Sign up to save your podcasts
Or
Share The Power Allocation
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
The Power Allocation episodes
- Liquid Cooling Hits 40% of AI Data CentersLiquid cooling is no longer optional for AI infrastructure — it's the baseline requirement. The data center cooling market is projected to reach $31 billion by 2034. This episode examines the technologies, infrastructure implications, and heat recovery opportunities.
Key topics covered:
- 40% of AI data centers deploying liquid cooling in 2026
- Cooling market growing from $11.65B to $31B by 2034
- Direct-to-chip, immersion, and CDU-based systems
- AI-driven thermal management systems
- Microsoft-Fortum heat recovery for district heating
Related keywords: liquid cooling data centers, immersion cooling AI, direct-to-chip cooling, data center thermal management, heat recovery data centers
Brought to you by Spring Street Management Group.
]]>3 min - The Permitting Problem Is Getting WorseData center permitting timelines have tripled in legacy markets, and community opposition has blocked billions in projects. This episode examines the permitting landscape, regulatory fragmentation, and strategies for navigating local politics.
Key topics covered:
- Permitting extending from 6-12 months to 2-3 years
- New York moratorium bill for 20+ MW data centers
- Community opposition and misconceptions
- State and local regulatory fragmentation
- Brownfield guidance and federal expediting efforts
Related keywords: data center permitting, data center moratorium, community opposition data centers, data center zoning, permitting delays AI infrastructure
Brought to you by Spring Street Management Group.
]]>3 min - Meta Compute and the 5-Gigawatt CampusMeta has created Meta Compute to centralize and accelerate its AI infrastructure buildout. With Prometheus in Ohio and Hyperion scaling to 5 gigawatts in Louisiana, Meta is building at scales that rival utility infrastructure projects. This episode examines Meta's compute strategy and its implications.
Key topics covered:
- Meta Compute division reporting directly to Zuckerberg
- Prometheus supercluster activating in Ohio 2026
- Hyperion campus scaling to 5 GW in Louisiana
- $135 billion AI investment potential in 2026
- Llama 4 and distributed inference requirements
Related keywords: Meta Compute, Meta data centers, Hyperion Louisiana, Prometheus Ohio, Meta AI infrastructure, Zuckerberg compute
Brought to you by Spring Street Management Group.
]]>5 min - Fuel Cells Just Became a $5 Billion BusinessFuel cell adoption has moved from small-scale pilots to multi-billion-dollar primary power deployments for AI data centers. This episode examines the technology, business models, and market shift driving fuel cells as a time-to-power solution.
Key topics covered:
- Bloom Energy's $5 billion Brookfield partnership
- $2.65 billion American Electric Power agreement
- Solid oxide fuel cell efficiency advantages
- ECL's 1 GW hydrogen-powered AI Factory in Texas
- Hydrogen as storage medium for 24/7 carbon-free energy
Related keywords: fuel cells data centers, Bloom Energy Brookfield, solid oxide fuel cells, hydrogen data centers, on-site power generation AI
Brought to you by Spring Street Management Group.
]]>3 min - The Copper Crisis No One Is Pricing InAI-driven data centers require about four times more copper than traditional facilities. With a global deficit forecast and prices projected to hit $12,500-$13,000 per ton, copper is becoming a critical constraint on AI infrastructure. This episode examines the copper supply chain and its implications for investors.
Key topics covered:
- AI data centers requiring 4x copper per facility
- Global refined copper deficit of 150,000 metric tons in 2026
- JPMorgan and UBS copper price projections
- Strategic stockpiling and blanket agreements
- Aluminum as partial substitute for busway systems
Related keywords: copper supply data centers, copper deficit AI, copper price forecast, data center materials, copper demand AI infrastructure
Brought to you by Spring Street Management Group.
]]>3 min - The $690 Billion Capex Arms RaceFive tech giants are deploying nearly $700 billion in a single year to build computing infrastructure — more than the GDP of most countries. This episode examines the capital deployment driving the AI buildout and its downstream effects on utilities, construction, and materials.
Key topics covered:
- Amazon leading with $200 billion capex
- Alphabet nearly doubling to $175-185 billion
- Meta Compute targeting $115-135 billion
- Microsoft Azure expansion at $110-145 billion
- Supply chain strain from unprecedented capital deployment
Related keywords: hyperscaler capex, AI infrastructure spending, Amazon AWS data centers, Meta Compute, Microsoft Azure expansion, tech capital expenditure 2026
Brought to you by Spring Street Management Group.
