GM Built It First. Tesla Gets the Credit.

My recent "Third Bay" memo traced how GM understood the EV architecture insight before Tesla commercialized it.

A reader left a comment on LinkedIn naming this precisely as the Innovator's Dilemma.

Clayton Christensen's thesis holds that incumbents don't fail because they lack talent or technology. They fail because leadership optimizes for the business they have, not the business that's coming.

The robotaxi sector may be offering early evidence of that thesis playing out again. We are still in the early stages of what a robotaxi business can become. But the signals from incumbent automakers are worth examining now.

A recent DOAC podcast episode (with Daniel Priestley, March 2026) described Tesla's Cybercab as the world's first vehicle produced without a steering wheel or pedals.

It's a compelling story. It's also wrong.

GM built that vehicle first.

The Cruise Origin was unveiled in January 2020. No steering wheel. No pedals. Campfire seating for six. Engineered to last one million miles. Purpose-built for autonomous rideshare from the ground up. Production began at Factory Zero in Detroit, the same facility that builds the GMC Hummer EV.

GM's leadership called it "the first scalable vehicle ever designed specifically for autonomous rides."

Then they killed it.

In October 2023, a Cruise robotaxi dragged a pedestrian in San Francisco. The pedestrian had first been struck by a human-driven vehicle. Cruise lost its California permits and operations halted nationwide.

By July 2024, GM took a $583 million write-down and cancelled the Origin.

By December 2024, the retreat was complete. GM formally exited the robotaxi business, merged Cruise into its internal teams, and pivoted to personal vehicle autonomy.

Eight months after leaving Cruise, Kyle Vogt, the entrepreneur who founded Cruise in 2013 and sold it to GM for $581 million in 2016, posted on X: GM "repeatedly finds themselves with a 5-10 year head start, but then fumbles the ball, shuts things down, and loses the lead."

The critique is pointed. It also invites a fair counterpoint. Vogt resigned under pressure following the safety scandal that unfolded on his watch. And it was Vogt who chose to sell Cruise to GM. If he had doubts about GM's ability to hold a disruptive position, that was a decision he made with full information.

His quote also raises a broader observation worth stating plainly. Some companies with early head-starts are no longer in the race.

GM. Ford. VW. Hyundai. Apple.

All had early positions. None sustained them. That is not a coincidence. It is the pattern.

The technology is not gone. That distinction matters.

GM is actively building autonomous capability into its personal vehicle lineup. Super Cruise is currently a Level 2 system, available on many models and covering 750,00 miles of North American roads. GM has targeted a Level 3 version, which would allow hands-off, eyes-off driving in 2028. That roadmap is not yet delivered, but it is real.

GM also hired former Tesla Autopilot chief Sterling Anderson to lead the next phase of personal AV development. That is a serious signal about intent.

The engineering capability survives. The institutional knowledge survives. This is not a story about a company that couldn't do the work.

What GM walked away from is the robotaxi business model. And that is where the strategic debate lives.

Before treating this as a GM-specific failure, the record deserves to be read in full.

Ford and VW jointly backed Argo AI starting in 2017, eventually raising over $3.6 billion and reaching a peak valuation of $12.4 billion. Ford shut it down in October 2022, took a $2.7 billion write-down, and retreated to Level 2 driver assistance for personal vehicles. Ford CEO Jim Farley's exit line: "profitable, fully autonomous vehicles at scale are a long way off and we won't necessarily have to create that technology ourselves."

Hyundai and Aptiv backed Motional. Aptiv pulled financing in early 2024. And the Las Vegas robotaxi service was suspended shortly after.

Apple spent an estimated $1 billion per year on autonomous vehicle development for nearly a decade and shut the program down in 2024.

The pattern is not just a GM story. The pattern extends well beyond GM. Every major automaker that entered the robotaxi race eventually walked away and redirected toward incremental ADAS on vehicles they already sell.

Part of this story is not about strategic courage. It is about capital structure.

