Google at the center as AI infrastructure trade shifts away from “cheap box” makers, market commentators say

A market column making the rounds on July 2 argues that investors are starting to look past Super Micro Computer as a proxy for the next wave of datacenter buildouts. The piece says the crowded “buy the cheapest server builder” trade is losing favor as capital consolidates toward companies that own more of the full artificial-intelligence infrastructure stack.

The argument is rooted in recent stock performance and a focus on profitability at the hardware assembler level. The column points to a sharp selloff in Super Micro, saying the shares are down about 14.79% over the prior week and 41.02% over the prior month. It characterizes those moves as consistent with a fundamentals-driven unwind in hardware exposure rather than a pure valuation reset.

On operational results, the post claims Super Micro’s margins have been under pressure, describing GAAP gross margin as falling to 6.3% in Q2 FY26 before recovering to 9.9% in Q3 FY26. It also cites what it says was an earnings and revenue miss in Q3 FY26, including revenue of $10.24 billion and EPS of $0.84, contrasted with a consensus EPS estimate mentioned in the post.

The column further emphasizes the balance sheet as a risk factor. It states that Super Micro has $8.8 billion in bank debt and convertible notes, and it says cash used in operations was $6.6 billion during the quarter. In addition, it describes corporate governance and disclosure uncertainty, saying the board is running an independent review of certain transactions tied to export-control issues, with numbers in filings characterized as preliminary and unaudited.

Alphabet is positioned in the article as the alternative the market is gravitating toward. The post highlights Alphabet’s cloud business as a source of demand visibility, stating that it has a $460 billion cloud backlog, and it points to operating margins of 36% to support the view that the company can fund and scale AI infrastructure through its platform rather than relying on thin-margin hardware production.

The piece also frames Alphabet as benefiting from patient capital and analyst sentiment. It says Berkshire Hathaway made a private placement in Alphabet valued at $10 billion, and it claims there are 57 analyst buy ratings with zero sells, using that to argue the stock has stronger institutional support than the hardware proxy trade.

Zooming out, the central debate in AI infrastructure investing is whether the datacenter buildout is primarily a semiconductor and system-integration story or a platform story. The column’s thesis leans toward the latter, arguing that AI spend flows to the firms that package compute, networking, and model access into an end-to-end offering, leaving server assemblers more exposed to pricing pressure and supply-chain economics.

Still, important details are not provided in the post, and some claims cannot be independently verified from the text available here. For instance, the column does not lay out the methodology behind its “rotation underway” conclusion, and it does not reconcile its Alphabet backlog and margin figures with the most recent company reporting. It also does not provide primary documentation for the specific Super Micro export-control transaction review description beyond referencing what it says is in filings.

For investors watching this theme next, the key indicates to track are guidance and margin trajectories for hardware assemblers, plus any updates from Alphabet on cloud and AI infrastructure demand. On Super Micro specifically, developments from the independent review referenced in the post could shape how confidently investors discount governance and compliance risk, while Alphabet’s disclosures on cloud AI capacity and operating leverage would indicate whether the “owning the stack” narrative matches reality.