]]>3 min - PJM Just Authorized Redirecting Data Center Power to HomesPJM—the grid operator for 13 states—now has authorization to redirect power from data centers to households during grid emergencies. This episode examines what this policy shift means for data center operators and the evolving relationship between AI infrastructure and grid reliability.
Key topics covered:
- PJM's new authority to require data centers to switch to backup generators
- The blurring line between backup power and dispatchable generation
- Large load tariffs and demand response provisions for data centers
- Google and Fervo Energy's clean transition tariff with NV Energy
- Implications for backup generator maintenance, fuel contracts, and emissions permits
Related keywords: PJM data centers, grid reliability, demand response, backup generators, large load tariff, data center curtailment, Google Fervo geothermal
The Power Allocation is brought to you by Spring Street Management Group.
]]>3 min - Blackwell Changes Everything About Data Center DesignNVIDIA's Blackwell GPUs demand 10-30x the power density of traditional data centers, rendering most existing facilities obsolete for AI workloads. This episode breaks down the infrastructure overhaul required to support next-generation AI hardware.
Key topics covered:
- Blackwell rack power requirements: 60-140 kilowatts vs. traditional 4-6 kilowatts
- Why fewer than 5% of data centers can support 50+ kW racks
- Liquid cooling as a hard requirement, not an option
- Electrical infrastructure overhaul: transformers, PDUs, bus bars, and wiring
- 400-800 Gbps networking with Quantum-X800 InfiniBand
Related keywords: NVIDIA Blackwell, data center power density, liquid cooling AI, NVL72, high-density computing, AI infrastructure design, colocation AI
The Power Allocation is brought to you by Spring Street Management Group.
]]>3 min - Why Texas Will Have 40 Gigawatts of Data Centers by 2028Texas is on track to reach 40 gigawatts of data center capacity by 2028—a 142% increase from current levels and more than the total capacity of most countries. This episode examines why the Lone Star State is winning the race for AI infrastructure investment.
Key topics covered:
- ERCOT's regulatory advantages over FERC-jurisdictional regions
- OpenAI's 1.2GW Stargate data center in Milam County with SB Energy
- Galaxy's 1.6GW Helios campus expansion in West Texas
- West Texas oil patch transformation into a data center corridor
- Co-location of gas generation with compute to eliminate grid dependencies
Related keywords: Texas data centers, ERCOT, Stargate Texas, SB Energy OpenAI, Galaxy Helios, West Texas AI infrastructure, data center permitting
The Power Allocation is brought to you by Spring Street Management Group.
]]>3 min - The Labor Crisis Threatening the AI BuildoutThe AI infrastructure buildout faces a constraint that capital alone cannot solve: a severe shortage of skilled construction labor. This episode explores how 82% of construction firms report difficulty filling positions, and why projects are being postponed or canceled despite secured funding and permits.
Key topics covered:
- Peak crew sizes jumping from 750 to 4,000-5,000 workers per site
- 82% of firms struggling to fill hourly craft positions
- 60% of projects postponed or canceled due to labor constraints
- Worker migration from Arizona to Dallas as markets shift
- Competition for electrical workers from EVs, heat pumps, and grid upgrades
Related keywords: data center construction, skilled labor shortage, electrical workforce, construction megaprojects, AI infrastructure labor, DataBank Red Oak
The Power Allocation is brought to you by Spring Street Management Group.
]]>3 min
About The Power Allocation
From the publisher's feed
The AI boom isn't constrained by chips, algorithms, or talent. It's constrained by electricity.
The Power Allocation is a daily briefing on AI infrastructure — where capital is actually being deployed. Each episode cuts through the hype to examine the physical realities shaping the AI buildout: power constraints, grid interconnection, land acquisition, data center financing, cooling infrastructure, and utility relationships.
This isn't a software podcast. This is an infrastructure podcast.
Who it's for: Institutional investors, infrastructure allocators, data center developers, utilities, family offices, and anyone positioning capital for the physical layer of artificial intelligence.
What you'll learn:
Why power availability — not GPU supply — is the binding constraint on AI compute
How hyperscalers are locking in multi-decade power purchase agreements
Where data centers are relocating and why grid geography is reshaping the industry
The financing structures turning compute facilities into bond-like assets
What execution timelines, permitting delays, and interconnection queues mean for capital deployment
Format: 3-6 minute episodes. Dense. Clear. No hype.
New episodes daily. Subscribe wherever you listen.