The AI opportunity is moving from chips to capacity — and that may change who benefits
For the past several years, mega-cap technology companies have dominated the AI investment narrative. They built the platforms, trained the models, and delivered the returns. But as AI spending accelerates into its next phase, the investment opportunity appears to be broadening in a direction many allocators haven't fully appreciated: the physical infrastructure required to bring compute online.
Polen Capital's white paper, From Compute to Capacity: A Real SMID Opportunity, makes the case that the AI capex cycle is creating a fundamental earnings-cycle opportunity for small- and mid-cap companies — not just a valuation mean-reversion argument.
Key Takeaways
- AI capex is flowing into the real economy.1 Approximately half of real U.S. GDP growth in 2025 was attributable to AI spending, now reaching into land, power, and construction.
- SMID companies are the capacity builders. Our analysis shows the same $1B revenue opportunity could drive 20–44% stock-price impact for a SMID company vs. ~1.6% for a large cap.
- Valuations are not stretched. SMID allocations have declined for two decades, and fund flows remain depressed relative to history.
1Barclays Americas Tactical Playbook, 4Q25: Riding the AI Wave (December 2025)
How big is AI infrastructure spending, and why does it matter for the real economy?
The scale of AI capital expenditure is now macroeconomically significant. The five largest hyperscalers — Alphabet, Amazon, Microsoft, Oracle, and Meta — are collectively spending hundreds of billions of dollars annually on data centers, chips, networking, power systems, and supporting infrastructure.
~50%
of real U.S. GDP growth in 2025 attributable to AI spending 1
~1%
data center vacancy rate, with pipeline already pre-committed 2
According to research from Barclays, approximately half of real U.S. GDP growth in 2025 was attributable to AI spending. That spending is no longer confined to semiconductor supply chains. It is flowing into land acquisition, substations, transformers, switchgear, generators, cooling systems, fiber connectivity, construction labor, and grid interconnection — turning AI capex into real-economy capex.
This transmission mechanism is what creates the bridge from mega-cap technology budgets to a potential small- and mid-cap opportunity. The hyperscalers are the source of capital; SMID companies may be the source of capacity.
1Barclays Americas Tactical Playbook, 4Q25: Riding the AI Wave (December 2025)
2 Source: As of February 2026. Intex, JLL Research. Note: vacancy reflects leased and hyper-owned inventory.
What are the biggest bottlenecks in the AI data center buildout?
The binding constraints are increasingly physical and local rather than technological:
Power and grid access
Data center demand is arriving faster than reliable new supply can be added. Interconnection queues stretch multiple years, pushing operators toward behind-the-meter solutions.
Skilled labor
Every gigawatt of new capacity requires thousands of workers across electrical, mechanical, plumbing, construction, and engineering. The supply of qualified workers is barely growing.
Electrical and thermal infrastructure
The majority of fully loaded data center investment flows outside the GPU and CPU complex into networking, power distribution, backup power, thermal management, land, and buildings.
These bottlenecks create multi-year scarcity rather than near-term overbuild risk. Data center vacancy hovers around 1%, and a significant share of the development pipeline is already pre-committed. 2
2 Source: As of February 2026. Intex, JLL Research. Note: vacancy reflects leased and hyper-owned inventory
Why might small- and mid-cap stocks benefit more than large caps?
The portfolio construction implication is subtle but significant. Large-cap indices are heavily concentrated in the companies funding AI capex — the hyperscalers and dominant semiconductor providers. Small- and mid-cap indices, by contrast, have greater representation in the industrial, electrical, construction, materials, energy, and infrastructure-related businesses that execute the physical buildout.
"A hypothetical $1 billion of incremental AI infrastructure revenue is negligible for a mega-cap company but potentially transformational for a SMID business."
Are investors already paying peak prices for this exposure?
Generally, no — and that's a key part of the thesis. SMID allocations have generally declined for two decades. Valuations sit at a wide discount to large caps, and fund flows into SMID strategies remain depressed relative to history. Investors are not being asked to pay a peak-enthusiasm price for exposure to the AI infrastructure buildout.
If earnings growth accelerates for the companies enabling the physical AI buildout, SMID stocks may benefit from both fundamental improvement and multiple recovery — a combination that is more difficult to achieve in the large-cap space where expectations and positioning are already elevated.
