Where Capital Flows—and Where It Doesn't
SIGNALS BRIEF #2 | Week of January 19, 2026 (2026.01.19)
This week’s signals confirm what we’ve been tracking: AI infrastructure capital is flowing at unprecedented scale while climate finance for frontline communities remains structurally starved. The infrastructure gap isn’t subtle—it’s systematic.
BY THE NUMBERS
$50 billion — Anthropic’s new investment in US data centers
500,000 — Trainium2 chips in AWS Project Rainier (deployed in under 12 months)
$226 million — Electric manufacturing sector lobbying spend in 2025 alone
80% — Share of climate solutions delivered by SMEs in emerging markets
<10% — Share of climate finance flows reaching those same SMEs
AI INFRASTRUCTURE SIGNALS
Meta doubles down on compute sovereignty
Mark Zuckerberg announced Meta Compute—a top-level initiative to build “tens of gigawatts” of AI capacity this decade. The move formalizes infrastructure as core strategy, with dedicated executives managing capacity planning separate from operations. Meta’s $72B capital spend in 2025 positions AI infrastructure as business-critical, not back-office. (Source: Axios, January 12, 2026; TechCrunch, January 12, 2026)
Anthropic goes vertical
The $50B data center announcement (Texas, New York, more locations coming) marks Anthropic’s shift from model developer to infrastructure player. Partnership with Fluidstack targets “gigawatts of power” delivery at speed. Combined with AWS Project Rainier’s 500,000 Trainium2 chips (operational October 2025) and expanded Google Cloud deals, Anthropic is diversifying computing power while ensuring capacity for 300,000+ business customers. (Source: Anthropic, November 12, 2025; Data Center Dynamics, October 30, 2025)
Lobbying spend tracks infrastructure growth
Amazon, Microsoft, Google, and Meta reported tens of millions in federal lobbying during 2025. Meta hired 21 additional lobbyists this year; OpenAI scaled from 3 (2023) to 18 (2024). The Data Center Coalition tripled quarterly spending from $123K to $360K. The electric manufacturing sector poured $226M into grid access advocacy—on pace for its largest lobbying year ever. (Source: OpenSecrets, November 5, 2025)
CLIMATE FINANCE SIGNALS
The deployment disconnect
At COP30, Global Climate Finance Forum founder Marilyn Waite framed the problem clearly: “SMEs are leading the way across climate solutions, and yet they are left out of accessing the capital needed to scale.” Despite delivering 80% of locally-developed climate solutions in emerging markets, these enterprises receive less than 10% of climate finance flows. Standard Chartered’s Dana Barsky noted SMEs “do not usually fit in a typical risk profile of a global bank”—highlighting structural barriers, not absence of viable projects. (Source: ImpactAlpha, December 2, 2025)
The intermediary question
Green Banks, CDFIs, and community-based organizations exist precisely to bridge capital deployment gaps. Yet the intermediary infrastructure remains dramatically undercapitalized relative to both need and the scale of capital being deployed in other sectors. The question isn’t whether these institutions can deploy capital effectively—many have proven track records. The question is what prevents them from operating at the scale the climate challenge requires.
Building the connective infrastructure
Justice Climate Fund CEO Amir Kirkwood’s year-end message offers a counterpoint to the infrastructure gap. JCF’s network of 400+ community lenders, green banks, and local partners is building the ecosystem for equitable climate finance deployment. Their partnership with Ceres brought community lenders and investors together for “candid conversations on scaling private investment into community climate and clean energy finance.” JCF launched the Climate Opportunity & Resilience Fund and developed impact measurement frameworks with Sorenson Impact Institute. This is what capital deployment infrastructure looks like when built intentionally for community-scale resilience. (Source: Justice Climate Fund, December 18, 2025)
WHAT THIS MEANS
AI infrastructure deployments happen at unprecedented speed—$50B commitments, 500,000 chips operational in under a year, coordinated policy advocacy across multiple levels of government. The capital flows because the infrastructure to deploy it exists: dedicated coalitions, systematic advocacy, multi-jurisdictional negotiation.
Climate resilience capital faces the opposite problem. The projects exist—SMEs deliver 80% of climate solutions in emerging markets. The need exists—communities require $1.3 trillion in climate finance. What’s missing isn’t viable projects or available capital. What’s missing is the deployment infrastructure: the intermediaries, the policy advocacy, the systematic pathways that connect capital to communities.
The gap isn’t about technology or economics. It’s about infrastructure—who builds it, who funds it, and whose needs it serves.
