Archive · 26 Jul 2025
New AI Chip Designs Aim to Solve Energy Consumption Problem
New AI supercomputers need a lot of power, and experts are worried about how much energy they'll use.
Companies are creating new computer chips to make AI more energy-efficient.
One startup, Positron, is making chips that could be much better than the current leader, Nvidia.
Companies like Cloudflare are testing these new chips.
Even with these new chips, the demand for AI is growing so fast that it might still need more energy overall.
So, the challenge is to find new ways to produce energy for AI.
Cloudflare is testing new microchips from Positron that could be more energy-efficient for AI inference.
Positron has secured $51.6 million in new investment, with total investment reaching $75 million.
Several companies, including Google, Amazon, and Microsoft, are investing in AI chip development.
Nvidia's new Blackwell systems are more energy-efficient than previous generations, but faces competition.
Experts predict that even with more efficient chips, overall AI energy consumption will likely continue to grow.
- Who
- Cloudflare, Positron, Nvidia, Google, Amazon, Microsoft, and other tech companies.
- What
- Development of new, energy-efficient chips for AI.
- When
- Ongoing, with new startups and investment rounds in 2023 and beyond.
- Where
- Silicon Valley and data centers globally.
- Why
- To reduce the energy consumption of AI and lower costs.
Supporters of New AI Chips
Skeptics on Energy Savings
Energy Efficiency Impact
Supporters of New AI Chips
New, efficient AI chips can dramatically cut energy use, thus reducing operational expenses and making AI development more sustainable.
Skeptics on Energy Savings
Energy production, not data centers and chips, could be the real bottleneck for future development of AI.
Chip Design Approach
Supporters of New AI Chips
Specialized chips for inference offer significant advantages in speed and efficiency.
Skeptics on Energy Savings
Flexibility and broader application are critical, and these chips aren't necessarily the best fit.
- Positron Investment
- $51.6 million (new round), $75 million (total)
- Nvidia Gross Margin
- ~60%
- Cloudflare's Role
- Testing Positron chips
- Primary Focus
- AI inference
- Companies Involved
- Positron, Groq, Nvidia, Cloudflare