Which economic bottleneck has become most critical for AI adoption, and why?
Most answers cite physical infrastructure—like data centers, hardware, and power—as the main bottleneck, though some point to organizational, financial, or workforce constraints instead.
12 answers
Answers

David Vellante
Co-founder, co-CEO SiliconANGLE, theCUBE, theCUBE Research (formerly Wikibon) · Siliconangle & Thecube
Networking
David Vellante says networking is the critical bottleneck for AI adoption, calling it the common thread because it makes data move faster and changes the game.
The answer starts at 31:42.
theCUBE Podcast · Listen on Apple Podcasts

Patrick O'Shaughnessy
Founder and CEO · Positive Sum
Hardware supply
Patrick O'Shaughnessy says the actual real limitation right now is hardware supply, adding that it is not capital or debt, and agrees with power as well.
Invest Like The Best · Watch on YouTube

Larry Cheng
Managing Partner and Co-Founder · Volition Capital
Compute capacity
Larry Cheng says AI for the moment seems to be gated by capacity, specifically compute capacity and data centers, though he notes compute is sold out for the next couple of years with massive demand.
N/A · N/A · Listen at 25:14

Jon Gray
President & Chief Operating Officer · Blackstone
AI infrastructure
Jon Gray says the scarce thing today is AI infrastructure, because supply is falling further behind rapidly growing demand due to entitlements, community pushback, turbine and chip shortages, and the difficulty of building power plants.
Blackstone · Watch on YouTube

Evan Conrad
Co-Founder · San Francisco Compute Company
Credit
Evan Conrad says the bottleneck in compute and AI is not power or chips but entirely credit, because lenders cannot underwrite the value of a GPU cluster as collateral.
Cognitive Revolution "How AI Changes Everything" · Watch on YouTube

Callum Williams
Senior economics writer · The Economist
Cost
Callum Williams says the critical bottleneck is probably cost, since companies burn through AI budgets quickly; he suggests that if cost stopped constraining use, AI diffusion would increase.
The answer starts at 5:41.
MTS · Listen on Apple Podcasts
David George
General Partner · Andreessen Horowitz
Data-center capacity
David George says data-center capacity is the critical constraint, noting you can't get it until 2028 because there are bottlenecks in every step of the supply chain.
The answer starts at 4:54.
a16z Podcast · Listen on Apple Podcasts

Doug Chapman
System Development Engineer3 · Amazon Web Services (Aws)
Adoption lag
Doug Chapman says the critical bottleneck is adoption lag, which he describes as companies' inability to adopt, potentially making them "too big to change."
Capgemini · Watch on YouTube

Rene Haas
Chief Executive Officer · Arm
Data center buildout
Rene Haas says building out data centers is going to be a bottleneck, noting that few projects are ahead of schedule or need less labor than expected, and that local pushback on data center development could add a headwind.
No Priors: AI, Machine Learning, Tech, & Startups · Watch on YouTube

Paul McDonagh-Smith
Visiting Senior Lecturer (IT Group and Executive Education) · Mit Sloan School Of Management
Organizational structure
Paul McDonagh-Smith says hardly any of the obstacles to AI deployment are technology problems, attributing them instead to organizational structure across many dimensions.
The answer starts at 21:44.
The AI Forecast: Data and AI in the Cloud Era · Listen on Apple Podcasts
Andrew Feldman
Founder and CEO · Cerebras
Real estate
Andrew Feldman says the constraint on AI compute is real estate, citing data centers and buildings as the bottleneck even though chips are extraordinary.
New York Stock Exchange · Watch on YouTube

Raymond Hawkins
Chief Customer Officer · Compass Datacenters
Electricians
Raymond Hawkins identifies electricians as the key workforce bottleneck, saying labor is trickier than other problems because the AI industry can throw money at chips and power but workforce investment sits outside its comfort zone.
Latitude Media · Watch on YouTube