Full autonomous capability is extraordinarily expensive to develop. The sensor arrays, compute stacks, software development, safety validation, regulatory navigation, and the fleet vehicles required to run a service at scale represent billions in sustained investment before a single profitable mile is driven. Waymo has spent an estimated $8 billion to reach its current state.

Alphabet's market cap provides a balance sheet that can absorb that burn for years. Amazon's logistics empire gives Zoox a strategic rationale extending well beyond ride-hailing economics. Tesla's stock valuation has provided access to capital that few traditional manufacturers can match.

Legacy OEMs operate on automotive margins, answer to institutional shareholders on quarterly cycles, and compete in a business where a single product misstep costs billions. The pressure to redirect capital from a long-horizon speculative business back to the core franchise is not purely a failure of leadership vision. It is also a structural reality.

This may be the most precise version of the Christensen argument: incumbents are not always wrong to retreat. The organizations funding the disruption simply have different financial architectures. They can afford to be patient in ways that a Ford or a GM cannot.

Super Cruise is a Level 2 system today. The driver stays in the loop. The more autonomous version is targeted for 2028.

That is a fundamentally different business from what Waymo, Tesla, and Zoox are building.

Robotaxis represent a new revenue stream on top of vehicle sales: miles-as-a-service, fleet economics, urban mobility infrastructure.

Waymo completed more than 4 million paid rides in 2024 alone. ARK Invest has projected the global robotaxi TAM at $4 trillion in net revenue by 2030. That number may prove optimistic. The direction of the opportunity is not in dispute.

The U.S. is not the whole story. China's Baidu Apollo Go surpassed 11 million total rides by mid-2025. Pony.ai and WeRide are scaling through Uber partnerships across the Middle East and Europe. The robotaxi race is global and accelerating.

A parallel opportunity is forming in autonomous over-the-road trucking. Aurora Innovation launched the first commercial driverless trucking service on public roads in Texas in April 2025. Kodiak Robotics, Waymo Via, and Torc Robotics are active in the same space. Fixed routes, defined customers, acute driver shortage. The near-term unit economics in freight may be more practical than urban robotaxi.

In the U.S., three operators remain committed to consumer robotaxi at scale: Waymo, Tesla, and Amazon via Zoox.

Globally, a Chinese tier led by Baidu, Pony.ai, and WeRide is already operating at significant volume and expanding westward.

Note what the U.S. survivors share. None of them manage a legacy ICE franchise.

Tesla is an automaker. It earns most of its revenue selling vehicles to consumers. But it carries none of the defensive reflexes that tend to kill disruptive bets inside legacy OEMs. No ICE platform to cannibalize. No franchise dealer network to protect. No century of manufacturing identity to defend.

Sterling Anderson led the Tesla autopilot organization and co-founded Aurora Innovation in 2017, one of the leading autonomous trucking companies now operating driverless commercial routes in Texas. GM hired Anderson in 2025 to lead its next phase of personal AV development.

The engineer building the disruptive future is now working inside the incumbent trying to catch up. That is either a very good sign for GM, or a very good illustration of how incumbents acquire talent after ceding the lead.

Waymo has no vehicles to sell. Amazon has a logistics empire to feed. Neither faces the conflict of cannibalizing a product line that still generates most of their revenue.

The Innovator's Dilemma is not about old versus new. It is about what an organization has to lose by succeeding at something disruptive.

Legacy OEMs had the most to lose. They acted accordingly.

This is not a story about incompetence. The engineers at Cruise, Argo AI, and Motional were exceptional. They built real things that operated on public roads.

What the robotaxi retreat may reveal is a leadership culture across the legacy auto industry that struggles to hold a disruptive position when the core business is under pressure.

One incident, one bad quarter, one investor call becomes the pretext. Regulatory complexity becomes the rationale. Capital efficiency becomes the cover.

The robotaxi race is not over. The TAM is still forming. The technology is still maturing. It is entirely possible that the pivot to personal autonomy proves to be the right long-term call for GM and Ford.

But the companies with early head starts are no longer in the race.

That is worth watching.

If you have a perspective or disagreement, reply directly. I read every response.