New Street Research analyst Pierre Ferragu predicts 2025 will mark the beginning of a new artificial intelligence era, citing breakthrough developments in Tesla Inc.‘s TSLA Full Self-Driving technology and OpenAI’s latest O3 model.
What Happened: In a detailed research note, Ferragu emphasizes the remarkable performance of OpenAI’s O3 model on the Advanced Reasoning Corpus – Artificial General Intelligence benchmark, a testing framework that assesses the reasoning and problem-solving abilities of AI systems.
The model achieved an impressive 87.5% score, nearing human-level capabilities and marking a substantial improvement over previous AI models, which typically scored below 10%.
“This is not merely incremental improvement, but a genuine breakthrough, marking a qualitative shift in AI capabilities,” says François Chollet, founder of the ARC-AGI foundation, as quoted in the research report.
The O3 model, set for release in January, demonstrates remarkable improvements across multiple benchmarks. It achieved a 71.7% accuracy in software engineering tasks, matching human professional performance, and scored 2,727 Elo in competitive coding, surpassing many expert programmers.
“Between FSD v13.2.2 driving me around with literally no intervention and o3 scoring 88% at ARC-AGI, I do believe the inflection is now. 2025 will be year zero of the AI era,” Ferragu wrote on X.
See Also: Benzinga’s Top 5 Stories In 2024: The Year Of The Cybertruck
Why It Matters: However, these advances come with substantial computational costs. The high-performance version of O3 requires approximately 2,000 times more computing power than its predecessor, with a single AGI task costing about $3,500 to process.
Ferragu suggests this breakthrough has significant implications for AI infrastructure providers. “We likely are still only at the very beginning of the growth curve for AI infrastructure and its supply chain,” he notes, indicating potential investment opportunities in the sector.
While O3 represents a major step toward Artificial General Intelligence, Ferragu clarifies it hasn’t achieved full AGI status yet. The model is expected to face challenges with next-generation tests, where estimates suggest it may not exceed 25% accuracy.
Read Next:
Image Via Shutterstock
Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.
© 2025 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.
Comments
Trade confidently with insights and alerts from analyst ratings, free reports and breaking news that affects the stocks you care about.
