Chamath Warns Anthropic Could Become the Friendster of the AI Era
Billionaire Chamath Palihapitiya warned that if Anthropic cannot solve its computing and power limitations, it will become the "Friendster of the AI era."
Chamath pointed out that Claude refused to complete stock screening prompts, while competitors could, highlighting that the computing bottleneck is becoming the biggest threat to AI leaders.
Investors in AI infrastructure and computing power suppliers in the market are driving the expansion of power data centers. Chamath's public warning aims to guide capital to focus on infrastructure, benefiting platforms like xAI with ample computing power, while constrained labs like Anthropic face short-term pressure, with funds accelerating from model optimization to power and data center concentration.
Source: Public Information
ABAB AI Insight
Chamath Palihapitiya, as a former Facebook executive and founder of Social Capital, has long invested in AI and tech infrastructure. He has previously emphasized in the All-In Podcast that power shortages are the next battleground for AI, pointing out that data center projects face local opposition and regulatory hurdles. This Friendster analogy continues his judgment that infrastructure determines outcomes.
On the capital path, Chamath calls for Anthropic to collaborate with power-rich entities like Elon Musk, while shifting focus to underlying resources such as power, cooling, and minerals. His motivation is to remind AI labs to shift from a race of model parameters to a competition for physical infrastructure, transforming limited computing power into long-term competitive barriers through strategic partnerships.
Similar to how Friendster was surpassed by MySpace and Facebook due to its inability to handle traffic growth, and how early internet companies fell due to server limitations, Anthropic is currently at a critical juncture as AI transitions from model capability leadership to infrastructure bottlenecks. If it cannot quickly resolve power issues, it will face rapid market share loss.
Structural judgment: Essentially, this is about capital concentration. The extreme dependence of AI training and inference on power and data centers allows a few entities that control large-scale infrastructure to efficiently concentrate computing resources. The mechanism is that physical supply growth lags far behind model demand, forcing value to shift from independent model labs to platforms that control energy and data centers.
ABAB News · Cognitive Law
No matter how strong the model, without enough power, it remains Friendster.
Infrastructure bottlenecks make technological advantages mere bubbles.
In the era where computing power reigns, power is worth more than code.