Archive · 28 Dec 2025
Groq CEO Identifies Key Flaws in ChatGPT
Jonathan Ross, the CEO of Groq, thinks ChatGPT and similar AI models have two big problems.
First, they are too slow to respond, which makes them feel less like a natural partner in conversations.
Second, running these AI models costs a lot of money and uses a lot of energy.
Ross believes that the current computer chips used for AI, called GPUs, are not the best for the job.
His company, Groq, is making special chips called LPUs that he says are much faster.
He thinks better hardware is just as important as improving the AI models themselves.
Groq CEO Jonathan Ross identifies latency and high compute costs as major flaws in ChatGPT's evolution.
Ross argues that current AI models are too slow for real-time interaction, masking delays with a 'typing' effect.
The economic and environmental costs of running large language models are deemed unsustainable by Ross.
Groq's Language Processing Units (LPUs) are positioned as a faster alternative to traditional GPUs.
Ross's critique comes as Groq competes with Nvidia in the AI hardware space.
- Who
- Jonathan Ross, CEO of Groq
- What
- Identified two major flaws in ChatGPT's evolution
- When
- At a recent industry event
- Where
- Not specified
- Why
- To highlight the need for faster, more efficient AI hardware
- CEO
- Jonathan Ross
- Company
- Groq
- Flaws Identified
- Latency and high compute costs
- Event
- Industry event
- Hardware Focus
- Language Processing Units (LPUs)
- Current Hardware
- General-purpose GPUs
- Claim
- Groq LPUs faster than traditional GPUs