General Motors has put 200 vehicles equipped with autonomous driving technology on public roads. Not test tracks. Not closed courses. Public streets where people walk dogs, run red lights, and double-park without warning.
The milestone, confirmed by GM in recent days, marks a significant escalation in the automaker’s push to become a credible player in autonomous transportation — a field where it has already spent billions, absorbed a spectacular failure, and now appears to be mounting a comeback that few in the industry expected to happen this quickly.
According to MSN, GM’s autonomous vehicle subsidiary Cruise has reached the 200-vehicle deployment figure across its active operating areas, a number that represents both the company’s ambitions and the distance it still needs to cover to catch its chief rival, Alphabet’s Waymo. The vehicles are operating in select cities, gathering data, refining their software, and — critically — doing so without a human safety driver behind the wheel in certain zones.
That last detail matters enormously.
The entire autonomous vehicle industry was shaken in October 2023 when a Cruise robotaxi in San Francisco struck a pedestrian who had already been hit by another car, then dragged the person approximately 20 feet. The incident triggered a cascade of consequences. California’s Department of Motor Vehicles revoked Cruise’s permit to operate driverless vehicles in the state. The National Highway Traffic Safety Administration opened an investigation. GM’s then-CEO Mary Barra ordered a sweeping internal review. And Cruise’s CEO Kyle Vogt resigned.
For months, the operation went dark. Cruise pulled its driverless vehicles off the road nationwide — not just in California — and the company hemorrhaged talent. GM had already poured more than $10 billion into Cruise since acquiring it in 2016. The question hanging over Detroit was stark: Had all of that investment been incinerated in a single nighttime accident on a San Francisco street?
The answer, apparently, is no. But the path back has been anything but simple.
GM restructured Cruise’s leadership, installed new safety protocols, and shifted the subsidiary’s strategy. Rather than operating an independent robotaxi fleet — the model Cruise had been pursuing in direct competition with Waymo — GM began integrating the autonomous technology more tightly with its own vehicle lineup. The company signaled that it would focus on deploying the technology in personal vehicles sold to consumers, not just in ride-hail fleets. It was a philosophical pivot with massive financial implications.
The 200 vehicles now on the road reflect this hybrid approach. Some are operating in supervised autonomous mode, meaning they can drive themselves but have a human onboard ready to intervene. Others are running fully driverless in geofenced areas where regulators have granted permission. GM has been deliberately quiet about the exact breakdown, a communications strategy that contrasts sharply with the company’s earlier tendency to make bold public promises about autonomous timelines — promises it repeatedly failed to meet.
And yet 200 vehicles is still a modest number. Waymo, which has been operating its robotaxi service commercially in Phoenix, San Francisco, Los Angeles, and Austin, now completes over 150,000 paid rides per week, according to recent disclosures by Alphabet. The company has expanded aggressively into new markets, including Atlanta and Miami, and has announced plans for additional cities. Its fleet numbers in the hundreds in each major market. By any measure, Waymo has established a commanding lead in commercial autonomous ride-hailing.
So what exactly is GM’s play here?
The theory, articulated by GM executives in recent earnings calls and industry presentations, is that the real money in autonomy won’t come from operating taxi fleets — a capital-intensive, low-margin business — but from selling autonomous-capable vehicles directly to consumers. If GM can embed Level 4 autonomous capability into its Ultium-platform electric vehicles, it could charge a premium that transforms the economics of every car it sells. Think of it as the Tesla Full Self-Driving subscription model, but with hardware and software that actually delivers on the promise of hands-off, eyes-off driving in defined conditions.
This is a fundamentally different bet than the one Waymo is making. Waymo wants to own the ride. GM wants to own the car.
Both approaches have serious risks. Waymo’s model requires enormous capital expenditure on fleet vehicles, maintenance, mapping, and operations in each new city. GM’s model requires convincing regulators — and consumers — that a car sold to an individual can safely operate without human oversight in certain scenarios. No automaker has yet achieved this at scale. Not Tesla, despite Elon Musk’s years of projections. Not Mercedes-Benz, which has received limited Level 3 approval in Germany and Nevada but only at speeds below 40 miles per hour. Not anyone.
The regulatory picture in the United States remains fragmented and uncertain. There is no federal framework for autonomous vehicle deployment. States set their own rules, and those rules vary wildly. California has among the strictest oversight regimes. Arizona has been far more permissive, which is one reason Waymo chose Phoenix as its first commercial market. Texas has been welcoming. New York has been essentially closed to autonomous testing.
GM’s 200-vehicle deployment is spread across jurisdictions where the company has secured the necessary permits, though it hasn’t disclosed the full geographic breakdown. What’s clear is that the company is being more cautious this time — a posture driven as much by legal exposure as by genuine safety concerns. The 2023 San Francisco incident generated multiple lawsuits and a federal investigation that, while largely resolved, left deep institutional scars.
