BlackRock's AI Sentiment

By analyzing BlackRock Investment Institute (BII) publications over the years alongside current portfolio allocations in the US-listed iShares A.I. Innovation and Tech Active ETF (BAI)[1] and the EU-listed iShares AI Innovation Active UCITS ETF,[2] we can estimate where we are in the capital inflow cycle. BAI is managed by Tony Kim and Reid Menge, who both have a track record in managing technology portfolios. Portfolio positioning indicates a shift away from the indiscriminate accumulation phase. The strategy has shifted toward physical bottlenecks, private market allocations, and a reduction in exposure to vulnerable software providers.

BlackRock’s Sentiment

Let’s look at BII’s shifting commentary on AI over the past three years.

In late 2023, BII published “Beyond the Buzz.”[3] BlackRock was measuring AI patent filings to forecast corporate earnings. The tone of the reports was relatively cautious. Analysts acknowledged the disruption potential but remained focused on cybersecurity risks and model hallucinations. The investment thesis was broad and exploratory.

By November 2024, the tone shifted to emphasize structural growth. While some market commentary drew comparisons to a dot-com redux, BII rejected these comparisons.[4] They argued the heavy concentration of market returns in a handful of mega-caps was a necessary feature of the intelligence buildout. Unlike the late 90s, BlackRock pointed out that tech leaders were generating massive free cash flow to fund their capital expenditures.

If a handful of companies were borrowing enough money to meaningfully impact long-term Treasury yields, BII argued that traditional broad diversification might be less effective. BII referred to this as the “diversification mirage,” warning that hiding in equal-weighted indices or long-dated bonds might fail to provide portfolio ballast.

A notable evolution in BlackRock’s sentiment occurred during their 2026 Outlook (published in late 2025). AI was upgraded from a technology sector theme to a dominant macroeconomic force in the global economy.[5] BII introduced the “Micro is Macro” framework. The capital expenditure of a few hyperscalers became large enough to impact traditional economic models. BlackRock estimated that AI investment was driving nearly half of all US GDP growth. BII observed that this force outweighed conventional macroeconomic headwinds at the time like the US-China tariff conflict and the late-2025 US government shutdown.

Hardware Rotation

Nvidia is no longer the top holding of BlackRock’s AI thesis. SK Hynix and Micron hold the top two spots in both the US and EU active portfolios, combining for roughly 13% of total assets. Nvidia sits lower down the roster at 4.2% (as of June 12th 2026).

BlackRock’s portfolio allocations suggest a view that compute has matured enough to shift pricing power directly to High Bandwidth Memory (HBM): SK Hynix crossing the $1 trillion market cap threshold in May[6] aligned with BlackRock’s assessment that HBM is a primary bottleneck for next-generation cluster deployments. Notably, SK Hynix is now reporting a 72% operational profit margin (Q1 2026).[7]

The portfolios also reveal a shift in the inference phase. AMD commands a higher weighting than Nvidia. BlackRock’s managers appear convinced that AMD’s MI-series architecture has captured market share from Nvidia during the initial training-heavy years of 2023 and 2024. Indeed, AMD’s Data Center segment revenue grew significantly in Q1 2026 (57% YoY to $5.8 billion).[8]

SaaSpocalypse

In February 2026, the BII noted a shift in the AI narrative from indiscriminate buying to identifying active threats to existing business models.[9]

AI agents were perceived to be eroding traditional SaaS competitive advantages. This triggered the severe software selloff earlier this year, dubbed the “SaaSpocalypse” (for Software as a Service). In BlackRock’s portfolios, legacy cloud and software providers like Hubspot and Atlassian are held at minor 1% allocations. The portfolios are instead heavily concentrated in power management, liquid cooling and optical transceivers, featuring outsized positions in companies like Vertiv, Lumentum and Asia Vital Components.

Subsequent market performance has largely reflected this thesis. Out of 15 noteworthy SaaS stocks checked, only three have risen YTD: Datadog, Snowflake and Cloudflare. What these three have in common is that they have actively integrated AI into their brand and service offerings. For Datadog and Snowflake it is featured on their websites front and center, and Cloudflare’s services have been expanded to cover AI crawler management and an AI model gateway (serving both AI adopters and AI critics). The remaining 12 SaaS tickers, which are mostly frontend services, have fallen 37% since the beginning of the year on average.

Physical Constraints and the “Leveraging Up” Theme

The AI thesis has evolved from theoretical model scaling to infrastructure constraints. BlackRock’s April 2026 publications emphasized land, permitting and power as the primary constraints on the $5 to $8 trillion global buildout.[10] The recent abandonment of the $550 million Sentinel data center in Maine over local political resistance serves as an example of the friction BlackRock warned about.

To bypass grid delays, hyperscalers are pivoting to “behind the meter” power generation. BII notes this geographic reality gives the US a structural advantage over Europe and Asia. Middle East conflicts have kept natural gas prices elevated abroad, while US domestic gas supplies provide a cheaper baseload for independent data center power.

Funding this infrastructure requires massive debt issuance. BlackRock identifies this as the “leveraging up” phase. Tech giants are front-loading capital expenditures and issuing bonds to bridge the gap until new native AI software revenues materialize. BlackRock is favoring private credit and infrastructure equity to finance this gap. They are actively avoiding long-term US Treasuries, warning that the massive corporate borrowing spree will keep upward pressure on term premiums.

Retail Access to Private Equity

The biggest contrast between BlackRock’s US and EU ETF listings lies in differing regulatory frameworks. US retail investors holding BAI are getting indirect exposure to foundation models: OpenAI and Anthropic shares. The European UCITS fund has no exposure to these private companies.

The US fund utilizes SEC Rule 22e-4, which permits holding up to 15% of the portfolio in illiquid assets. This hybrid ETF structure allows BlackRock to allocate hundreds of millions of dollars into late-stage private equity rounds. European investors are precluded from this exposure due to ESMA’s strict daily liquidity requirements.

Takeaway

The innovation cycle in AI is fast enough to genuinely warrant a reevaluation of passive indices as foundation for sector ETFs.