From Solar Tiles to Starship: The Decade Musk Failed to Deliver Puts Semi At Risk

Sign up for daily news updates from CleanTechnica on email. Or follow us on Google News!

Tesla says the Semi is entering high-volume production in 2025, nearly eight years after its 2017 unveiling. At the time, Elon Musk promised a revolution in freight: 500 miles of electric range, faster acceleration than a diesel, and a cost-per-mile that would make it a no-brainer for fleet managers. It would be an incredibly useful and important product, unlike the Boring Company’s tunnels to nowhere, the Cybertruck that cuts carrots in half with its doors, a Roadster that could fly and a Cybercab that would only increase congestion.

Semi production was supposed to begin in 2019. It didn’t. The first units weren’t delivered until December 2022, and by early 2025, fewer than 100 are confirmed to be in customer hands. The 50,000 units per year target, floated as recently as late 2022, has now shifted to 2026. The Nevada Gigafactory expansion is still underway. And with it comes a larger question: has Musk lost the ability to deliver anything except controversy and hype?

The Boring Company began as a solution to traffic, albeit a brain dead one that ignored subsurface tunneling risks and induced traffic demand. It proposed autonomous electric pods moving at high speed through narrow tunnels. That vision, announced in 2016, included top speeds of 150 mph and a goal of transporting up to 4,000 passengers per hour per tunnel. By 2021, internal planning documents suggested up to 700 miles of tunnels across multiple U.S. cities. Since then, projects in Chicago, Los Angeles, Washington DC, San Jose, Fort Lauderdale, Ontario (CA), and San Antonio have been canceled or indefinitely paused. In some cases, like San Antonio, city officials say Tesla simply stopped responding to inquiries after initial proposals were accepted. In 2021, the company removed references to nearly all of its non-Vegas projects from its website, without public explanation.

The one exception is Las Vegas, where the Boring Company has built a small network of tunnels under the Convention Center and a handful of nearby casinos. The system was originally described as fully autonomous, high-speed, and high-throughput. In practice, it consists of manually driven Tesla vehicles operating at speeds well under 40 mph. The tunnels have a single lane in each direction, limiting capacity and complicating emergency response scenarios. Throughput, even during major conventions, has not exceeded 1,500 people per hour—well below early projections and vastly below the 50,000-70,000 passengers per hour of subways. Even in Las Vegas, where local regulators and developers remain friendly to the company, the expansion beyond the city core remains incomplete.

Despite raising hundreds of millions in funding and receiving local subsidies in some jurisdictions, the Boring Company has generated minimal direct revenue. Most of its capital has gone toward tunnel boring machine development and the Vegas Loop buildout. Outside of Las Vegas, it has yet to complete a public infrastructure project or enter revenue-generating operation. The disconnect between the company’s stated goals and its demonstrated output is growing more difficult to ignore.

SpaceX’s Starship was supposed to be operational by 2020. Elon Musk first claimed it would reach orbit within six months in 2019. The first orbital test flight didn’t occur until 2023, and it ended in mid-air disassembly. A second flight, in late 2023, also failed to complete its objectives. As of 2025, Starship has not reached orbit. It has not delivered a payload. It remains in the testing phase, and while the launches have provided spectacular visuals visible from hundreds of miles away, the technical results have been disappointing. Repeated structural failures, guidance issues, and engine anomalies have delayed progress.

Musk initially claimed that Starship would lift over 100 metric tons to low Earth orbit and do so at a cost of just $10 per kilogram. Today, those figures appear increasingly unrealistic. Internal reports suggest the current hardware under delivers by more than 50% on payload mass. Meanwhile, the costs of the program—including the development of Super Heavy boosters, orbital launch mounts, ground systems, and the repeated loss of test vehicles—are mounting. When adjusted for inflation, the cost per launch is likely comparable to or even above historic heavy-lift systems like the Saturn V or early SLS flights. Starship has not yet demonstrated any form of reusability.

No commercial payloads have flown. No national security contracts have been fulfilled. The business case for Starship relies on dramatic cost and volume breakthroughs that, five years after its originally promised debut, have not materialized.

In the meantime, NASA has delayed its Artemis III lunar mission, which depends on Starship to deliver astronauts to the surface. That contract, initially awarded in 2021, now carries the risk of extended schedule slippage, forcing NASA to reevaluate backup options. The idea of point-to-point Earth transit using Starship, once promoted as a near-term goal, has disappeared from official communication. SpaceX has raised billions in funding to support Starship development, but revenue from the program is essentially nonexistent. Without successful payload delivery or a viable commercial service, Starship remains a cost center with unproven scalability.

Tesla’s solar tile roof was expected to outpace its car business. In 2016, Musk claimed Tesla Energy could exceed vehicle revenue. He predicted one million solar roofs installed within a few years. As of early 2025, the number is likely below 25,000. Tesla no longer reports solar roof figures separately from its broader solar business. Costs remain high. Installation requires a full roof replacement. Competing systems like GAF Energy’s Timberline Solar have outpaced it in deployment and ease of installation.

The reality is that building-integrated photovoltaics, especially when embedded in roofing materials, present serious challenges. Roofs are complex structures that must withstand extreme temperature changes, precipitation, wind uplift, and mechanical stress over decades. Integrating PV cells into individual tiles introduces thousands of potential failure points, complicates flashing and waterproofing, and adds weight and fragility to what is already a finely tuned building envelope. Unlike conventional solar panels, which are mounted on top of an existing roof and can be installed in a matter of days, solar tile installations are labor-intensive and require trained crews familiar with roofing, electrical systems, and Tesla-specific components. Tesla has repeatedly raised prices and revised installation protocols in response to these challenges, often surprising customers mid-contract.

Despite years of iteration and re-launches, the solar roof remains a marginal product with limited traction outside of Tesla enthusiasts and high-end custom homes. It is unclear whether the underlying concept can ever scale given the technical integration and economic tradeoffs involved.

Cybertruck was revealed in 2019 with the goal of mass production by late 2021. Deliveries began in late 2023. Tesla sold just under 39,000 Cybertrucks in 2024. That’s a fraction of the more than 1.6 million reservations the company claimed. From the outset, the vehicle’s extreme styling and materials introduced substantial manufacturing complexity. The stainless steel exoskeleton, a defining design element, proved difficult to stamp, weld, and paint. Its angular surfaces required new tooling and manufacturing tolerances that delayed production line readiness. Even by Tesla’s own admission, getting Cybertruck into production involved more engineering and factory challenges than anticipated. When released, its doors and hood sliced carrots in half, making it clear that they were hazards to children’s fingers. And, of course, its massive front end was a danger to children as well, just as all US pickup trucks are.

The vehicle has faced persistent quality concerns, including panel alignment, inconsistent trim, and component failures. In early 2025, Tesla recalled over 46,000 Cybertrucks due to defects in the accelerator pedal assembly. Meanwhile, demand appears to have softened. Q4 2024 saw a 22% sales decline relative to Q3. Tesla has since shifted workers away from the Cybertruck line to boost Model Y output, suggesting internal resource constraints and potentially weaker-than-expected sales velocity.

The price has also deviated sharply from original expectations. Instead of a sub-$40,000 truck, the initial trims are priced at more than double that, pushing the vehicle into a premium category that alienates some of its early reservation holders. Beyond the economics, Cybertruck has become the most polarizing product from an increasingly polarizing company. Its radical design has inspired loyalty and derision in equal measure. It now stands as a symbol not just of Tesla’s ambitions, but also of its struggle to translate vision into manufacturable, supportable reality. It’s most likely to fade into the annals of quirky automotive failures.

Tesla’s Full Self-Driving (FSD) journey has been marked by ambitious promises and persistent delays. In 2016, Elon Musk asserted that by 2017, a Tesla would autonomously navigate from Los Angeles to New York without human intervention—a milestone that remains unachieved. Between 2018 and 2024, the anticipated arrival of Level 5 autonomy was repeatedly projected and deferred. As of 2025, FSD remains at Level 2, necessitating full driver attention. While software iterations, particularly FSD Beta versions 12 and 13, have shown measurable improvements, Tesla has yet to demonstrate consistent hands-free driving in unstructured environments. The robotaxi initiative, initially slated for a 2020 launch, is now tentatively planned for later in 2025, beginning in Austin. Despite the software’s advancements, it continues to require active supervision.​

Tesla’s approach to autonomous driving has diverged from mainstream AI methodologies. Initially, the company leveraged a neural network architecture that processed vast amounts of data from its fleet, utilizing human driver interventions as feedback to refine its models. This strategy aimed to emulate human-like driving behaviors through reinforcement learning. However, the broader AI community has shifted focus toward large language models and visual question-answering systems, which, while impressive, do not directly address the real-time decision-making challenges inherent in autonomous driving. This divergence has left Tesla to navigate the complexities of FSD development largely independently.​

In 2021, Tesla made the contentious decision to remove radar sensors from its vehicles, opting to rely solely on camera-based vision systems. Elon Musk defended this move by highlighting the capabilities of advanced neural networks to process visual data, drawing parallels to human vision. While humans drive without radar, the absence of complementary sensor modalities like lidar—dismissed by Musk as “stupid, expensive, and unnecessary”—has sparked debate. Critics argue that this sensor reduction compromises the system’s ability to perceive and react to complex driving scenarios, especially under adverse conditions. Notably, the National Highway Traffic Safety Administration (NHTSA) initiated an investigation into FSD following incidents where the system reportedly failed in low-visibility situations, including a fatal pedestrian accident. ​Most recently former JPL engineer and now YouTuber Mark Rober released a video in which he managed to get it to run into a wall painted like the road and into mannequin of a child hidden in fog.

The challenges Tesla faces are multifaceted. Reinforcement learning, the foundation of Tesla’s FSD strategy, has proven to be more data-intensive and slower to yield results than initially anticipated. While early systems leapt forward, every new release is asymptotic to the curve, forever getting half way to the goal with each step. Moreover, the machine learning community’s pivot toward applications like ChatGPT and visual question-answering has left Tesla’s sensor and speed requirements misaligned with prevailing AI research trajectories. This misalignment necessitates that Tesla shoulder the majority of its R&D efforts independently, without the benefit of broader industry collaboration.​

And so, to the Tesla Semi. In 2022, it completed a 500-mile run with a full load. Its acceleration, grade performance, and energy efficiency are real. But building dozens of trucks in a pilot line is different from producing tens of thousands annually at a price anyone will pay. The company is still building the Nevada production facility. Megacharger infrastructure is sparse. Service networks are minimal. Maintenance programs are still being worked out with early partners like PepsiCo.

Tesla’s 4680 battery cell, unveiled in 2020, was heralded as a game-changer for electric vehicles, promising enhanced energy density and reduced production costs. However, the journey from prototype to mass production has been fraught with challenges. Technical hurdles, such as issues with dry electrode technology and thermal management, have impeded the scaling process. Reports indicate that during test production, Tesla experienced cathode scrap rates as high as 70% to 80%, a stark contrast to the sub-2% rates typical in conventional battery manufacturing. ​

Compounding these internal challenges are external critiques from industry leaders. Robin Zeng, CEO of Contemporary Amperex Technology Co. Limited (CATL), the world’s largest battery manufacturer and a key supplier for Tesla’s Shanghai Gigafactory, has expressed skepticism about the viability of the 4680 cell. Zeng reportedly told Elon Musk that the 4680 battery “is going to fail and never be successful,” highlighting concerns over its cylindrical design and manufacturing complexities.

Prismatic cells, favored by CATL and others, offer advantages in energy density per unit volume and can be more easily stacked for optimal thermal management. Cylindrical cells like Tesla’s 4680 are structurally robust and familiar in terms of high-throughput manufacturing, but the larger format introduces challenges with heat dissipation, mechanical stability, and consistent quality at scale. Zeng’s criticism reflects broader skepticism in the industry that cylindrical cells, particularly in oversized formats like the 4680, may not deliver the promised performance or cost breakthroughs once deployed in mass-market applications.

To be fair, not all of Musk’s ventures over the past decade have failed to deliver. SpaceX’s Falcon 9 and Starlink systems have both scaled dramatically since 2015. Falcon 9’s reusable booster program began development in 2011 and saw its first landing in 2015, well before Starship consumed attention and resources. Starlink began internal development in 2015 and launched its first satellites in 2019, quickly becoming the most deployed satellite constellation in history by 2024. Tesla’s Model 3 and Model Y programs also bore fruit, though both were announced well before 2016 and benefited from mature EV platforms and years of iterative improvement. Giga Shanghai, launched in 2019, is an outlier in terms of construction and scale-up speed, though it was aided heavily by local partnerships and government incentives. Tesla’s Supercharger network, initiated in 2012, and its grid battery products like Megapack, which began scaling in earnest in 2020, have also found commercial traction. Even there, Musk axed the team that delivered the win and it’s been struggling since. But these programs either predate Musk’s more ambitious claims or owe much of their success to teams, partners, and conditions largely outside his direct control. Where Musk has dominated the narrative—particularly with newer programs post-2016—consistent execution has been harder to find.

The success of the Tesla Semi is intricately linked to the consistent and scalable production of the 4680 battery. While Tesla’s recent milestones are promising, the initial production inconsistencies and external skepticism cast a shadow over the company’s ability to meet its ambitious targets. As Tesla continues to navigate these challenges, the broader question remains: can the company effectively translate its innovative vision into reliable, large-scale manufacturing success while Musk is not paying attention to the useful but boring stuff?

Almost every major new program in existing or new companies Musk has launched in the past decade—from tunneling, to solar tiles, to trucks, to rockets, to autonomy—has missed its timelines by years. Most have been scaled back, delayed, or indefinitely deferred, failures by most objective measures. The Semi may be different, but the pattern is deeply concerning. The United States needs the Semi that delivered 1,000 mile plus days with a couple of brief charging stops in NACFE’s 2023 Run On Less for a reasonable price point, not something that costs double or triple the reported $250,000 price tag.