By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panelists agree that BlackRock's push into AI infrastructure is a strategic bet, but they differ on the timing, risks, and potential returns. While some see it as a durable, multi-year capex cycle, others caution about overbuilding for AI, operational complexity, and the risk of slower AI adoption or higher energy costs.
Risk: The single biggest risk flagged is the potential for slower AI adoption or higher energy costs to squeeze returns on investment and compress fees.
Opportunity: The single biggest opportunity flagged is the potential for BlackRock to earn management fees on trillions in AUM growth as hyperscalers invest in data centers and infrastructure.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Some people say artificial intelligence has gone too far.
For example, AI can now create realistic images and audio known as "deepfakes" that fool the public, chatbots can store information from "conversations" with users (1), and 80% of surveyed Gen Zers reported that they would marry an AI (2).
- Thanks to Jeff Bezos, you can now become a landlord for as little as $100 — and no, you don't have to deal with tenants or fix freezers. Here's how
- Robert Kiyosaki says this 1 asset will surge 400% in a year and begs investors not to miss this ‘explosion’
- Dave Ramsey warns nearly 50% of Americans are making 1 big Social Security mistake — here’s how to fix it ASAP
But Larry Fink, billionaire CEO of BlackRock [NYSE:BLK], argues that AI hasn't gone far enough. He believes that AI is the future, and he's putting his money where his mouth is.
"I don't believe we're moving fast enough," Fink said at the Milken Institute Global Conference on May 5, according to Business Insider (3). "There is not an AI bubble. There is the opposite."
At the Milken conference, Fink announced that BlackRock will partner with a hyperscaler to expand on AI infrastructure, including building data centers and investing in energy. Fink has not yet revealed the name of the hyperscaler, and BlackRock reportedly hasn't responded to Business Insider's request for comment.
Fink is a known AI advocate, and BlackRock has a history of investing in AI infrastructure. In 2024, the asset management giant acquired private market asset manager Global Infrastructure Partners for $12.5 billion (4).
In March 2025, BlackRock and Global Infrastructure Partners joined forces with MGX, Microsoft [NASDAQ:MSFT], Nvidia and xAI to invest in data centers (5).
"AI infrastructure will play an increasingly critical role in driving economic growth across every industry and every region of the world," Satya Nadella, chairman and CEO of Microsoft, said in the BlackRock press release (6). "We're thrilled to welcome these new companies to the AI Infrastructure Partnership as we invest together to build the infrastructure of the future (7)."
Fink's words — and actions — express that he does not believe in an AI bubble and that he's not worried about such a bubble popping. But many investors and industry experts disagree with his stance.
Four tech hyperscalers — Alphabet, Amazon, Meta and Microsoft — plan to spend more than $650 billion on AI in 2026 (8). But OpenAI has reported $25 billion in annualized revenue (9), while Anthropic claims its annualized revenue run rate is over $30 billion (10).
Four leading AI models discuss this article
"The investment thesis for AI has shifted from software application growth to the physical energy and real estate constraints of the hardware stack."
Larry Fink’s pivot from asset management to infrastructure development via the GIP acquisition is a structural shift, not just a thematic bet. By integrating power generation and data center real estate, BlackRock is betting on the 'physicality' of AI—the bottleneck isn't software, it's the grid. While hyperscalers like MSFT and GOOGL face massive CapEx pressure, BlackRock is capturing the 'picks and shovels' of this build-out. However, the $650 billion spending projection cited in the article is speculative and ignores the potential for a 'utility trap' where energy costs outpace the monetization of AI models, compressing margins for the entire tech sector.
The 'infrastructure' play may simply be a way for BlackRock to offload long-term, illiquid energy assets onto institutional clients while masking the slowing adoption rates of generative AI software.
"BLK's GIP integration and AI infra partnerships could accelerate AUM growth to 15%+ annually on hyperscaler capex tailwinds."
Larry Fink's 'opposite of AI bubble' claim highlights BlackRock [BLK]'s strategic bet on infrastructure via its $12.5B Global Infrastructure Partners acquisition and March 2025 partnership with Microsoft [MSFT], Nvidia [NVDA], xAI, and MGX for data centers. Hyperscalers' $650B+ 2026 capex (Alphabet, Amazon, Meta, MSFT) signals multi-year demand, positioning BLK to earn 1-2% management fees on trillions in AUM growth—potentially 15%+ annually if utilization hits 80%. Energy investments address grid strains, with U.S. data center power needs projected to double to 35GW by 2030 (EIA data). This diversifies BLK beyond ETFs into high-margin infra.
Hyperscalers' $650B capex dwarfs AI firm revenues ($25B OpenAI, $30B Anthropic run-rates), risking overbuild if ROI disappoints and utilization lags, hammering BLK's infra assets.
"A 12:1 ratio of capex to current revenue suggests the market is pricing in speculative future demand, not proven business models—and BlackRock's involvement signals even asset managers are now funding the infrastructure gap that tech companies themselves won't fully finance."
