG has had commercial conversations with both customers in the last 6-9 months on cross-selling for both tactical and torque sensors, albeit those conversations have not moved to design conversations yet.
Torque sensors are not only what VPG is better at but also the “harder” more defensible sensor. Tactile is arguably a race to the bottom with new technologies threatening to push margins to near 0. Temperate independent torque sensors on the other hand are vastly harder to make and are not only protected by patents but also “black magic” manufacturing procedures.
Torque sensors have a much larger TAM than tactile sensors as there are arguably 20x+ more torque sensors in each robot than tactile.
Given Tesla has already had their torque sensors in Optimus and is now re-configuring their factories for Optimus production it is very likely that the switching costs for these torque sensors is extremely high. You not only have to rip and replace the entire “connective tissue” or the nervous system of the robot but you also have to re-collect the mountain of training data that was configured with the VPG torque sensors specifically. Not to mention re-tooling a factory assembly process to have a new nervous system would likely lead to massive down time and delays.
Lastly it is likely that Figure is using some in house or alternative torque sensor but I would argue that they will find these to be inferior in temperate exposed environments (factories, dish washing, etc) so much so that they will opt for the top of line VPG sensors. We can see this by the ongoing cross sell conversations that VPG is holding. It is likely that they (like Tesla) tried their hand at in housing this tech and will realize that the reward is simply not worth the risk or cost.
How are you thinking about the fact that Tesla is not purchasing both tactile and torque from VPG?
I feel like this is a massive signal that they have already in-housed the hand sensors that are "good enough" for their applications.
Furthermore, if Figure only needs tactile sensors but no torque sensors that would also signal to me that either 1. it is possible to in house the torque sensors too or 2. there is another viable torque sensor supplier.
Both of those facts make me extremely weary to put long term faith into the revenue opportunity at Tesla.
I think it's a legitimate point to make and something I actually plan to cover with the IR team when I meet with them next. They are not the only sensor supplier in the marketplace, but they are the premium sensor supplier. Also, these design phases with Tesla and Figure have been many years, which signals that once they are embedded into the supply chain, it may not be as simple to take them out. Also, VPG has scale production capacity to support a humanoid hockey stick ramp and their pricing goes down with volume. I don't think either company is making their sensors in house, but could be using another supplier for torque/tactical. One thing that is a big positive is that VPG has had commercial conversations with both customers in the last 6-9 months on cross-selling for both tactical and torque sensors, albeit those conversations have not moved to design conversations yet.
Did some more research and here are my thoughts:
Torque sensors are not only what VPG is better at but also the “harder” more defensible sensor. Tactile is arguably a race to the bottom with new technologies threatening to push margins to near 0. Temperate independent torque sensors on the other hand are vastly harder to make and are not only protected by patents but also “black magic” manufacturing procedures.
Torque sensors have a much larger TAM than tactile sensors as there are arguably 20x+ more torque sensors in each robot than tactile.
Given Tesla has already had their torque sensors in Optimus and is now re-configuring their factories for Optimus production it is very likely that the switching costs for these torque sensors is extremely high. You not only have to rip and replace the entire “connective tissue” or the nervous system of the robot but you also have to re-collect the mountain of training data that was configured with the VPG torque sensors specifically. Not to mention re-tooling a factory assembly process to have a new nervous system would likely lead to massive down time and delays.
Lastly it is likely that Figure is using some in house or alternative torque sensor but I would argue that they will find these to be inferior in temperate exposed environments (factories, dish washing, etc) so much so that they will opt for the top of line VPG sensors. We can see this by the ongoing cross sell conversations that VPG is holding. It is likely that they (like Tesla) tried their hand at in housing this tech and will realize that the reward is simply not worth the risk or cost.
Love this research. Do you mind if I share this to my X community? Can give you credit.
Sure just link my Substack (: