Yazılar

US Grants Exemption for Zoox Self-Driving Vehicles, Ends Safety Probe

The National Highway Traffic Safety Administration (NHTSA) has certified Amazon’s (AMZN.O) self-driving unit Zoox vehicles for demonstration use and closed its investigation into whether the vehicles complied with federal safety standards.

The agency began probing in 2022 after questions arose about whether Zoox’s purpose-built autonomous vehicles, which lack traditional driving controls like steering wheels and brake pedals, met federal safety requirements under the company’s self-certification.

In June, Zoox applied for an exemption from some regulatory requirements, which NHTSA granted, allowing the vehicles to operate legally on U.S. public roads under this exemption. The agency also required Zoox to remove any claims that their vehicles fully comply with applicable federal motor vehicle standards.

The Trump administration had signaled in June an intent to accelerate approvals for self-driving vehicle exemptions, following stalled proposals by General Motors and Ford to deploy vehicles without steering wheels and pedals.

In May, Zoox recalled 270 driverless vehicles after an unoccupied robotaxi was involved in a crash in Las Vegas due to a scenario where the automated system might misjudge another vehicle approaching perpendicularly and failing to stop, potentially causing collisions. Zoox paused operations briefly, conducted a safety review, and deployed a software update to fix the issue.

NHTSA also closed a related probe from May 2024 into 258 Zoox vehicles after a software update resolved a braking problem that led to two rear-end collisions involving motorcyclists.

Europe’s Ageing Power Plants Set for AI-Driven Data Centre Transformation

Big tech firms, including Microsoft and Amazon, are eyeing Europe’s retiring coal and gas plants as prime locations for new data centres — tapping into their existing power grid connections, water infrastructure, and cooling systems to meet surging AI energy demands. Utilities such as Engie, RWE, and Enel see these conversions as a way to offset decommissioning costs, secure lucrative long-term power contracts, and underwrite future renewable projects.

Many of the EU’s and UK’s 153 remaining hard coal and lignite plants are scheduled to close by 2038, joining the 190 that have shut since 2005. Repurposing these sites offers utilities stable, high-margin revenues, with tech companies reportedly paying up to €20/MWh in “green premiums” for low-carbon electricity. Depending on scale — some data centres can require up to a gigawatt — such premiums could translate into contracts worth hundreds of millions to billions of euros over decades.

The approach also addresses one of Europe’s key data centre bottlenecks: grid connection delays, which can stretch over a decade. Converting old plants offers “speed to power,” significantly accelerating deployment timelines. Projects range from retrofitting existing sites to building “energy parks” pairing renewable generation with direct supply to data centres.

Engie is actively marketing 40 potential sites worldwide, including its decommissioned Hazelwood coal plant in Australia, while EDP, EDF, Enel, and Britain’s Drax are pursuing similar strategies. Some developments, such as a planned 2.5 GW facility at a former German coal plant and multiple UK sites, are already in motion — though details remain scarce for security reasons.

Industry analysts say the trend represents a diversification of utility business models, creating new revenue streams and fostering long-term tech–energy partnerships. For hyperscalers, the premium is worth paying if it secures earlier market entry in the AI race.

Amazon eyes deeper investment in Anthropic to stay ahead in AI race

Amazon is reportedly considering another multibillion-dollar investment in Anthropic, the artificial intelligence firm behind the Claude AI models, according to the Financial Times. The potential move would strengthen Amazon’s position as a major player in the rapidly intensifying global AI race.

The report, citing sources familiar with the matter, says Amazon wants to expand on the $8 billion investment it committed to Anthropic in November 2023. That initial deal, which included an upfront $4 billion, made Amazon one of the company’s largest stakeholders, alongside Google, which has invested more than $3 billion into Anthropic.

Both Amazon and Anthropic declined to comment on the renewed talks when contacted by Reuters.

A race to stay relevant in AI

Amazon’s increasing interest in Anthropic highlights its urgency to catch up to rivals OpenAI and Google, who have made significant consumer-facing advances in generative AI over the past two years. Anthropic’s Claude family of AI models competes directly with OpenAI’s ChatGPT and Google’s Gemini.

“We quickly realized that we had many shared goals that were fundamentally critical,” said Dan Grossman, Amazon’s VP of worldwide corporate development. “The size of the (existing investment) represents our ambition.”

Amazon’s deepened partnership with Anthropic could also help it attract top AI talent, an increasingly competitive space where companies are offering equity, massive compensation packages, and research freedom to lure leading minds in machine learning and large language models.

Strategic implications

Amazon’s AI ambitions are closely tied to its cloud business, AWS, where Anthropic’s models are being integrated into services for enterprise customers. The ongoing partnership gives Anthropic priority access to AWS’s Trainium and Inferentia chips, optimizing both model development and deployment.

Beyond infrastructure, Amazon is aiming to embed Claude-powered AI tools deeper into Alexa, Amazon Web Services, and its e-commerce ecosystem, which could give it an edge in personalized search, voice interfaces, and customer service automation.

The prospective increase in funding would also help Amazon maintain equity leadership in Anthropic amid growing investor interest in the startup. With AI startups commanding soaring valuations, Amazon appears determined not to lose strategic control over a potential future titan in the field.