The world of artificial intelligence is experiencing a new wave of innovation. Two ambitious startups are taking different approaches to AI development that differ significantly from the current standards of major technology giants. Flapping Airplanes has raised $180 million (approximately CZK 4.1 billion), and former OpenAI researcher Jerry Tworek is seeking up to $1 billion (approximately CZK 20.6 billion) for his Core Automation project.
Flapping Airplanes: A Young Lab with a Grand Vision
Flapping Airplanes has reached a valuation of $1.5 billion (roughly CZK 34 billion), and its goal is to create artificial intelligence that matches human intelligence without having to "swallow half the entire internet." The startup was founded by brothers Ben and Asher Spector, who are only 25 and 26 years old. Ben is the founder of Prod, which quickly became one of Silicon Valley's most prominent talent incubators. Asher is a former debate champion and a statistics PhD student at Stanford.
The third co-founder is Aidan Smith, a former Thiel Fellow who worked at Neuralink for three years while studying at Georgia Tech. The company's advisers include prominent figures such as Andrej Karpathy (the "father" of vibe coding) and Jeff Dean. Investment firm Sequoia Capital described their approach as a "young AGI lab."
Data Is the Problem, Not Computing Power
Flapping Airplanes is built on a controversial idea: scaling alone is not enough to achieve AGI (artificial general intelligence). As David Cahn of Sequoia Capital explains, data, not computing power, is now the main obstacle to further AI development. There is only one internet, and we have already exhausted it.
According to the founders, current AI models are extremely inefficient. The name Flapping Airplanes refers to a more biologically inspired concept of machine intelligence. The startup is focused on finding more data-efficient artificial intelligence inspired by biology.
Core Automation: AI That Learns from the Real World
Jerry Tworek, who left OpenAI this month, wants to build artificial intelligence that continuously learns from real-world experience – a capability that current systems lack. His plan centers on a single continuously learning model called Ceres, with ambitions ranging from factory automation to the eventual terraforming of planets.
Tworek's Core Automation project represents a radical departure from the current approach of major AI labs. Instead of training on enormous datasets from the internet, he wants to create a system that learns continuously through interactions with the real world.
A Bet Against $200 Billion Giants
This new wave of startups represents a serious contrarian bet against frontier labs with budgets exceeding $200 billion (approximately CZK 4.6 trillion). Leading AI researchers are openly questioning whether scaling alone can lead to AGI.
Flapping Airplanes believes that young people have always unlocked fundamental science – Albert Einstein was only 26 during his "miracle year." The company also believes that AI research is not a mystical domain reserved for experts, but a field that anyone smart and ambitious can learn.
Investors are betting billions on the chance that one of these teams will crack the code that major AI leaders may be overlooking. It is a high-risk game, but the potential reward – a breakthrough toward genuine artificial general intelligence – could transform the entire technology industry.
While most of the AI industry is focused on scaling current large language models, these startups are betting on fundamental research and long-term projects with a 5-10-year horizon. The question remains whether their approach will deliver the anticipated breakthrough sooner than the traditional path of massive scaling.
Sources: sequoiacap.com and theinformation.com
