GM's AI writes 90% of autonomy code, simulating 100 years of driving daily

GM now generates 90% of its self-driving car code using AI, stress-testing the system in a simulation that runs the equivalent of 100 years of human driving every single day.

General Motors Co. said artificial intelligence now writes 90% of the code for its autonomous driving team, a shift that could cut development costs and accelerate the timeline for hands-free driving technology across its vehicle lineup.

"Today, nearly 90% of the code written by our autonomy team is generated by AI," Chief Executive Officer Mary Barra said on the company's Q1 2026 earnings call, calling it proof of "how seriously we're embracing AI across the enterprise."

The AI-written code powers the next-generation eyes-off, hands-off Super Cruise system, set to launch on the Cadillac Escalade IQ in 2028. GM is validating the software in a digital environment that simulates roughly 100 years of human driving daily, Barra said, with supervised on-road testing already underway in California and Michigan. The current Super Cruise system has logged 1 billion hands-free miles and is on pace to exceed 850,000 subscribers by year-end, with renewal rates between 30% and 40%.

The 2028 Escalade IQ launch will be the verdict event for GM's AI-first approach to autonomy. A single high-profile failure of machine-generated safety code would carry significant reputational risk, but success could give GM a cost and speed advantage over Tesla Inc. and Waymo in the race to deploy Level 3 autonomous driving at scale.

The Financial Engine Behind the AI Bet

GM has the cash flow to fund aggressive AI tooling investment. Q1 adjusted earnings per share came in at $3.70 versus the $2.64 consensus estimate, a 40% beat — the fourth consecutive quarter GM has topped Wall Street EPS forecasts. EBIT-adjusted reached $4.25 billion, up 22% year over year, with margin expanding 2 percentage points to 10%. Management raised full-year adjusted EPS guidance to a range of $11.50 to $13.50.

Digital services provide another revenue stream to support the autonomy push. OnStar revenue topped $750 million in Q1, up more than 20% year over year, with calendar-year revenue expected to reach $3.1 billion and deferred revenue approaching $7.5 billion. Chief Financial Officer Paul Jacobson said Super Cruise attachment rates after the free trial sit near 40%, calling himself "very optimistic" about the conversion math.

Regulatory Scrutiny and Competitive Risk

Barra's announcement comes as regulators examine rival approaches to autonomous driving. Two U.S. senators are urging the National Highway Traffic Safety Administration to review Tesla's self-published Full Self-Driving crash statistics, while European regulators have accused Tesla of publishing misleading data on FSD safety. Tesla's robotaxi fleet in Texas stands at 69 vehicles, compared with Waymo's 620.

GM is pitching its AI-written, simulation-validated approach as the disciplined alternative — a contrast to what it frames as Tesla's less rigorous methodology. But the strategy carries its own risks: the 90% figure applies specifically to GM's autonomy team's pre-launch code, not to production software in customer vehicles. The company has not disclosed how it plans to extend the same AI-driven development pipeline to its internal combustion engine vehicles, many of which were not designed with the computing power required for Level 3 autonomy.

The market has rewarded the pitch so far. GM shares are up 66% over the past year and trade at $80.04, against an analyst consensus target of $94.81 and a forward price-to-earnings ratio of 7. Citi recently raised its price target to $108 after the Q1 beat.

For investors, the question is whether GM's AI-first autonomy strategy can deliver a return on the billions invested. The company trades at a steep discount to Tesla's forward P/E of roughly 60, reflecting the market's skepticism about GM's ability to execute on its technology vision. If the 2028 Escalade IQ launch proves the AI-written code is safe and reliable, that valuation gap could narrow. If it fails, the reputational damage could set GM's autonomy program back years.

This article is for informational purposes only and does not constitute investment advice.