June 2026: Tesla’s Robotaxi Reality Check, Google’s AI Search Headaches, and EVs Conquering South Korea

The Robotaxi Illusion: Hype Versus Reality

It started with a 14-second video posted over the weekend — a Tesla Model Y cruising the streets of Dallas without anyone behind the wheel. Elon Musk re-shared it. Headlines followed. Tesla had, the narrative went, launched unsupervised robotaxi rides in two major Texas cities at once.

The reality, as crowdsourced tracker Robotaxi Tracker showed within hours, was far more modest. Dallas listed 31 square miles of service area; Houston, just 25. As of Monday morning, both cities were showing the service as "unavailable." By contrast, Tesla’s Austin operation — the same company, the same technology — had maintained a stable fleet of roughly 46 vehicles for months. The pattern is familiar: Tesla has a habit of announcing robotaxi milestones in the days immediately before quarterly earnings, sending the stock jumping, only for the services to quietly contract afterward.

The disconnect between Musk’s rhetoric and the lived experience of actual riders is widening. Two years ago, Musk predicted unsupervised robotaxis would be available to half the US population by the end of 2025. Today, in June 2026, the service remains a small experimental roll-out in a handful of Texas cities. Bloomberg’s analysis recently mapped that "growing chasm" between the bombastic predictions and the reality: a service that is slow, geographically limited, and inconvenient enough that paying comparison to Uber or Lyft still tilts heavily in favor of the incumbent apps for most trips.

Safety remains the other unignorable variable. Tesla reported 14 crashes involving robotaxis in the 9 months following launch — a figure the company redacts heavily in its federal filing, making severity difficult to assess. A viral video over the weekend appeared to show a Tesla robotaxi veer onto a freeway, requiring remote human intervention to find an off-ramp. The incident was brief, the outcome was fine, but it captured precisely the kind of edge case that autonomous systems must handle flawlessly — not "mostly."

Tesla is not alone in grappling with these growing pains. Waymo, Alphabet’s robotaxi unit, also recently launched in Dallas and Houston with similarly modest footprints. But Waymo has been transparent about starting small; Tesla’s pattern of declaring victory before the service is robust has left investors and regulators skeptical.

Autonomous Driving’s Land Rush: Who Is Actually Building Infrastructure

The robotaxi race is not just about software — it is about physical proving grounds, and those do not come cheap. Waymo just paid $220 million for Apple’s former 5,458-acre testing facility in Wittman, Arizona. Apple had acquired the site in 2021 for $125 million for Project Titan, its ill-fated attempt to build a self-driving car. Tim Cook pulled the plug in 2024, and the property has been sitting on the market ever since.

Waymo’s purchase is strategically significant: it gives the Alphabet-owned company a private, purpose-built track for edge-case testing — the kind of controlled environment that is essential for training autonomous systems on rare, dangerous scenarios before they enter public roads. For a company that is now operating or testing in a dozen American cities, a dedicated 5,500-acre facility is not a luxury; it is a competitive necessity.

The inflection point for robotaxis has always been infrastructure, and 2026 is the year that reality is settling in. Deployment at scale requires not just the cars but the maps, the regulatory frameworks, the maintenance depots, the remote supervision centers, and the public trust. Waymo’s purchase of a former Apple proving ground is proof that the most valuable real estate in AI right now is not a server farm — it is a stretch of empty desert where you can crash a car without hurting anyone.

The Electric Pivot: Tesla Dominates South Korea — A Market Historically Hostile to Imports

In May 2026, Tesla’s Model Y did something no foreign carmaker has ever accomplished in South Korea: it became the best-selling vehicle in the entire country. Not just the best EV — the best vehicle, full stop. With 8,762 units sold, it pushed the Kia Sorento to second place and the Hyundai Grandeur to third.

The significance of the milestone is easy to understate. South Korea has historically been one of the most protected and brand-loyal automotive markets on earth. Hyundai and Kia together command roughly 70 percent of domestic sales, built on decades of consumer trust and aggressive import tariffs. Foreign entrants — even premium European brands — have struggled to gain meaningful ground. BMW, the second-best-selling import brand in Korea last month, sold 6,555 units, well under three-quarters of Tesla’s total.

Tesla imported 10,866 cars into South Korea in May, making it the top import brand for the fourth consecutive month. The EV category represented 48.6 percent of all imported passenger car registrations. Tesla’s year-to-date growth in the country sits at 250.8 percent versus the same period last year, and it now holds a 30.8 percent share of the entire imported car segment.

The breakthrough appears driven by the updated Juniper refresh of the Model Y, which addressed earlier criticisms around range, interior quality, and ride refinement — all of which matter intensely in a market that compares Tesla directly against premium domestic alternatives. Korea’s EV tax incentives and an expanding Supercharger network have further tilted the economics in Tesla’s favor. What was once the hardest import market for an automaker to crack is now Tesla’s.

Charging Infrastructure Gets Unusually Creative

With EV adoption accelerating across Europe and North America, the bottleneck is shifting from vehicle availability to charger density — and Tesla is solving it with industrial design. In March 2026, the company introduced its "Folding Unit" V4 Supercharger: a factory-pre-assembled charging station built on a heavy-duty hinge system that folds flat for transport and unfolds on site. The first unit arrived in Europe in early June, with Tesla teasing a motorway rest-stop installation that is likely in Norway, its traditional launch market for new EV infrastructure.

