3 Tech Stocks That Are Screaming Buys as the Indexes Reach Record Highs

The article highlights investor hesitation amid frothy AI valuations (Shiller CAPE ~42 near record highs) but argues there are value opportunities in three ad/AI infrastructure names. Meta plans $115B–$135B capex in 2026, with liquidity of ~$81B and trailing-12-month free cash flow of ~$46B, alongside a drop in P/E to 21. The Trade Desk’s growth slowed (Q1 2026 revenue growth to 12% vs >20% historically; ~10% expected for full-year 2026) but retention stayed >95% and P/E fell to 22, while Oracle faces leverage/capex concerns after the OpenAI deal yet reports a 47% YoY jump in cloud revenue (Q4 FY2026) and a backlog cited as up to ~$330B even under worst-case assumptions.

Analysis

The market is still underpricing the asymmetry in hyperscaler capex: the near-term earnings drag is visible, but the monetization path is much slower to show up in reported numbers. That favors the infrastructure layer first — semis and networking beneficiaries like NVDA — while it leaves software and ad-tech names exposed if they cannot prove they are converting AI spend into measurable ROI. META is the cleanest balance-sheet holder here, but the stock only works if investors are willing to look through 2-4 quarters of capex compression.

TTD looks less like a cheap cyclical and more like a share-regain story against walled gardens that now control both demand and measurement. The key second-order effect is that transparency friction can cause advertisers to consolidate budgets into META/GOOGL/AMZN if the optimization loop is easier inside their ecosystems; that would keep a lid on TTD growth even if digital ad spend improves. A sub-15% growth profile is probably enough to stabilize the stock, but not enough to justify rerating without evidence that product migration stopped hurting conversion.

ORCL is the most fragile setup because the market is starting to discount backlog quality rather than backlog size. High capex plus elevated leverage means any delay in converting contracted demand into cash flow forces the equity to trade more like a levered infrastructure utility than a growth compounder. The contrarian view is that the market may be over-discounting client-funding risk: if cloud growth stays near current levels, the stock can re-rate, but that requires visible execution over the next 1-2 quarters rather than faith in the backlog number alone.

Key Decisions for Investors

• Long NVDA vs. short TTD into the next 1-2 earnings cycles: the trade is that AI capex monetizes first in chips, not in ad-tech. Risk/reward favors NVDA if hyperscaler spend remains elevated; invalidation is a sharp slowdown in datacenter orders or a reacceleration in TTD revenue back above the low-teens.
• Avoid outright shorting META; instead buy on a 5-8% post-earnings drawdown or when the market starts discounting capex as permanent margin pressure. Time horizon 6-18 months. Falsifier: if ad pricing or engagement weakens enough to force downward revenue revisions, the valuation support breaks.
• Use TTD as a relative short against META/GOOGL rather than a naked short: the thesis is distribution power consolidation, not business failure. Entry is best after any rally on 'AI platform' messaging; cover if quarterly growth reaccelerates above expectations or customer retention slips from the 95%+ range.
• Do not chase ORCL on headline backlog alone; wait for evidence that backlog converts into cash flow while capex moderates. If the stock stabilizes but free cash flow remains pressured, consider a cautious long only after the next cloud-growth update; invalidation is continued debt-funded capex without backlog monetization.
• If you need a cleaner expression of AI infrastructure demand, prefer the picks-and-shovels basket over single-name platform risk: NVDA plus an underweight in TTD/ORCL. This captures the current phase where spend is real but end-demand proof is still pending.