I Just Put More Than $10,000 Into These 3 Tech Stocks. Here's Why.

Author highlights AMD's data-center opportunity and GPU deals: OpenAI and Meta each committed ~6 GW of AMD GPUs and received warrants roughly equal to 10% of shares apiece, with the final tranche strike near $600 (~3x current price). ServiceNow's Now Assist ACV hit $600M last quarter and is forecast to exceed $1B by year-end, with revenue growing >20% and recent Armis and Veza acquisitions bolstering security and agentic-AI orchestration. Pinterest grew revenue ~14% last quarter, trades at ~11x 2026 EPS and <8.5x 2027 EPS, and benefited from Elliott's $1B convertible-note purchase to fund accelerated buybacks, leading the author to call it deeply undervalued.

Analysis

Agentic AI is shifting compute economic plumbing: as models orchestrate many external tools and long-running agent processes, high-performance CPUs (I/O, context switching, model orchestration) become a gating factor rather than pure GPU FLOPS. That change amplifies demand not just for cores but for platform pieces: memory controllers, coherent interconnects, NICs with RDMA/GPUDirect, and chiplet-friendly packaging — creating multi-vendor TAM expansion where a successful CPU entrant can monetize adjacent supply chain leverage (board vendors, HBM vendors, interposer fabs).

For capital allocators, the interesting asymmetric outcomes come from adoption frictions. If an alternative software stack (open ROCm or other runtimes) secures cloud hyperscaler scale, it creates an install-base moat that translates into recurring wafer/tape-out economics and enterprise CPU share gains over several years; failure to attract that scale leaves capacity underutilized and forces price competition. Enterprise SaaS winners that have deep workflow lock-in and embed agent orchestration will see disproportionate revenue stickiness and improved ARPU as AI features become differentiated — but operational risk rises where models generate incorrect actions and trigger SLAs or security incidents, making recent security/SaaS acquisitions strategically valuable.

Pinterest is a classic optionality case: cheap equity plus an activist-led buyback-financing path creates optional EPS re-leveraging if ad monetization gains continue, but near-term cyclicality in retail and privacy headwinds are real and can compress multiples quickly. Timing matters — the next 6–18 months will determine whether AI-induced ARPU lift and buybacks are realized (dislocation -> re-rate) or if macro-driven ad weakness and regulatory friction re-price the story.

Key Decisions for Investors

• Pair trade (9–18 months): Long AMD (2–3% portfolio) / Short INTC (1–1.5% portfolio). Rationale: capture potential CPU share shift and superior platform monetization while hedging sector beta. Target: 40–60% gross upside on AMD leg if EPYC wins share; max downside: limit to 20–25% portfolio loss via stop/size if foundry allocation or software adoption stalls.
• Options trade on AMD (12 months): Buy AMD 12-month LEAP call spread (buy 1x 2027 Jan 90C, sell 1x 2027 Jan 150C) sized to 1% portfolio. Rationale: asymmetric bet on CPU-driven re-rating with defined capital at risk. Reward: ~3–4x if shares double; risk = premium paid.
• Income + convexity on ServiceNow (6–12 months): Buy NOW stock (1.5–2% portfolio) and sell near-term 6–8 week 10–12% OTM covered calls to finance position. Rationale: monetize elevated option premia while holding a SaaS franchise with embedded AI orchestration optionality. Target: 20–35% annualized return; downside: protect or trim at 18–22% drawdown.
• Event-hedged long on Pinterest (9–12 months): Buy PINS (2% portfolio) and purchase a 9–12 month ~15% OTM protective put (cost ~2–3% of position) or size with a trailing stop. Rationale: activist-funded buybacks plus AI-driven conversion improvements create asymmetric upside while the put caps black-swan ad-cycle losses. Target: 40–60% upside to catalyst realization; capped downside equal to put cost plus any residual share loss.