PwC's Dallas Dolen, the firm's technology, media, and telecom leader, argues that value in AI is shifting away from SaaS software toward hardware and infrastructure. The trillion-dollar infrastructure buildout is compressing margins at the software layer while rewarding compute providers. Dolen coins the term 'TokenMaxxing' to describe how enterprises are deliberately maximizing token consumption to extract more output from models, a behavioral shift with direct cost implications.
Agentic AI is getting real traction inside enterprises, but not uniformly. Dolen identifies which use cases are actually working and which are still theoretical, and he addresses whether compute constraints are a genuine bottleneck or a convenient narrative. The model ecosystem battle between hyperscalers and frontier labs gets specific treatment here, with Dolen naming which players he sees holding structural advantages.
The workforce and shadow AI sections are where this interview earns a second read. Dolen gives a practitioner's view of how AI is reshaping headcount decisions and what happens when employees adopt AI tools faster than IT policy can respond. If you work in enterprise technology strategy, the gap between his grounded framing and most analyst commentary is worth your time.
[WATCH ON YOUTUBE →]