Palantir CEO Alex Karp Criticizes Closed-Source Large Models on CNBC, Calls Products from OpenAI, Anthropic, etc. 'Garbage'
Karp stated that when companies pay to use large models, their core data is stolen by big firms for training, effectively spending money to support competitors; closed-source models are overly hyped, outsourcing the lifeblood of American enterprises and the military to a few laboratories.
This statement highlights Palantir's emphasis on data security and corporate control.
In market mechanisms, AI enterprise clients, as buyers, are concerned about data security. This incident drove Karp's performance on the live broadcast, directing capital towards platforms like Palantir that emphasize data sovereignty; Palantir benefits from reinforcing its market positioning through this narrative, while closed-source large model companies face short-term pressure from trust issues.
Source: Public Information
ABAB AI Insight
Alex Karp has previously criticized closed-source AI models multiple times, and this CNBC live broadcast continues Palantir's consistent theme of "data sovereignty" and corporate control, similar to its early emphasis on data security in intelligence analysis.
On the capital front, Palantir guides enterprise client resources through public criticism, motivated by promoting its platform's advantage of keeping data in-house while generating momentum for its business model.
Similar to critiques by Thiel or Andreesen against closed systems, the AI industry is currently in a critical phase of public opinion battle between closed-source and open-source/corporate control models.
Essentially, this is a capital reallocation: Karp's criticism of closed-source large models highlights the risk of data leaks, accelerating the reallocation of corporate capital from token payments to self-controlled platforms and strengthening the market share of data security solutions like Palantir.
ABAB News · Cognitive Law
When data is stolen for training, companies are effectively spending money to support competitors.
Closed-source models are overly hyped, outsourcing risks to a few laboratories.
With Palantir-style data sovereignty, corporate control is restored.