Fink's 'opposite of a bubble' framing is rhetorically clever but empirically weak. The capex math is damning: $650B spent by four hyperscalers against ~$55B combined annualized revenue from OpenAI and Anthropic suggests a 12:1 ratio of infrastructure investment to actual revenue generation. That's not confidence—it's a bet that monetization models don't yet exist at scale. BlackRock's $12.5B GIP acquisition and partnership with Microsoft/Nvidia/xAI signals real capital deployment, but also reveals the problem: asset managers are now funding infrastructure because tech companies alone can't justify the ROI. The article omits critical context: utilization rates on existing data centers, power grid constraints, and whether demand for compute will materialize fast enough to absorb this supply.
Fink may be right that we're *under*-investing relative to long-term AI's potential—if transformative AGI arrives in 5-10 years, today's $650B capex looks cheap in hindsight. The revenue gap could simply reflect that killer applications (autonomous systems, scientific discovery) haven't monetized yet, not that they won't.
"The multi-year AI data-center capex cycle will be a durable growth driver for hyperscalers' top lines and cloud margins, supported by AI workloads scaling and efficiency gains."
The article frames Fink's AI stance as the anti-bubble bet and highlights BlackRock's push into AI infrastructure alongside Microsoft, Nvidia, and others. That signals a durable, multi-year capex cycle rather than a short-lived craze. The strongest bullish read is that hyperscalers monetize AI by expanding data-center capacity, custom accelerators, and energy efficiency, which should lift long-term revenue growth and cloud margins. But the piece glosses over key risks: energy and power costs, cooling bottlenecks, supply-chain tightness, and the risk that AI workloads grow slower or shift to open-source solutions, depressing ROI. Regulatory and geopolitical risks also loom, and near-term stock returns hinge on execution of multi-year capex plans.
Counterpoint: The AI infrastructure wave could be a perpetual capex treadmill with diminishing returns; if AI demand slows or power costs spike, the huge build-out won’t translate into commensurate cash flow, and valuations could compress.
"The 12:1 capex ratio is misleading because it ignores the massive, non-AI legacy enterprise cloud workloads that these new data centers will inevitably absorb."
Claude highlights the 12:1 capex-to-revenue ratio, but misses the 'platform' nature of this spend. Hyperscalers aren't just building for current AI revenue; they are replacing legacy enterprise IT stacks. If MSFT and GOOGL successfully migrate existing cloud workloads to these new, more efficient data centers, the ROI isn't just 'AI'—it's total infrastructure modernization. The risk isn't overbuilding for AI; it's the operational complexity of integrating these energy-intensive assets into aging grids.
"Hyperscalers' existing cloud FCF funds capex, making narrow AI revenue comparisons misleading."
Claude's 12:1 capex-to-AI-revenue ratio ignores hyperscalers' $200B+ annual cloud revenues (MSFT Azure ~$80B TTM, AMZN ~$100B) generating ample FCF for the $650B spend over 3-5 years—it's not a pure AI gamble. Gemini's modernization point strengthens this: BLK's GIP play monetizes total infra demand. Unflagged risk: rising interest rates could squeeze BLK's 20x P/E if fee growth lags.
"The $650B capex conflates legacy cloud modernization with speculative AI workload growth—conflating the two obscures the real ROI question."
Grok's cloud-revenue reframe is stronger than Claude's pure-AI-capex framing, but both miss the timing mismatch. MSFT Azure's $80B is mature, declining-margin business. The $650B spend targets *new* AI workloads with unproven unit economics. Modernizing legacy infrastructure doesn't require this scale—it's a bet on AI adoption acceleration. If that stalls, BLK's 20x P/E becomes indefensible, and Grok's interest-rate sensitivity risk compounds.
"BLK's infra bet is highly sensitive to utilization and energy costs; if AI demand stalls or power costs rise and rates climb, ROIC and fee income may disappoint."
Picking up Grok's ROI framing, the real fragility is reliance on a 80%+ utilization path and benign energy costs. The 650B capex vs. $200B cloud revenue underestimates the sensitivity to power prices, cooling, and grid constraints; a slower AI adoption or higher rates could squeeze ROIC and compress fees, especially if utilization lags. The bear case isn't that infra is worthless, but that the timing and cost backdrop could derail the boom.
The panelists agree that BlackRock's push into AI infrastructure is a strategic bet, but they differ on the timing, risks, and potential returns. While some see it as a durable, multi-year capex cycle, others caution about overbuilding for AI, operational complexity, and the risk of slower AI adoption or higher energy costs.
The single biggest opportunity flagged is the potential for BlackRock to earn management fees on trillions in AUM growth as hyperscalers invest in data centers and infrastructure.
The single biggest risk flagged is the potential for slower AI adoption or higher energy costs to squeeze returns on investment and compress fees.