Jeff Bezos' EV Startup: A Week in Review

This Week's Top Tech Stories

Welcome to this week’s review of significant developments in the technology sector! A diverse range of stories have emerged, including investment in an electric vehicle startup, allegations against a major social media company, and potential developments concerning autonomous vehicle data usage. Let's delve into the details.

Electric Vehicle News: Slate's Ambitious Plans

Slate, a new entrant in the electric vehicle market, is pursuing the development of a cost-effective, two-seater pickup truck. The target price point for this vehicle is $25,000, making it an appealing option for budget-conscious consumers.

The company has secured substantial funding to support its objectives, notably receiving backing from Jeff Bezos. Production of the vehicle is currently anticipated to commence as early as late 2026.

Meta Faces Serious Accusations

Sarah Wynn-Williams, a former executive at Facebook (now Meta), recently provided testimony to the U.S. Senate. This testimony stemmed from her experiences and insights detailed in a book she authored about her time with the company.

Wynn-Williams alleged that Meta engaged in direct collaboration with the Chinese Communist Party. She stated this collaboration was intended to compromise U.S. national security and violate core American principles.

Nikola's Potential Revival

Trevor Milton, the founder of Nikola who received a pardon following a conviction for securities fraud, is attempting to reacquire the assets of his former company. Nikola is currently navigating bankruptcy proceedings.

It remains uncertain whether any other entities have submitted offers to purchase Nikola’s assets at this time.

Stay Informed with TechCrunch's Week in Review

This has been TechCrunch’s summary of the week’s most important tech news. To receive this recap directly in your inbox each Saturday, you can subscribe to the newsletter here.

Recent Developments

Data Collection Practices: An unreleased draft of Waymo’s privacy policy indicates the company intends to leverage data gathered from its robotaxis. This includes video footage from interior cameras linked to rider identification. The collected data will be utilized for training generative AI models.

Users will reportedly have the option to decline participation in this data collection process.

Energy and Policy

Coal's Continued Role: President Trump recently issued an executive order supporting the use of coal for powering data centers. This directive mandates the designation of coal as a critical mineral.

Furthermore, the order aims to prevent the decommissioning of certain coal-fired power plants, ensuring their continued operation.

Executive Insights

Elon Musk's Preferences: A data breach at NetJets, a private jet company owned by Berkshire Hathaway, revealed internal guidelines for serving Elon Musk. These guidelines detail his specific preferences during flights.

Musk reportedly prioritizes speed and direct routes, disregarding fuel conservation. He also prefers a cabin temperature of 65 degrees Fahrenheit.

Personnel Moves

OpenAI Alumni Join New Venture: Thinking Machine Labs, a new AI company founded by former OpenAI CTO Mira Murati, has secured prominent advisors. Bob McGrew, previously OpenAI’s chief research officer, is among those joining the team.

Alec Radford, a former OpenAI researcher instrumental in the development of key innovations, will also contribute as an advisor.

Dropbox Leadership Change: Eric Cox, Dropbox’s chief customer officer who joined the company in 2023, is departing from his role. An SEC filing confirmed this transition.

The search for a replacement is currently underway.

Innovation and Reflection

HVAC Focus: Matt Rogers, the co-founder of Nest, remains deeply engaged with the field of HVAC systems. He acknowledged that Nest’s current trajectory differs from his initial vision.

Despite selling the company, Rogers’ passion for HVAC technology persists.

Humorous Note

A1 vs. AI: During a summit on the impact of AI on education, U.S. Secretary of Education Linda McMahon playfully referred to AI as “A1,” similar to the steak sauce.

While initially using the correct terminology, she occasionally slipped, suggesting a lighthearted moment rather than a genuine misunderstanding.

Cost Considerations in AI Model Evaluation

The operational expenses associated with Artificial Intelligence are substantial for businesses. Beyond the costs of running AI systems, the process of testing and evaluating these models is proving to be significantly expensive.

Evaluation Costs of Advanced Models

For instance, assessing OpenAI’s o1 reasoning model incurs a cost of $2,767. Benchmarking Anthropic’s Claude 3.7 Sonnet, a recently released “hybrid” reasoning model, using the same testing parameters amounts to $1,485.35.

Comparison with Previous Generation Models

These figures contrast sharply with the evaluation costs of earlier models. Evaluating OpenAI’s o1-mini is priced at $141.22, while its predecessor, Claude 3.7 Sonnet’s non-reasoning version, costs only $81.41.

As Kyle Wiggers reports, the increasing complexity and scale of AI models are directly contributing to the rising expenses associated with their benchmarking and evaluation.

Factors Driving Up Benchmarking Costs

• Model Size: Larger models require more computational resources for testing.
• Complexity: More intricate models necessitate more comprehensive and time-consuming evaluations.
• Reasoning Capabilities: Models with advanced reasoning abilities demand more rigorous testing scenarios.

Therefore, AI model evaluation is becoming a considerable financial undertaking for companies seeking to leverage the latest advancements in the field.