AI's Growing Pains: From Carbon Credits to Memory Crunch and Job Fears

This week, the AI industry's rapid expansion collided with hard realities across hardware, environment, and labor markets. Google, Anthropic, Salesforce, and others pledged nearly $1 billion to buy future carbon removal credits, while budget phone maker Nothing scrapped its CMF Phone 3 Pro because AI's memory demands made a meaningful upgrade impossible. Meanwhile, a new benchmark revealed that even the best AI models fully solve only 3 percent of realistic knowledge work tasks, and top CEOs including Jeff Bezos and Sam Altman reversed their earlier warnings, now predicting AI will create a labor shortage instead of mass unemployment.

The carbon removal commitment, announced this week, aims to help startups scale direct air capture and other technologies by guaranteeing future demand. The group's pledge totals roughly $1 billion, according to TechRadar Pro, with the goal of proving that carbon removal can work at industrial scale.

Memory demands kill a budget phone

Nothing's decision to cancel the CMF Phone 3 Pro illustrates a growing hardware bottleneck. The company's co-founder said the phone could not deliver a genuine step forward without expensive memory chips, which AI features require in increasing quantities. The Next Web reported that this marks a shift from a price story to a product story: AI's hunger for memory is no longer just raising costs but killing products entirely.

• Nothing scrapped the CMF Phone 3 Pro because AI memory requirements made it impossible to offer a budget device that felt like an upgrade.
• The move follows a year of rising memory chip prices driven by AI demand.
• Apple has also reportedly flinched, adjusting its own hardware plans in response to the same constraints.

Benchmark reveals AI's knowledge work limits

A new benchmark from The Decoder shows that even the most advanced AI models struggle with realistic knowledge work. The best model fully solved just 3 percent of tasks, exposing a wide gap between hype and practical utility. The benchmark was designed to test tasks that mirror real office work, such as research, analysis, and decision-making.

Separately, Anthropic overhauled its Claude Design tool to improve the handoff between design and engineering, but a designer and an engineer disagreed on whether the fix worked, according to The New Stack. The disagreement underscores the difficulty of integrating AI into complex workflows.

CEOs flip the script on jobs

In a notable narrative reversal, top tech leaders are now downplaying AI's threat to jobs. Jeff Bezos said in Paris that AI would cause a labor shortage, not mass unemployment, and would unlock endless demand for builders and entrepreneurs. Sam Altman made similar comments days earlier, according to The Next Web. The shift comes after two years of warnings that AI would replace workers across industries.

What comes next: The industry faces a balancing act. Carbon removal credits may help offset AI's growing energy footprint, but hardware constraints and performance gaps remain unresolved. The job market narrative may continue to flip as companies deploy AI in real workplaces and discover its limits.