Develop a hive mind chip to be embedded into every humans.
Robotics like those headless robotic dogs from the Boston guys? Those are freaking scary. 
Mine is more software. But purpose is general purpose
Software is called AI 
As long as google is wasting money on science projects it’s no good in my books. Anyone wants to buy ahead of earnings on Monday?
1 Like
That has not been my experience. I’ve worked with many great French engineers. I think they were from Ecole Poly something-or-other.
France was an early leader in telecommunications, wireless, modem technology for phones, etc.
1 Like
Lol dude. Way to go to make me feel good about my career choice effective september/october
Do you think waymo is gonna make money?
Not enough to move the needle for GOOG. What’s the business model?
1 Like
If Waymo destroys Uber, can add 100B
I think the business model is:
- make self driving car
- …
- profit!
Just because something is hard doesn’t make it valuable.
Running a ride sharing network is a big commitment. Self driving may be ready for freeway driving in 10 years but it won’t be able to navigate city streets for another 20 years I bet. That means ride sharing will still need a human driver.
So no google won’t run a ride sharing network.
Will it sell its system to car makers? That’s the model of Nvidia. But Nvidia already sells chips. And bundling its own software makes even fatter margin. It’s a perfect fit for Nvidia. Intel is going after the same model. Most car makers either partner with Nvidia or intel. Not sure about google.
And then there’s GM that’s going alone. People are underestimating GM I think. Of all the self driving companies I see GM’s cars most often in SF.
1 Like
Nvidia will never do self driving software. You are dreaming. Nvidia sucks at software and doesnt have the talent to make it work.
The German car companies (and technology providers) also have very advanced efforts in this space. Most are taking the approach of phasing into level 5 autonomy over many years. The existing auto companies understand this space very well and have been working on these problems for many years.
I’m not sure that is true. Nvidia is better than almost any other HW vendor at SW. Their deep learning success is largely due to the fact that they invested heavily (and early) in the SW work required to make most ML frameworks “just work” on Nvidia. Not a trivial task.
1 Like
Cuda and others are just that… Framework. Tensor flow is already displacing it. Most of nvidia run up was crypto
Nvidia doesnt do software. They are not even a comoetitor to googlr.
Nvidia may not be the one to build the high level driving software, but it’s building a rich foundational software layer on top of its hardware. That hardware includes lots of sensors that need to talk to Nvidia’s compute thru high speed link. To make an OS so to speak making everything just works is non trivial.
Strategy wise that’s the smart thing to do. Kind of like making the picks and shovels for everyone to build their self driving tech on top of Nvidia platform. Nvidia can become the intel and Microsoft, combined, of self driving cars.
1 Like
I agree Nvidia is not a competitor to google. Nvidia makes money selling silicon and systems, and google makes money selling ads.
tensorflow is not displacing cuda. They are at different layers of the stack. Nvidia did the hard SW engineering work of making sure tensorflow, mxnet, caffe, caffe2, etc is all accelerated seamlessly on Nvidia gpus.
There are tpus that is the goal. No dependency on nvidia.
That said, nvdia will fail at self driving anything. I am 95% sure.
I can see the hardware support, but i also know googlr is good at computing hardware. Customized hardware and customized realtime os, google can do.