What does Polen Capital conclude?
The first phase of AI rewarded ownership of compute. The next phase may reward ownership of capacity — the physical, electrical, mechanical, networking, and labor-intensive infrastructure required to bring compute online.
That shift creates what Polen Capital believes is a far more favorable environment for SMID companies than investors have seen in many years. The AI opportunity is broadening: from models to data centers, from chips to power, from software to infrastructure, and from capital-light platforms to capacity-constrained suppliers.
Frequently Asked Questions
Hyperscaler AI capex is flowing into physical infrastructure — power, land, construction, electrical systems — where small- and mid-cap companies have significant market presence. We believe this creates a fundamental earnings-cycle opportunity.
No. SMID allocations have declined for two decades, valuations sit at a wide discount to large caps, and fund flows remain historically depressed.
Power and grid access, skilled labor shortages, and specialized electrical and thermal infrastructure are the primary physical constraints limiting the pace of AI capacity expansion.
About Polen Capital
Polen Capital is a growth-oriented investment firm managing assets across U.S. and global equity strategies including large cap, SMID cap, and international portfolios.
Important Disclosures: This information has been prepared by Polen Capital without taking into account individual objectives, financial situations or needs. As such, it is for informational purposes only and is not to be relied on as legal, tax, business, investment, accounting, or any other advice. Recipients should seek their own independent financial advice. Investing involves inherent risks, and any particular investment is not suitable for all investors; there is always a risk of losing part or all of your invested capital.
No statement herein should be interpreted as an offer to sell or the solicitation of an offer to buy any security (including, but not limited to, any investment vehicle or separate account managed by Polen Capital). This information is not intended for distribution to, or use by, any person or entity in any jurisdiction or country where such distribution or use would be contrary to local law or regulation.
Unless otherwise stated, any statements and/or information contained herein is as of the date represented above, and the receipt of this information at any time thereafter will not create any implication that the information and/or statements are made as of any subsequent date. Certain information contained herein is derived from third parties beyond Polen Capital’s control or verification and involves significant elements of subjective judgment and analysis. While efforts have been made to ensure the quality and reliability of the information herein, there may be limitations, inaccuracies, or new developments that could impact the accuracy of such information. Therefore, the information contained herein is not guaranteed to be accurate or timely and does not claim to be complete. Polen Capital reserves the right to supplement or amend this content at any time but has no obligation to provide the recipient with any supplemental, amended, replacement or additional information.
Any statements made by Polen Capital regarding future events or expectations are forward-looking statements and are based on current assumptions and expectations. Such statements involve inherent risks and uncertainties and are not a reliable indicator of future performance. Actual results may differ materially from those expressed or implied.
There is no assurance that any securities discussed herein are currently held in a Polen Capital portfolio nor that they are representative of the entire portfolio in which they are or were held. It should not be assumed that any transactions related to the securities discussed herein were (or will prove to be) profitable or that any future transactions will equal the investment performance of the securities discussed herein.
References to specific securities, asset classes and financial markets are for illustrative purposes only and are not intended to be, and should not be interpreted as, recommendations.
The Global Industry Classification Standard (“GICS”) was developed by and is the exclusive property and a service mark of MSCI Inc. (“MSCI”) and Standard & Poor’s, a division of The McGraw-Hill Companies, Inc. (“S&P”) and is licensed for use by Polen Capital Management, LLC. Neither MSCI, S&P nor any third party involved in making or compiling the GICS or any GICS classifications makes any express or implied warranties or representations with respect to such standard or classification (or the results to be obtained by the use thereof), and all such parties hereby expressly disclaim all warranties of originality, accuracy, completeness, merchantability and fitness for a particular purpose with respect to any of such standard or classification. Without limiting any of the foregoing, in no event shall MSCI, S&P, any of their affiliates or any third party involved in making or compiling the GICS or any GICS classifications have any liability for any direct, indirect, special, punitive, consequential or any other damages (including lost profits) even if notified of the possibility of such damages.
Past performance is not indicative of future results.
This information may not be redistributed and/or reproduced without the prior written permission of Polen Capital.
Please see full disclosures in the white paper.