Sign up to access an inspiration library of 30,000+ posts
Vilas Dhar's Recent LinkedIn Posts
Vilas Dhar
@vilasdhar
President, Patrick J. McGovern Foundation ($1.5B) | Global Authority on AI, Governance & Social Impact | Independent Director | Shaping Leadership in the Digital Age
Posts
Vilas Dhar
Uncategorized
3mo
We would never let five companies design the national highway system behind closed doors with no public oversight. But we are doing exactly that with AI.
AI is now the largest infrastructure buildout in American history. Capex totaled $427 billion last year and is projected at $562 billion this year. Five companies are building computational capacity that rivals mid-sized nations. If you hold an index fund, you are funding this buildout. Those companies and their peers now account for more than 40% of the S&P 500.
Congress held years of public hearings before authorizing highways. Before the Telecom Act. Before rural electrification. AI infrastructure is larger than all of them combined, and it has had none. Not one hearing on who it serves or what happens to the communities it bypasses.
The cost of skipping that step is already visible in the chaotic nature of adoption. PwC's 2026 CEO survey found that 56% of global executives report zero measurable productivity gains from AI. J.P. Morgan estimates the current infrastructure would require $650 billion in new annual revenue, in perpetuity, just to return 10%.
We are buying infrastructure without defining its architecture.
On Friday, Laurie Segall and I discussed this at the Commonfund Forum with 500 institutional investors as part of a speaker lineup that included Kenneth Rogoff, Platon, Admiral James Stavridis and Jen Easterly.
I lead a foundation that has deployed more than $500 million in AI capital across 30 countries. Across a decade of work, we've learned deployments that deliver national-scale results define who the technology is for before spending a dollar on infrastructure.
That pattern holds whether the investment is a $2 health tool in rural India or a $200 million enterprise integration. People-centered design is the decisive variable.
If AI infrastructure is going to function as public infrastructure, it needs public design standards. At a minimum:
→ Define public purpose before capital deployment. Large-scale systems should specify who they are built for and what outcomes they are meant to produce.
→ Attach obligations to scale. The greater the computational footprint, the greater the accountability and transparency.
→ Require distributional impact assessment. Major deployments should disclose who benefits, who bears risk, and which communities may be bypassed.
→ Align capital markets with public value. Investors should see not only projected returns, but the social assumptions embedded in those projections.
The architecture is being set now, and we could design it with the intent to determine who benefits so no one is left outside the system.
#AI #AIGovernance #Infrastructure #PublicPolicy #Technology #InstitutionalDesign The Patrick J. McGovern Foundation Mitsuru Tsukamoto Mark Anson Anthony Ialeggio Nicole Melwood
230
Vilas Dhar
Uncategorized
3mo
We would never let five companies design the national highway system behind closed doors with no public oversight. But we are doing exactly that with AI.
AI is now the largest infrastructure buildout in American history. Capex totaled $427 billion last year and is projected at $562 billion this year. Five companies are building computational capacity that rivals mid-sized nations. If you hold an index fund, you are funding this buildout. Those companies and their peers now account for more than 40% of the S&P 500.
Congress held years of public hearings before authorizing highways. Before the Telecom Act. Before rural electrification. AI infrastructure is larger than all of them combined, and it has had none. Not one hearing on who it serves or what happens to the communities it bypasses.
The cost of skipping that step is already visible in the chaotic nature of adoption. PwC's 2026 CEO survey found that 56% of global executives report zero measurable productivity gains from AI. J.P. Morgan estimates the current infrastructure would require $650 billion in new annual revenue, in perpetuity, just to return 10%.
We are buying infrastructure without defining its architecture.
On Friday, Laurie Segall and I discussed this at the Commonfund Forum with 500 institutional investors as part of a speaker lineup that included Kenneth Rogoff, Platon, Admiral James Stavridis and Jen Easterly.
I lead a foundation that has deployed more than $500 million in AI capital across 30 countries. Across a decade of work, we've learned deployments that deliver national-scale results define who the technology is for before spending a dollar on infrastructure.
That pattern holds whether the investment is a $2 health tool in rural India or a $200 million enterprise integration. People-centered design is the decisive variable.
If AI infrastructure is going to function as public infrastructure, it needs public design standards. At a minimum:
→ Define public purpose before capital deployment. Large-scale systems should specify who they are built for and what outcomes they are meant to produce.
→ Attach obligations to scale. The greater the computational footprint, the greater the accountability and transparency.
→ Require distributional impact assessment. Major deployments should disclose who benefits, who bears risk, and which communities may be bypassed.
→ Align capital markets with public value. Investors should see not only projected returns, but the social assumptions embedded in those projections.
The architecture is being set now, and we could design it with the intent to determine who benefits so no one is left outside the system.
#AI #AIGovernance #Infrastructure #PublicPolicy #Technology #InstitutionalDesign The Patrick J. McGovern Foundation Mitsuru Tsukamoto Mark Anson Anthony Ialeggio Nicole Melwood
230
Vilas Dhar
Uncategorized
3mo
We would never let five companies design the national highway system behind closed doors with no public oversight. But we are doing exactly that with AI.
AI is now the largest infrastructure buildout in American history. Capex totaled $427 billion last year and is projected at $562 billion this year. Five companies are building computational capacity that rivals mid-sized nations. If you hold an index fund, you are funding this buildout. Those companies and their peers now account for more than 40% of the S&P 500.
Congress held years of public hearings before authorizing highways. Before the Telecom Act. Before rural electrification. AI infrastructure is larger than all of them combined, and it has had none. Not one hearing on who it serves or what happens to the communities it bypasses.
The cost of skipping that step is already visible in the chaotic nature of adoption. PwC's 2026 CEO survey found that 56% of global executives report zero measurable productivity gains from AI. J.P. Morgan estimates the current infrastructure would require $650 billion in new annual revenue, in perpetuity, just to return 10%.
We are buying infrastructure without defining its architecture.
On Friday, Laurie Segall and I discussed this at the Commonfund Forum with 500 institutional investors as part of a speaker lineup that included Kenneth Rogoff, Platon, Admiral James Stavridis and Jen Easterly.
I lead a foundation that has deployed more than $500 million in AI capital across 30 countries. Across a decade of work, we've learned deployments that deliver national-scale results define who the technology is for before spending a dollar on infrastructure.
That pattern holds whether the investment is a $2 health tool in rural India or a $200 million enterprise integration. People-centered design is the decisive variable.
If AI infrastructure is going to function as public infrastructure, it needs public design standards. At a minimum:
→ Define public purpose before capital deployment. Large-scale systems should specify who they are built for and what outcomes they are meant to produce.
→ Attach obligations to scale. The greater the computational footprint, the greater the accountability and transparency.
→ Require distributional impact assessment. Major deployments should disclose who benefits, who bears risk, and which communities may be bypassed.
→ Align capital markets with public value. Investors should see not only projected returns, but the social assumptions embedded in those projections.
The architecture is being set now, and we could design it with the intent to determine who benefits so no one is left outside the system.
#AI #AIGovernance #Infrastructure #PublicPolicy #Technology #InstitutionalDesign The Patrick J. McGovern Foundation Mitsuru Tsukamoto Mark Anson Anthony Ialeggio Nicole Melwood
230