In the first phase of the generative AI boom, the winning strategy was straightforward: own the physical bottleneck. Alphabet’s plan announced this week to raise $80 billion suggests that the next phase may hinge on something else—the ability to finance AI capacity at scale without undermining returns.
Physical scarcity has defined the AI era so far. Companies scrambled to secure GPUs, advanced packaging, memory and data-center shells. With grid connections, power equipment and even land near cheap electricity in short supply, the market rewarded companies sitting directly in the path of frantic infrastructure demand.
That phase isn’t over. But we think the marginal bottleneck is changing.
The scarce object in AI is no longer just a GPU. It’s a financeable megawatt: a unit of AI computing power backed by energy, land, cooling, accelerators, networking and credible demand—supported by a capital structure that doesn’t impair the balance sheet.
The Google Signal: Endurance, Not Distress
As Google’s parent, Alphabet is one of the world’s most profitable businesses. It has enormous operating cash flow, backed by dominant distribution, a scaled cloud platform and proprietary models.
So, why has Alphabet chosen to raise outside equity capital instead of tapping into its healthy balance sheet? As we see it, this wasn’t a distress signal. Rather, Alphabet is effectively saying that the opportunity is too large, the spending curve is too steep and the uncertainty is too high to fund the entire AI build-out only through internal cash flow and incremental debt alone.
For the past decade, mega-cap technology companies were treated as self-funding machines that could grow, invest, acquire and return capital at the same time. AI challenges the model of capital-light businesses with extraordinary margins and huge excess cash flow.
While the AI opportunity may still be software-led, the infrastructure required to compete is capital-intensive. In our view, Alphabet’s raise suggests the AI build-out is changing from a normal corporate capex cycle to a true capital-formation cycle.
Why Berkshire Hathaway Matters
As part of the raise, Berkshire Hathaway is investing $10 billion in Alphabet. By doing so, Berkshire Hathaway is providing something increasingly scarce for AI development: patient, long-duration capital.
Anchoring to Berkshire Hathaway also signals that this isn’t merely a dilutive equity issuance. Berkshire Hathaway has lowered the perceived cost of capital, while monetizing the scarcity value of its own balance sheet.
The form of financing matters too.
If AI infrastructure were like a predictable utility project, debt would be the obvious instrument for financing dependable future cash flows.
But AI is different. While many of its physical assets may be long lived, the computing layer evolves quickly, and the technology can become obsolete. This creates a mismatch between longer-duration physical assets and shorter-duration technology risk.
Equity Financing Helps Absorb Uncertainty
Equity is better suited than debt to cope with that type of uncertainty. It can absorb delays, lower utilization and variable returns without fixed repayment obligations.
That’s why we think Alphabet’s use of equity and equity-like capital is so interesting. It suggests AI infrastructure sits at the intersection of technology, energy and infrastructure, with elements of venture risk.
What Will Define Phase Two?
If phase one of the AI build-out was defined by access to computing power, we think the next phase will be about access to financeable computing power.
Physical bottlenecks won’t disappear. But rising costs and persistent uncertainties create financing constraints that eventually weigh on a company’s balance sheet. So, today, the key question for investors is whether companies can fund capacity at acceptable returns.
From Scarcity to Capital Efficiency
The first AI trade rewarded scarcity. But the next stage could reward capital efficiency. Equity investors should increasingly ask: Does this company help the AI ecosystem lower the cost of useful computing power by producing revenue, improving productivity or fostering more durable economics?
This marks a shift in how the AI market is evaluated. In our view, spenders will increasingly need to balance technical capability with financing discipline, and investors will need to pay closer attention to how this changing dynamic filters through various AI end markets.
Creating Quality AI Capacity
Capacity build-out will no longer be an arms race. In phase two, the quality of AI capacity is likely to matter more.
Not every megawatt is equal. Without reliable power, high utilization or efficient financing, capacity may not generate acceptable returns. In contrast, a megawatt supported by demand and low-cost capital should be more valuable. This is where the AI market may become more selective. In a capital-abundant world, customers prioritize speed and availability. While that still matters in a capital-constrained world, AI customers will pay much closer attention to performance per dollar, performance per watt and the payback period on their investments.
The same logic applies across the AI stack. Businesses that lower the cost per unit of useful computing power should gain an advantage.
Demand Matters Too
Ultimately, profitability will also depend on turning AI infrastructure into revenue. We expect the market to increasingly focus on which companies can translate computing power to real use cases, customer demand and productivity gains.
Financial firms will also benefit as access to long-duration capital is likely to become more important. The next phase of the AI build-out will require more capital-market activity across the spectrum, depending on the specific spending objectives. We believe that spells more demand for balance sheets and structuring capabilities across the financial industry, which will often determine competitive advantages in an industry shaped by ongoing obsolescence risk.
So, what should equity investors learn from Alphabet’s capital raise? In our view, it marks the beginning of the financial bottleneck phase of the AI build-out, where both technology and finance will determine success. Investors must ask new questions about hurdle rates and revise valuation frameworks accordingly to discover companies that improve the economics of the AI build-out.
In our view, AI winners will need to combine infrastructure, demand and capital structure to generate durable returns. And in the next phase, losers might include companies with good technology stories but flawed funding models. We believe equity investors who can distinguish between the two will be able to avoid AI traps and build portfolios that can capture the potential of the next stage of the AI revolution.
A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.
The views expressed herein do not constitute research, investment advice or trade recommendations, do not necessarily represent the views of all AB portfolio-management teams and are subject to change over time.
References to specific securities are presented to illustrate the application of our investment philosophy only and are not to be considered recommendations by AB. The specific securities identified and described do not represent all of the securities purchased, sold or recommended for the portfolio, and it should not be assumed that investments in the securities identified were or will be profitable.
© AllianceBernstein
More Innovative ETFs Topics >
