Google will accelerate its custom artificial intelligence chip rollout from a two-year cycle to two processors annually while expanding its Taiwan research footprint by 60 per cent.
The revised schedule shifts the company away from its traditional multi-year hardware cadence as competition for proprietary cloud computing silicon intensifies across the tech sector.
Faster Silicon Cadence
Amin Vahdat, Google’s senior vice president and artificial intelligence infrastructure chief, announced the accelerated timeline during a keynote address at Semicon Taiwan in Taipei on Wednesday. The company plans to release two bespoke chips every year and expects to lift that frequency even further over time.
Backing the faster production tempo requires more engineering capacity on the ground. Google is growing its dedicated research and development floor space across Taiwan by 60 per cent to house expanded design and testing teams.
Regional Hardware Footprint
Taiwan sits at the centre of global advanced semiconductor fabrication and packaging. By enlarging its local engineering hubs, Google tightens operational proximity to key foundry partners, contract assemblers and component supply chains that manufacture its custom processing units.
For cloud platforms and enterprise services operating across Asia-Pacific, in-house silicon helps control operational power costs and workload efficiency in regional data centres. Competing tech operators across the region are running similar programs to secure custom processing capacity.
The company will now focus on staffing the expanded Taiwan design facilities ahead of its next scheduled processor rollouts.