Centralized research labs give researchers the opportunity to look at different ideas across the company and connect ideas. This is almost impossible if all res

Centralized research labs give researchers the opportunity to look at different ideas across the company and connect ideas. This is almost impossible if all researchers are siloed in decentralized research groups. Also, there's 'why can't we have both?'. In honesty, Google, like most tech companies today follows a hybrid model. The public statements of executives are not to be taken as gospel, because their primary audience are the company's shareholders. In reality, you have PhDs in Google X working on moon-shot problems (product focused or otherwise, they are long-term projects) as well as in direct P&L divisions. Centralized research labs also have their own versions of P&L metrics (e.g. number of papers published, product $s resulting indirectly from research, number of patents, etc.). Financially, centralized research labs working on moon-shot problems are a hedge against short term fluctuations and gives companies an opportunity to pursue long-term trends. Also, in companies that have both, there is nothing to prevent researchers from moving between centralized research labs and decentralized when opportunities present themselves (e.g. product commercialization/scale-up project resulting from research). I think it is dogmatic to pursue centralized or decentarlized exclusively.