The financial stakes for GM are staggering. The company spent $1.9 billion on Cruise operations in 2023 alone, a figure that drew sharp criticism from investors and analysts who questioned whether the returns would ever materialize. In response, GM cut Cruise’s burn rate significantly in 2024, reducing headcount and narrowing the scope of operations. The 2025 budget for autonomous development hasn’t been publicly detailed, but executives have indicated it will be lower than the 2023 peak while still representing a substantial commitment.
Wall Street has been skeptical. GM’s stock has traded at a persistent discount to its intrinsic value, and many analysts attribute that gap partly to uncertainty about the company’s autonomous spending. Morgan Stanley analyst Adam Jonas has repeatedly flagged the Cruise investment as a source of valuation risk, while also acknowledging that success in autonomy could dramatically re-rate the stock. It’s the classic binary outcome problem: the technology either works and becomes enormously valuable, or it doesn’t and the investment is largely written off.
There are technical reasons for cautious optimism. The sensor hardware available today — including lidar, radar, and high-resolution cameras — is significantly better and cheaper than what existed even three years ago. Machine learning models for perception and prediction have improved markedly, aided by the sheer volume of real-world driving data that companies like Cruise and Waymo have accumulated. The computational power available in vehicle-grade chips, particularly Nvidia’s Orin and upcoming Thor platforms, has increased by orders of magnitude.
But hardware and software improvements don’t automatically translate into safe, reliable autonomous driving in all conditions. The so-called “long tail” of rare and unusual driving scenarios — a mattress falling off a truck, a child chasing a ball into the street, a construction zone with contradictory signage — remains the fundamental engineering challenge. Every autonomous vehicle company in the world is grappling with this problem. None has fully solved it.
GM’s approach to the long tail has evolved. Cruise originally relied heavily on simulation to train its vehicles, running millions of virtual miles to expose the software to rare scenarios. That approach remains central, but the company has supplemented it with more structured real-world testing and a more conservative operational design domain — industry jargon for the specific conditions under which the vehicle is allowed to drive itself. Narrower domains mean fewer edge cases, which means fewer opportunities for catastrophic failure.
The 200-vehicle deployment also serves a critical data-collection function. Every mile driven generates sensor data that feeds back into the training pipeline, improving the models that control vehicle behavior. This creates a virtuous cycle: more vehicles on the road means more data, which means better software, which means safer operation, which means regulatory approval for more vehicles. Waymo has been riding this flywheel for years. GM is trying to spin it up.
Competition isn’t limited to Waymo. Amazon’s Zoox has been testing its purpose-built autonomous vehicle in several cities and recently began offering employee rides in its robotaxi. Tesla continues to expand its supervised Full Self-Driving system and has announced plans for an unsupervised robotaxi service, though the timeline remains unclear. In China, Baidu’s Apollo Go service and Pony.ai are both operating commercial robotaxi services at significant scale, with regulatory support from municipal and national authorities that exceeds anything available in the U.S.
The Chinese competition is particularly relevant for GM, which has significant operations and sales in China. If Chinese autonomous vehicle companies achieve widespread deployment before their American counterparts, it could create pressure on U.S. regulators to accelerate approvals — or, conversely, to impose restrictions on Chinese technology operating in the U.S. The geopolitical dimension of autonomous driving is becoming impossible to ignore.
For GM specifically, the 200-vehicle milestone is best understood not as an achievement but as a proof point. The company is demonstrating to regulators, investors, and its own board that it can deploy autonomous vehicles safely and at increasing scale after a period of crisis. Whether that demonstration leads to a commercially viable product — one that consumers will actually buy and trust — is a question that won’t be answered for years.
But GM doesn’t have unlimited time. The automotive industry is undergoing simultaneous transitions in electrification, software-defined vehicles, and autonomous capability. Companies that fall behind on any of these vectors risk losing market position permanently. Ford effectively exited the autonomous race by shutting down its Argo AI joint venture with Volkswagen in 2022, choosing to focus on driver-assistance features rather than full autonomy. Stellantis has been largely absent from the conversation. Among traditional automakers, GM stands essentially alone in pursuing Level 4 autonomy at scale.
That’s either visionary or reckless. Possibly both.
The next 12 to 18 months will be telling. GM has indicated it plans to expand the autonomous fleet further and begin integrating the technology into vehicles that will be available to retail customers. The company’s Ultra Cruise system — a highway-focused advanced driver-assistance feature — is already shipping in certain Cadillac models and is expected to gain capability through over-the-air updates. The bridge from Ultra Cruise to full autonomy is long, but it’s the path GM has charted.
Two hundred vehicles on public roads. A $10 billion-plus investment. A catastrophic setback that nearly killed the program. And a competitor in Waymo that is pulling further ahead every quarter. GM’s autonomous bet is one of the most consequential wagers in the modern auto industry. The cars are driving themselves now. The question is whether anyone — investors, regulators, consumers — will come along for the ride.
WebProNews is an iEntry Publication