He accused Tesla of "treating the public like guinea pigs."
Most vehicles available for sale today offer driver assistance features; even those with the most advanced aids require the driver to always monitor and be prepared to control the vehicle.
Investors, who are waiting anxiously for what is expected to be a dismal quarterly earnings report on Wednesday, will be considering whether Musk's ambitious vision will be translated into profitable products and services.
The new computer chip is positioned behind the glovebox. Most older Teslas can have it installed by a technician. The chips will work with radar, optical and ultrasonic sensors already installed in Tesla cars. Musk said any car built since October 2016 would be capable of full self driving once it has the new chip.
The chip provides "a new level of safety and autonomy ... without affecting our cost or range," said Tesla chip designer Peter Bannon. Bannon delivered a highly technical overview to investors gathered at Tesla's Palo Alto headquarters.
Tesla started developing the computer in 2016, the same year its relationship ended with self-drive chip supplier Mobileye. Since then, it's been using chips from Nvidia. But Musk said Monday that while Nvidia makes "great chips," company has to satisfy large numbers of customers and applications, including video gaming and cryptocurrency mining. The Tesla chip is designed exclusively for self-drive Teslas, Musk said. Tesla contracts out the manufacture of the chip.
Musk faces formidable challenges. Regulations governing autonomous cars remain a state-by-state patchwork with Congress still struggling to enact new federal legislation. Safety concerns are rising as some cars equipped with Tesla's Autopilot driver-assistance technology have crashed and drawn attention on social media and tech sites. The industry and government have not come up with the basic data and a common, transparent method of assessing the safety of driverless cars, so the general public is left to guess whether new technologies offer improvements over human drivers, and if so, by how much.
Tesla is also controversial in the driverless industry for its decision to forgo expensive lidar sensor technology. Lidar uses light beams to identify physical objects. Most autonomous vehicle developers are using a mix of lidar, radar and visual camera sensors. Musk believes that radar and optical sensors are sufficient by themselves.
"Anyone relying on lidar is doomed," Musk said Monday. "Lidar is lame. In cars, it's frigging stupid." He added that Tesla's competitors "are all going to dump lidar, is my prediction. Mark my words."
Tesla's robo-taxi business model is far from proven. App-based ride-share companies Uber and Lyft are have been operating driver fleets for years and now are selling stock to the public. If Tesla leapfrogged those competitors by getting out first with driverless cars, it theoretically could cut costs enough to gain market share. Musk said Tesla has an advantage in the data it is constantly picking up on the road via car sensors, which is wirelessly delivered to company's developers.