Editorial illustration for Thinking Machines Challenges OpenAI: Superintelligence Beyond Simple Scaling
Thinking Machines Defies OpenAI's Scaling Strategy for AI
Thinking Machines challenges OpenAI scaling, says superintelligence is a learner
The artificial intelligence race just got more intriguing. A bold challenge is emerging from Thinking Machines, a startup that's quietly positioning itself as a contrarian voice in the high-stakes world of AI development.
While tech giants pour billions into massive machine learning models, hoping sheer computational power will unlock artificial general intelligence, this company sees a different path. Their researchers aren't buying the conventional wisdom that bigger always means smarter.
The startup, known for its secretive approach and strategic thinking, is preparing to shake up fundamental assumptions about AI progress. Something significant is brewing - and it's not just another incremental technical improvement.
At the heart of their argument lies a fundamental question: Can superintelligence truly emerge from simple scaling? Or is there something more nuanced, more complex happening in the development of advanced AI systems?
These are the provocative questions Thinking Machines is about to put front and center in the global AI conversation.
While the world's leading artificial intelligence companies race to build ever-larger models, betting billions that scale alone will unlock artificial general intelligence, a researcher at one of the industry's most secretive and valuable startups delivered a pointed challenge to that orthodoxy this week: The path forward isn't about training bigger — it's about learning better. "I believe that the first superintelligence will be a superhuman learner," Rafael Rafailov, a reinforcement learning researcher at Thinking Machines Lab, told an audience at TED AI San Francisco on Tuesday. "It will be able to very efficiently figure out and adapt, propose its own theories, propose experiments, use the environment to verify that, get information, and iterate that process." This breaks sharply with the approach pursued by OpenAI, Anthropic, Google DeepMind, and other leading laboratories, which have bet billions on scaling up model size, data, and compute to achieve increasingly sophisticated reasoning capabilities. Rafailov argues these companies have the strategy backwards: what's missing from today's most advanced AI systems isn't more scale — it's the ability to actually learn from experience.
The AI landscape is shifting beneath our feet. Thinking Machines suggests scaling up isn't the silver bullet many believe it to be.
Rafael Rafailov's perspective challenges the current Silicon Valley orthodoxy of simply building bigger models. His core argument? Superintelligence will emerge through superior learning mechanisms, not just massive computational power.
This stance represents a provocative counterpoint to the prevailing industry narrative. While tech giants pour billions into increasingly complex neural networks, Thinking Machines proposes a fundamentally different approach focused on learning efficiency.
The implications are significant. If Rafailov is correct, the next breakthrough in artificial intelligence might not come from brute-force computational expansion, but from more nuanced, adaptive learning strategies.
Still, questions remain. How exactly would a "superhuman learner" function? What distinguishes this approach from current machine learning techniques?
For now, Thinking Machines has thrown down an intellectual gauntlet. Their challenge to OpenAI and other tech leaders suggests the AI race isn't just about size, it's about smarts.
Further Reading
- 2025 was the year AI got a vibe check - TechCrunch
- These Startups Went From Zero To Unicorn In Under 3 Years - Crunchbase News
Related Reading
Browse all OpenAI news →
Business & Startups OpenAI, a Series F San Francisco startup founded in 2015 by eight pioneers
Business & Startups Terminal-Bench 2.0 launches with Harbor, testing any container-installable agent
Business & Startups Zuckerberg Unveils Meta Compute to Build Global AI Infrastructure
Related Topic Gen AI app sessions up fivefold, downloads jump 778% as ChatGPT leads traffic
LLMs & Generative AI
Related Topic GPT-5 helps mathematicians offload tedious tasks, says Timothy Gowers
Open Source
From Archives Indian IT lays off staff, calls it restructuring; AI tasks need human effort
From Archives HCLtech signs undisclosed multi-year FY26 deal using AI Force platform
From Archives ChatGPT integrates company knowledge from connected tools for smarter answers
LLMs & Generative AI
From Archives ChatGPT Apps SDK lets 800 M users run apps in-chat without extra installs
LLMs & Generative AI
Common Questions Answered
How does Thinking Machines challenge the current approach to artificial general intelligence development?
Thinking Machines argues that the path to superintelligence isn't about building increasingly larger models with more computational power. Instead, the company believes that the breakthrough will come from developing superior learning mechanisms that create more efficient and adaptive artificial intelligence systems.
What is Rafael Rafailov's key perspective on achieving superintelligence?
Rafael Rafailov believes that the first superintelligence will be a superhuman learner, challenging the conventional wisdom of simply scaling up machine learning models. His view suggests that advanced learning capabilities, rather than raw computational size, will be the critical factor in developing truly intelligent AI systems.
Why does Thinking Machines consider the current AI development strategy problematic?
Thinking Machines sees the current approach of pouring billions into massive machine learning models as fundamentally flawed. The startup argues that tech giants are mistakenly believing that computational scale alone will unlock artificial general intelligence, when in fact, more sophisticated learning mechanisms are the key to breakthrough AI development.