AI Hardware Shift: Key Players and Trends
Nvidia’s GPUs currently dominate the AI hardware landscape, but companies like Alphabet and OpenAI are pivoting towards custom chips. Alphabet has developed its Tensor Processing Unit (TPU) aimed at enhancing its AI program, Gemini, while Anthropic plans to deploy 1 gigawatt of computing power using TPUs this year, signaling a shift away from Nvidia’s general-purpose GPUs.
Broadcom, which has partnered with Alphabet and OpenAI to design custom AI chips, is targeting $100 billion in AI chip revenue by the end of 2027, expecting revenue to double to $8.4 billion by 2026. Meanwhile, Taiwan Semiconductor Manufacturing Company controls 72% of the global market for semiconductor manufacturing, producing chips for major tech firms and generating $122.42 billion in revenue for 2025, marking a 35.9% increase from 2024.
With a projected 60% market share in AI-specific chips by 2027, Broadcom’s recent expansions underline the industry’s movement towards specialized hardware, positioning both it and Taiwan Semiconductor as key beneficiaries in the evolving AI landscape.