ON MY RADAR
The global race for AI dominance shapes infrastructure decisions
President Trump attends the World Economic Forum in Davos this week for the first time since 2020. BlackRock CEO Larry Fink, a key convener at Davos this year, is simultaneously financing Meta’s $27B private-debt deal for a Louisiana data center complex—the largest private-debt transaction in U.S. history—holding $3B in bonds to fund Hyperion, the largest data center ever built for Meta Platforms that occupies 2,700 acres of farmland in Louisiana. The gathering reveals how global AI competition is reshaping capital allocation priorities. (Source: TIME, January 20, 2026; Wall Street Journal, October 21, 2025; Wall Street Journal, March 30, 2025)
Agentic commerce: innovation meets accountability
Marc Benioff’s TIME essay positions “Agentic Enterprise” as defining the next decade—AI, data, apps, and people working as unified systems. Salesforce deployed Agentforce at Davos to manage attendee schedules and networking. Meanwhile, Consumer Reports’ Delicia Reynolds Hand asks whether we’re “automating errors at machine speed” without building consumer protection infrastructure first. The juxtaposition reveals opportunity: how do we build systems that enable innovation while protecting consumers? Both goals are achievable—the question is whether we’re building the connective tissue between capability and accountability. (Source: TIME, January 15, 2026; Open Banker, January 6, 2026)
Grid access rules are taking shape
In an unusual federal intervention, the U.S. Energy Secretary directed the Federal Energy Regulatory Commission to develop rules for large load grid connections—traditionally a state regulatory arena. The Commission proposed 100% “participant funding,” requiring large loads to cover full grid upgrade costs rather than socializing them across ratepayers. The final rules are expected on April 30.
A record $226M lobbying spend in 2025 reveals how large-scale infrastructure gets built. The electric manufacturing sector—including Microsoft, Oracle, and data center operators—coordinated across federal and state levels to address grid access bottlenecks. Their argument: expedited connections are essential for U.S. global competitiveness, economic growth, and the carbon-free energy transition. With data centers driving 22% power demand growth in 2025 (projected to triple by 2030), the sector positions infrastructure investments as critical to digital economy strength and national security.
States and utilities raised concerns about grid reliability, ratepayer protection, and who bears infrastructure costs. The policy response varied: Ohio, Virginia, and West Virginia enacted rules requiring data centers to fund minimum infrastructure costs or provide dedicated generation. California moved in the opposite direction; consumer advocates report that legislation protecting ratepayers was “successfully gutted” by industry lobbying.
The central question: should infrastructure costs be distributed across all users or paid directly by those creating new demand? Or is there some other way?
(Source: CalMatters, December 2025; Engineering News-Record, December 2025; OpenSecrets, November 5, 2025)
WHAT THIS MEANS
As the U.S. has withdrawn from major climate initiatives, tech leaders continue engagement with the administration on AI infrastructure priorities and regulation. The framing is explicit: this is a race for global technological dominance, with implications for which infrastructure needs get prioritized in policy—and whose resilience gets funded (or doesn’t).
The $226M spend wasn’t just lobbying—it funded coordinated policy advocacy through the Data Center Coalition, systematic grid access negotiation across multiple jurisdictions, multi-level government engagement (federal, state, local), and 53% of sector lobbyists being former government officials. This is what capital deployment infrastructure looks like when it exists.
Climate resilience capital for frontline communities operates without equivalent infrastructure. No $200M+ coordinated advocacy. No systematic intermediary coalition spanning 50 states. No former officials-turned-advocates negotiating capital access pathways. The contrast isn’t about whether AI infrastructure should be built—it’s about recognizing what makes any large-scale capital deployment possible. Infrastructure exists when we choose to build it.
SHOW ME WHAT’S WORKING
Infrastructure exists when we choose to build it. Justice Climate Fund’s 400+ lender network and Ceres partnership proves intermediary infrastructure can be built intentionally. What other success stories deserve visibility?
For my February 4 deep dive on capital flows: Which institutions are bridging the gap? What conversations have you seen where capital and community are actually connected? What made those moments work?
This is live research. Help me find the patterns.
Send me a DM or share in the comments.
ABOUT THIS BRIEF: METHODOLOGY
Compound Impact uses a combination of human analysis and AI research assistance (Claude) to surface signals, analyze datasets, and test frameworks. I write, select, synthesize, and stand behind every conclusion. The Three Gaps framework, convergence thesis, and editorial voice are mine. Claude helps me do what I do, faster: find patterns in complexity.