As artificial intelligence infrastructure becomes increasingly concentrated among a handful of corporations with massive capital resources, market valuations have soared—while prominent investors warn that democratic institutions have failed to establish regulatory frameworks to govern the technology's rapid expansion.
Alphabet briefly passed Nvidia by market cap in after-hours trading this week as the stock rose about 160% in the past year, driven by a view on Wall Street that Google is well positioned across the AI landscape through its homegrown models, massive distribution network and cloud unit. The company closed the week with a market cap of $4.8 trillion, behind only Nvidia at $5.2 trillion. Following Alphabet's earnings report last week, JPMorgan analysts called the stock their "top overall pick" in the tech sector, pointing to a "standout quarter," accelerating growth and a cloud backlog that nearly doubled to $462 billion.
The market enthusiasm reflects a structural consolidation of power. Gene Munster, managing partner at Deepwater Asset Management, said, "Google is one of the two best-positioned AI companies because they own most of the stack. Chips, models, infrastructure and distribution. On top of that, they're nicely profitable." He said the other company in that category is Elon Musk's SpaceX, which merged with xAI in February in a deal valued at $1.75 trillion. This concentration of capability—where the largest companies control chips, models, infrastructure and distribution simultaneously—creates structural advantages that smaller competitors cannot easily overcome.
The Concentration Problem
Beneath the market enthusiasm lies a troubling pattern of capital concentration that extends far beyond Google's dominance. Microsoft, Oracle, Amazon and Google together have close to $2 trillion in reported cloud backlog, and nearly half of that traces back to commitments from just two companies: OpenAI and Anthropic.
This concentration became visible following a report that AI model developer Anthropic committed to spend $200 billion on Google Cloud over five years for 5 gigawatts of compute. The arrangement reveals a circular flow of capital: Anthropic is described as a cash-burning and richly valued startup that is raising tens of billions of dollars from Google and, in turn, is spending much of that money with Google on cloud services and TPUs.
Analysts raised concerns about the sustainability and transparency of this arrangement. Gil Luria, an analyst at D.A. Davidson, said the setup is reminiscent of Oracle, which saw its stock soar in September after the company reported a backlog increase of almost 360%, only for investors to later realize most of that was from OpenAI. Luria said, "They did it the same way Oracle did. They told us their backlog roughly doubled without telling us that almost the entire increase came from one deal with Anthropic." This pattern suggests that large infrastructure providers may be inflating their backlog figures without disclosing the concentration of those commitments among a small number of customers.
Luria also noted that some of the demand that Google and Amazon are seeing for their in-house chips is from their portfolio companies. He said, "When Google and Amazon talk up demand for their proprietary chips, much of that is captive demand. It's not organic." This observation highlights how vertically integrated tech giants can artificially inflate demand metrics by directing their own portfolio companies to use their services, creating the appearance of broader market adoption.
Regulatory Vacuum Acknowledged
Investor Paul Tudor Jones acknowledged the policy gap explicitly. In a separate report, CNBC noted that Paul Tudor Jones said the U.S. is late to regulating AI and said, "We should have already done it." His statement reflects concern from the financial community that democratic institutions have not established frameworks to govern the concentration of AI infrastructure and the market dynamics that accompany it.
Google is projecting capital expenditures of up to $190 billion this year, more than double its capex for 2025. Analysts at Argus said in a report after earnings that "risks of Alphabet's capex spend are salient," but they have a buy rating on the stock and view the company's ability to afford those expenditures versus the likes of OpenAI as a "competitive advantage." This framing reveals a fundamental market dynamic: the ability to deploy vast sums of capital becomes itself a competitive advantage, allowing dominant firms to entrench their position further.
Mizuho analysts raised their price target, writing that consensus estimates still significantly underestimate Google Cloud revenue and operating income over the next two years. Mizuho estimates roughly $61 billion of Google's cloud backlog through 2027 could come from sales of its TPUs, and most of that revenue will likely be recognized next year. This concentration of revenue sources—where a single product line (TPUs) may account for $61 billion of backlog—creates vulnerability to shifts in technology or customer preferences.
Some analysts suggest the concentration itself is inevitable given the scale of AI infrastructure needs. Munster said, "The deal underscores how early we are in AI. Even though the use cases are limited today, the need for compute is exponential. Google will ride that wave." He added, "The headlines about size and risk of any given customer miss the point. If one of those customers blows up, over time there will be dozens to take its place." Yet this perspective sidesteps the question of whether democratic societies should allow such concentration to develop unchecked while regulatory frameworks remain absent.
Market Valuation and Future Uncertainty
Munster said the biggest threat to Google's continued outperformance is that the stock is already baking in future gains, likening that scenario to Nvidia, which continues to see huge growth but is no longer getting rewarded by investors. Analysts expect Nvidia to report 78% revenue growth later this month, according to LSEG, but the stock is only up 15% this year, slightly outperforming the Nasdaq. Munster said, "The biggest risk to owning Google is that they don't have an opportunity to change the narrative with investors." He said that puts increased weight on the company to impress at Google I/O, which kicks off in less than two weeks.
Google has gone from AI laggard to infrastructure winner in short order, but the sustainability of current valuations depends on continued exponential growth and the company's ability to maintain its competitive advantages. The current market structure—where two companies control nearly half of $2 trillion in cloud commitments, and where the largest AI infrastructure providers are also among the largest AI developers—raises governance questions that markets alone may not resolve.
Why This Matters:
The consolidation of AI infrastructure among a handful of corporations with massive capital resources creates structural inequalities in who shapes this transformative technology. When two companies control nearly half of $2 trillion in cloud computing commitments, and when the largest infrastructure providers also develop AI systems, the potential for conflicts of interest and reduced competition becomes significant. Regulatory bodies have not yet established frameworks to address these dynamics, even as prominent investor Paul Tudor Jones has explicitly stated the U.S. is late to regulating AI. The current arrangement—where startups like Anthropic become dependent on Google's infrastructure while simultaneously funding Google's growth—concentrates decision-making power in ways that democratic institutions have historically sought to prevent through antitrust and regulatory oversight. The fact that much of the reported backlog growth comes from circular arrangements between parent companies and their portfolio companies raises questions about the transparency and authenticity of market signals. Whether government institutions can establish effective oversight before further consolidation occurs remains an urgent policy question.