Perplexity's Memory-Efficient Embeddings Drop as Iran Strikes Push AI Storage Costs Past $6,000 Monthly

The Day's Thesis

OpenAI's $110 billion funding round signals unprecedented capital flows into AI infrastructure just as geopolitical tensions threaten the supply chains and energy systems that underpin the entire stack. The collision between record AI investment and escalating Middle East risks is forcing a fundamental repricing of compute, storage, and critical materials across the technology ecosystem.

AI & Research Frontier

Perplexity's open-source embedding models deliver Google and Alibaba-level performance at dramatically reduced memory footprints, representing a critical breakthrough in AI efficiency as storage costs spiral beyond reach for many applications.

The timing couldn't be more crucial: Myrient's 390TB video game archive faces March closure after AI-driven memory price inflation pushed monthly costs above $6,000, demonstrating how frontier model training is cannibalizing storage capacity across the ecosystem.

Meanwhile, fundamental AI limitations are emerging that could reshape enterprise deployment strategies. Research reveals that even GPT-5-class frontier models lose up to 33% accuracy during extended conversations, a degradation that forces higher computational overhead to maintain service quality.

This performance decay coincides with OpenAI's legal characterization of AI safety researcher Stuart Russell as a "doomer" despite CEO Sam Altman previously endorsing Russell's extinction risk warnings—highlighting the tension between commercial scaling pressures and safety considerations as models approach more capable thresholds.

Technology & Infrastructure

OpenAI's record $110 billion funding round—anchored by Amazon's $50 billion, Nvidia's $30 billion, and SoftBank's $30 billion commitments—drives the company's valuation to $730 billion while virtually guaranteeing massive semiconductor demand acceleration.

This capital injection comes as Chinese humanoid robotics companies outpace US competitors in market deployment, leveraging manufacturing scale advantages that translate directly to higher consumption of rare earth magnets and advanced motor control semiconductors.