The engineering is more sophisticated than it sounds. Each Folding Unit pairs a single V4 power cabinet with eight charging posts, delivering up to 500 kW per stall for passenger vehicles and up to 1.2 MW for the Tesla Semi. The design fits 33 percent more stalls per delivery truck, cuts installation time roughly in half, and reduces total deployment costs by more than 20 percent compared to traditional ground-mounted installations — a meaningful margin when you are building dozens of new stations per quarter.

Cable length was also redesigned from scratch. Earlier Superchargers were physically limited to Tesla vehicles; the new stations use longer cables that are immediately compatible with Ford, GM, Rivian, Hyundai, Stellantis, and other brands joining the North American Charging Standard. The pivot from proprietary to universal is quietly one of the most consequential infrastructure decisions Tesla has made in the past two years. It transforms the Supercharger network from a competitive moat into a public good — and massively expands the addressable market for the hardware.

AI Search Is Creating a Legal Problem No One Expected

For as long as search has existed, courts have treated it as a discovery tool — a way to find and link to content created by someone else. That legal framing may be changing. A German court recently issued a preliminary ruling finding that Google is legally responsible for content generated by its AI Overviews feature, reasoning that the summaries constitute "independent, new, and substantive statements" rather than simple pointers to third-party websites.

The logic is straightforward and, in its implications, radical. When a conventional search engine returns results, it functions as an index; the content lives on the destination site, and the engine is generally shielded from liability for that content. AI Overviews are different: they synthesize, paraphrase, and sometimes invent information drawn from multiple sources into a single original response. The court’s reasoning is that because Google creates that response, only Google can verify its accuracy — and therefore only Google bears responsibility when it is wrong.

The ruling sits at the intersection of two accelerating trends: the deep integration of generative AI into consumer products, and the growing body of regulatory scrutiny on AI outputs. The European Union’s AI Act is already imposing transparency obligations on high-risk AI systems; a German court extending publisher-level liability to a search product is exactly the kind of precedent that will force every major tech company to rethink how it deploys generative features in consumer-facing products.

The Data Center Backlash Is Getting Organized

What began as scattered community opposition to new AI data centers is hardening into something resembling a movement. In early June, the Seattle City Council voted to enact a one-year emergency moratorium on new large-scale data center construction, responding to months of protest over water consumption, electricity costs, noise, and climate impact. Amazon — Seattle’s largest employer, the city’s most powerful corporate resident, and itself a massive data center operator — was effectively lobbied by its own employees.

Liesl Wigand, a senior software engineer at Amazon and member of Amazon Employees for Climate Justice, testified at a city council hearing: "In my job, I see the consequences of the all-costs-justified AI buildout. The biggest issue is a belief that AI should be how we solve everything, while ignoring the resources that it costs." More than 1,000 Amazon employees signed an open letter last year accusing the company of "casting aside its climate goals to build AI," and they have spent the months since pushing for 100 percent local renewable energy to power all new data center capacity.

The opposition is not unique to Seattle. Data center moratoria have been proposed or enacted across Utah, Virginia, the Netherlands, and Singapore, typically citing the same three concerns: electricity grids that were not built for hyperscale AI loads, water consumption measured in the millions of gallons per facility per day, and the physical footprint — power substations, cooling infrastructure, and the roads required to service them — that competes directly with housing and community space.

The technical optimist’s response is that these are first-world infrastructure problems that money can solve. That is true, but "money can solve it" is not the same as "it is being solved." Historically, permitting for large data centers takes two to three years; if cities begin refusing permits outright, that timeline stretches further. The AI buildout is running directly into the physical and political limits of the built environment.

The New Electric Era: Mitsubishi and a Reshaped Competitive Landscape

Tesla’s dominance masks a broader EV renaissance that is no longer centered on California. Mitsubishi is preparing to launch the 2027 Eclipse Sportback EV in North America later this fall — the company’s first dedicated electric passenger vehicle in the region in several years. Built on the next-generation Nissan Leaf platform, it is expected to carry a 75 kWh battery giving roughly 303 miles of range. Pricing has not been announced, but the underlying architecture suggests Mitsubishi is targeting the mainstream compact EV segment, not the premium end that Tesla and BMW have staked out.

The Eclipse’s arrival is part of a larger pattern: legacy automakers that appeared slow to electrify are now releasing credible EVs with competitive range, refined interiors, and aggressive pricing. The Mitsubishi-Nissan alliance also signals that smaller OEMs are going to compete through platform sharing and cost-efficient engineering rather than ground-up EV architectures — a sensible response to the enormous capital requirements of a fully bespoke electric platform.

The Week That Was: The Signals Behind the Noise

Reading the week’s tech headlines together, a picture emerges that is more coherent than any single story suggests. The AI boom is creating genuine economic value and genuine infrastructure stress simultaneously. Courts are beginning to draw legal lines around AI outputs for the first time. The employees who build these systems are starting to demand accountability for their environmental and social cost. The EV transition is accelerating in markets that were long written off, while autonomous vehicle rollouts remain slower and more fragile than their promoters admit.

What makes this moment unusual is the speed at which sectors are colliding: AI infrastructure depends on electricity grids that EV adoption is also draining; autonomous vehicle software draws on the same sensor and edge-compute advances that are powering consumer AI; and the engineers working on all of it share a growing concern about the resource intensity of the buildout. The next decade of technology will likely be defined by how well these tensions get resolved — and this week gave us a front-row seat to every one of them.