Cramer: Do not to sell these 2 stocks with major ties to the AI data center buildout

AI-driven hyperscaler spending was lifted to $695 billion, up more than 14% from prior commitments, reinforcing the market’s data-center buildout theme. Eaton and Corning were both highlighted as well-positioned beneficiaries of that capex cycle ahead of upcoming earnings and an Analyst Day, while the S&P 500 was slightly lower despite the sector tailwind. The commentary is constructive for AI infrastructure names but is mostly a market overview rather than a single stock catalyst.

Analysis

The market is increasingly pricing AI infrastructure as a multi-year capex cycle rather than a one-quarter enthusiasm trade, and that matters for second-order winners. The key implication is that the spend is no longer just a cloud-margin story for hyperscalers; it is becoming a throughput and utility bottleneck story for the entire electrical stack, which should keep order visibility elevated for names like ETN and GLW even if software monetization remains uneven.

The more interesting trade is not whether AI capex is real, but whether the market is underestimating how long it can stay real before returns on that spending come under pressure. If hyperscaler capex remains elevated into the next 2-3 quarters, the beneficiaries are the picks-and-shovels suppliers with pricing power and long lead times; if it rolls over, the first damage shows up in the higher-beta suppliers that have most directly re-rated on the theme. That makes the setup favorable for near-term earnings/catalyst upside in ETN and GLW, but less compelling for chasing the hyperscalers here after a strong run.

The contrarian miss is that consensus is treating AI infrastructure as one homogeneous trade, when the earnings durability is likely to diverge sharply by exposure type. Electrical, cooling, and fiber demand can compound even if model economics fluctuate, while more direct AI software beneficiaries need continued evidence of monetization to justify current multiples. In that sense, the best risk/reward is in the enablers, not the platform names, and the market may still be underappreciating how sticky datacenter buildout demand can be through 2026.

Key Decisions for Investors

• Add ETN on any post-earnings volatility or market dip over the next 1-2 weeks; use pullbacks as entry because the datacenter/utility capex cycle has a 12-24 month runway and downside is cushioned by backlog visibility.
• Hold or add GLW into Analyst Day; best risk/reward is for a 3-6 month horizon if management confirms hyperscaler deal monetization, with upside from long-duration fiber demand and downside limited unless capex guidance rolls over.
• Take partial profits in AMZN, GOOGL, META, and MSFT into strength rather than initiating new longs here; the trade is crowded and the next leg depends on proof of AI revenue conversion, not just spend announcements.
• Consider a pair trade: long ETN / short a basket of high-multiple AI beneficiaries with weak near-term monetization visibility over the next quarter, to isolate infrastructure spend from narrative beta.
• For new money, prefer a staged entry in GLW and ETN over 2-4 weeks rather than all at once; if AI capex momentum persists, these names can compound, but if the market rotates, they should outperform on relative earnings durability.