A technology forecaster who predicted robotaxi deployment in the mid-2020s turned out to be far more accurate than industry leaders making similar claims. Brad Templeton, an early Waymo employee, made a prediction back in 2013 that pilots would hit the road around 2020, followed by rapid scaling across cities in the middle of the decade. That timeline has largely come to pass, with Waymo, Baidu Apollo, and Wayve all announcing plans to deploy robotaxis in London in 2026, though regulatory delays have pushed the timeline to early fall.

The contrast between accurate forecasting and repeated misses raises a crucial question: what separates someone who gets it right from industry leaders who consistently overpromise? Templeton's track record offers insights into how revolutionary technology timelines actually work, and why even the smartest people in the field can get them spectacularly wrong.

What Made the 2013 Prediction So Accurate?

Templeton's forecast wasn't pulled from thin air. He had insider knowledge from his work at Waymo, where he helped create the robotaxi strategy the company still follows today. In 2007, he wrote a comprehensive series of articles about the future of self-driving cars that caught the attention of Larry Page, Google's co-founder and the true architect behind Waymo. Page hired Templeton to help shape the company's direction, giving him a front-row seat to the technology's actual progress.

Beyond his Waymo experience, Templeton had decades of software engineering and management expertise. Yet even with those advantages, he acknowledges that precise prediction was nearly impossible in 2013. The deep learning revolution was in its infancy; AlexNet, the first major breakthrough in deep neural networks, had only arrived a year earlier. Transformers, the technology that would eventually power large language models (LLMs) and reshape artificial intelligence, wouldn't emerge until 2017 and 2018.

What Templeton had that others lacked was a realistic understanding of the gap between impressive early demonstrations and production-ready robotaxis. Many teams could show impressive driving on selected routes with safety drivers ready to intervene. Chauffeur, the Google project that became Waymo, had racked up 100,000 miles in self-driving mode and 1,000 distinct miles of road coverage within a year of starting. But Templeton understood that this early progress masked a much longer journey to full autonomy.

Why Do Industry Leaders Keep Missing Their Own Timelines?

Elon Musk has made robotaxi predictions for nearly as long as Templeton has been forecasting, yet his timeline estimates have repeatedly proven too optimistic. This pattern reveals something important about how visionary leaders approach technology forecasting. Musk, who should arguably know better given his direct involvement in autonomous driving development, has consistently underestimated the complexity of moving from impressive demonstrations to reliable, scalable systems.

The problem isn't unique to Musk. In the early 2010s, traditional automakers nearly universally predicted 2020 as the date for real commercial robotaxi production. Chris Urmson, one of the world's leading autonomous driving experts, was asked in 2007 when robotaxis would arrive. His answer reflected General Motors' company line, though Urmson himself believed something would come sooner. Later, Urmson famously predicted his son wouldn't need a driver's license after turning 16 in 2019. While Waymo did deploy driverless cars in 2019, we're still not at the point in 2026 where teenagers can skip driver's licenses, though that may happen in select cities soon.

What these repeated misses reveal is a fundamental challenge in forecasting revolutionary technology. Templeton points to Amara's Law, a principle taught at Singularity University, which he helped build: we tend to overestimate progress in the short term while underestimating it in the long term. Visionary leaders often fall into the short-term overestimation trap, announcing timelines based on technical breakthroughs without accounting for the regulatory, safety, and scaling challenges that follow.

How to Think About Robotaxi Timelines and Forecasts

• Distinguish between impressive demos and production readiness: A self-driving car that performs well on selected routes with a safety driver ready to intervene is fundamentally different from a robotaxi operating unsupervised in dense urban traffic. The gap between these two states is far larger than most early predictions accounted for.
• Account for regulatory and scaling challenges: Technical capability is only one piece of the puzzle. Regulators must approve operations, cities must adapt infrastructure, and companies must prove safety at scale. Waymo's London deployment, for example, was delayed from mid-2026 to early fall due to regulatory readiness, not technical issues.
• Recognize survivorship bias in forecasting: When many people make different predictions, some will inevitably look prescient in hindsight. Templeton acknowledges that luck played a role in his accuracy, and that people rarely publicize their failed predictions, creating an illusion that forecasting is easier than it actually is.
• Understand the role of insider knowledge: Templeton's accuracy benefited from his direct involvement in building Waymo's strategy and his understanding of the actual technical challenges. General observers and even some industry leaders lack this granular knowledge of what's genuinely difficult versus what's merely complex.

Where Are Robotaxis Actually Heading?

The current state of robotaxi deployment suggests Templeton's original forecast is holding up reasonably well. Three major players are preparing London deployments in 2026, with regulatory approval expected in early fall. Wayve, a London-based company using large language models (LLMs) for autonomous driving, has yet to deploy an unsupervised robotaxi anywhere but hopes London will be its first major market. Baidu Apollo is also targeting European deployment, though recent technical challenges may interfere with their timeline.

The fact that multiple companies are converging on similar timelines and locations suggests the technology has genuinely matured beyond the hype phase. Unlike the 2010s, when predictions were often based on theoretical progress, today's announcements come from companies actively testing in real cities with real passengers. Waymo, for instance, has been operating with safety drivers in London and is now moving toward unsupervised deployment.

Yet the regulatory delays in London, pushing deployment from mid-2026 to early fall, underscore Templeton's broader point: technology timelines are set by when companies convince regulators they've made systems safe enough, not by when engineers finish the code. This regulatory reality is often invisible to outside observers but represents one of the largest sources of timeline uncertainty in autonomous vehicle deployment.

"I had insider's knowledge of the dawn of the industry, and had helped them create the robotaxi strategy they still follow to this day," explained Brad Templeton, an early Waymo employee and technology forecaster.

Brad Templeton, Technology Forecaster and Former Waymo Employee

The broader lesson from Templeton's accurate prediction is that good forecasting requires balancing optimism about technological progress with realism about the many non-technical barriers to deployment. Visionary leaders like Musk excel at imagining what's possible but often underestimate the time required to move from possible to practical. Templeton's 2013 forecast succeeded because it acknowledged both the genuine progress being made and the substantial work still required.