Why Is Everyone Making Their Own? :AI Boom Brings Hardware Back to the Spotlight

In the early 2020s, the tech world was dominated by software.

But since 2023, with generative AI going mainstream, hardware, especially AI chips has become central again.

Running models like ChatGPT, Claude, or Gemini requires massive computational power, primarily powered by GPUs.

And the market leader? NVIDIA.

Yet, in recent years, companies are challenging NVIDIA’s dominance.

AMD, Intel, ARM, and even AI firms like OpenAI and Anthropic are developing their own chips.

NVIDIA’s Dominance

Originally a graphics card manufacturer, NVIDIA captured the AI computing market with parallel-optimized GPU architectures.

• AI-focused GPUs: A100, H100, B100

• Global AI chip market share (2023): 80%+

• Market cap briefly reached $3 trillion, making NVIDIA the world’s most valuable company for a moment

But NVIDIA GPUs are expensive and supply-constrained.

Both startups and big tech companies see the need to reduce dependency.

New Challengers in the Market

The Essence of the AI Chip War

It’s not just about raw performance. The battle revolves around:

2. - Performance

3. → Faster training, more efficient inference

4. - Energy efficiency

5. → Large models consume enormous power; cost & carbon footprint matter

6. - Supply & pricing

7. → High-end GPUs cost tens of thousands USD; affordable, accessible chips are needed

Geopolitics and Chips

The AI chip war also has geopolitical dimensions.

• USA restricts GPU exports to China → AI chips become a national security concern

• China accelerates domestic chip development (Huawei Ascend, Alibaba Pingtouge, Biren)

• South Korea, Taiwan, Japan play key roles in the semiconductor supply chain, influencing global power balance

Future Scenarios

• Multipolar Market → NVIDIA’s dominance will diversify; AMD, Intel, and specialized startups gain share

• Internal AI Chips → AI companies like OpenAI, Google, and Amazon will deploy custom chips for in-house demand

• Edge AI Chips → Explosion in AI chips for smartphones, IoT, and autonomous vehicles

Why Every AI Company Wants Their Own Chip

AI chips aren’t just components—they are core strategic assets.

Companies want to:

• - Reduce NVIDIA dependency

• - Cut costs

• - Maximize performance

In the next decade, AI innovation will hinge not only on software but also on hardware dominance.

The 21st-century “new oil war” is really a battle over data and chips.