Morgan Stanley: AI Infrastructure Requires $10 Trillion, Capital Not Bottleneck, Power is Key

Overview

Morgan Stanley believes AI infrastructure investment cycle scale reaches $10 trillion, but capital is not the limiting factor. Global $256 trillion capital pool is sufficient to cover demand; core challenge lies in how to match financial structures to different investor preferences. Power supply and compute capacity are real bottlenecks; alternative asset management and banks will significantly benefit from financing ecosystem expansion.

Core views

Capital Adequacy: The report breaks down six capital sources, including $256 trillion asset owner capital (insurance 42T, pensions 42T, sovereign wealth funds 16T), $17 trillion annual debt issuance capacity, $2.6 trillion equity financing capacity, $4.9 trillion private market dry powder, $2.5 trillion bank balance sheet capacity, and $7.6 trillion money market liquidity. AI infrastructure spending accounts for only 4% of asset owner capital, no need to liquidate existing holdings on a large scale. Financing Structure Evolution: Early stage borne by hyperscale enterprise balance sheets, future will expand to diverse channels. Investment grade credit can finance stable infrastructure (e.g., data center ABS/CMBS), private credit covers early projects, equity markets support growth capital. 2025 global corporate bond issuance $13.3 trillion (43% of GDP), if peak levels resume could reach $17 trillion; equity issuance $1 trillion, if peak levels resume could reach $2.6 trillion. Bottleneck Shift: Power shortage is the core constraint, US data center demand expected to exceed grid capacity continuously before 2028. Compute aspect, advanced chip utilization near满载, inference load increases grid volatility. Enterprise adoption faces data preparation, talent shortage challenges. Beneficiary Targets: Alternative asset managers (APO, BX, KKR, BN, BLK) hold advantage with infrastructure platforms and dry powder size; Banks (BAC, C, GS, JPM, WFC) capture financing needs through investment banking and trading businesses, Citigroup listed as top pick due to AI infrastructure banking team and cross-border capabilities.

Analysis framework

Institutions adopt capital supply-demand framework analysis: first quantify total AI infrastructure demand ($10 trillion), then break down global capital supply sources and capacity. Highlight scale of this round via historical comparison (e.g. mobile/cloud cycle $1 trillion capex). Financial structure analysis focuses on term, liquidity, risk preference matching across different capital pools, e.g. insurance funds suitable for long-duration credit, pension funds can allocate equity and infrastructure. Bottleneck analysis expands from physical limits (power, chips) and enterprise adoption barriers (data, security), distinguishing capital constraints from physical constraints.

Methodology notes

• Industry/Industrial Analysis FrameworkSupply-demand framework

  Capital Supply Demand Analysis

  The report proves capital adequacy by comparing AI infrastructure capital demand ($10 trillion) with global capital supply ($256 trillion). This method helps investors understand financial feasibility of industry expansion.

• Company Fundamentals and Financial FrameworkFree cash flow analysis

  Cash Flow Matching Financing Structure

  The report emphasizes different capital pools need to match asset cash flow characteristics (e.g. insurance funds prefer stable returns), this method reveals key logic of financing structure design.

• Cycle and Prosperity FrameworkCapacity/Equipment Cycle (Juglar)

  Technology Cycle Capital Expenditure

  Analogize AI infrastructure to historical technology cycles (e.g. mobile internet 10x compute expansion), helping position current investment stage and duration.

• Industry/Industrial Analysis FrameworkUpstream-Midstream-Downstream Industry Chain Transmission

  Financing Ecosystem Chain Analysis

  The report breaks down financing chain from hyperscale enterprises to private capital, showing how capital allocates along industry chain layers, applicable for understanding complex infrastructure financing.

• Valuation methodsSOTP Sum-of-the-Parts Valuation

  Business Segment Capital Allocation

  Evaluate capacity separately for different financing channels (equity, credit, private market), this method helps identify investment opportunities in each sub-segment.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• APO (OW)

  Benefit from credit-intensive AI infrastructure financing

  Strengths

  Insurance capital, long-duration credit capability, structured solution platform

  Comparison

  Superior to peers in IG credit and asset-backed financing areas

  Risks

  Credit market volatility affects financing costs

• BX (OW)

  Benefit from data center and power exposure

  Strengths

  Natural gas power generation investment supports data center development, institutional vehicle demand

  Comparison

  Leading in infrastructure real estate sector

  Risks

  Power project execution risk

• KKR (OW)

  Most direct AI infrastructure investment channel

  Strengths

  Full-chain coverage of infrastructure, private credit, physical assets

  Comparison

  Dry powder size and data center platform superior to peers

  Risks

  Project returns below expectations

• C (OW, Top Pick)

  Benefit from AI infrastructure investment banking and cross-border financing

  Strengths

  Specialist AI infrastructure banking team, global network, increased investment banking share

  Comparison

  Cross-border capability superior to regional banks

  Risks

  Investment banking income volatility

• BAC (OW)

  Benefit from enterprise client financing needs

  Strengths

  Deep enterprise client base, $25 billion private credit commitment

  Comparison

  Mid-market coverage superior to large banks

  Risks

  Credit quality deterioration

Key data

• Total AI Infrastructure Demand~$10 TrillionEstimated total capex for this cycle
• Global Asset Owner Capital~$256 TrillionCovers retail and institutional capital pools
• Public Debt Annual Capacity~$17 TrillionAssuming peak issuance levels resume
• Public Equity Annual Capacity~$2.6 TrillionAssuming peak issuance levels resume
• Private Market Dry Powder~$4.9 TrillionUndeployed capital as of March 2026
• Bank Balance Sheet Capacity~$2.5 TrillionDerivation from excess capital of US large banks
• Data Center Securitization Credit~$65B → $200BExpected to grow 3x by 2028
• Hyperscale Enterprise Capex2026 $805B → 2027 $1.116TRevised up 5%/17% after 1Q26

Impact & implications

For Asset Managers: Alternative asset managers (APO, BX, KKR etc.) can capture data center, power-related financing needs凭借 infrastructure platforms and dry powder size. For Banks: Investment banking and trading businesses will benefit from AI infrastructure financing activity, Citigroup listed as top pick due to cross-border capabilities and specialist teams. For Market Structure: Equity issuance recovery may reduce concentration (history shows IPO volume negatively correlated with top head concentration), but tech weighting already at high level (S&P 500 Tech + Comm + Amazon account 50%).

Risks

• Power and grid capacity shortages delay data center deployment
• Permit delays, labor shortages, and equipment supply chain bottlenecks
• Enterprise AI monetization below expectations reduces capital expenditure motivation
• Model efficiency improvements reduce compute intensity demand
• Regulatory or geopolitical headwinds affect investment rhythm
• Infrastructure overbuilding leads to oversupply capacity

What to watch

• Hyperscale enterprise capex guidance changes
• Data center securitization credit issuance growth rate
• Grid expansion progress and permit approval speed
• Private market secondary market liquidity improvement status
• IPO market recovery impact on equity concentration