The Best Artificial Intelligence (AI) Growth Stocks to Buy on the Nasdaq as the Rally Heats Up

The article argues that AMD and Alphabet are well positioned to benefit from the next AI wave, with AMD highlighted for inference and agentic AI demand and Alphabet for its custom TPUs and lower AI compute costs. It cites AMD's partnerships with Meta and OpenAI for 6 gigawatts of next-generation GPU power and Alphabet's expanding TPU ecosystem as key growth drivers. Broader market risk is framed around the Strait of Hormuz and helium supply disruptions, but the piece is primarily bullish commentary rather than new hard data.

Analysis

The key second-order shift is not just “more AI spend,” but a re-architecting of the data-center bill of materials toward inference density and control-plane silicon. That favors vendors with memory-rich accelerators and CPU attach rates, because agentic workloads increase orchestration overhead even if raw token generation becomes cheaper. In that setup, AMD’s real opportunity is less about displacing the incumbent in frontier training and more about capturing the incremental rack buildout where buyers want a credible second source with enough software maturity to reduce single-vendor risk.

Alphabet’s advantage is more structural: custom silicon compresses its own unit economics while also letting it monetize compute as a differentiated cloud input, not just a commodity service. The hidden beneficiary is Broadcom, which becomes the “picks-and-shovels” commercialization layer if TPU adoption broadens beyond internal use; that raises the odds of durable high-margin revenue streams outside the core cloud narrative. By contrast, NVIDIA’s risk here is not immediate share loss, but mix pressure if cloud customers increasingly reserve top-end GPUs for training and shift inference to lower-cost custom silicon.

The main near-term risk is supply chain shock rather than demand weakness. A disruption to specialty gases or shipping lanes would likely hit timing more than ultimate demand, but the market could still de-rate AI hardware multiples by 10-15% on any sign of allocation constraints. Over a 3-6 month horizon, the bigger catalyst is enterprise proof that agentic AI requires more CPU-heavy racks than consensus models assume; if that plays out, the market may be underestimating AMD/INTC relative to the current GPU-centric positioning.

The contrarian angle is that the “AI winners” trade is broadening, not narrowing. Investors are crowding the obvious GPU beneficiaries while underpricing the enabling layers: CPUs, interconnect, custom ASICs, and software ecosystems that reduce inference cost. That argues for owning the second-order winners and fading the assumption that every AI dollar still flows to the same handful of training-centric names.

Key Decisions for Investors

• Go long AMD vs. NVDA on a 3-6 month horizon: pair the view that inference/agentic mix shifts favor memory-rich, CPU-attached platforms; target 15-20% relative upside if enterprise AI capex broadens beyond training.
• Add GOOGL on weakness and fund with a partial short in a basket of high-multiple AI infrastructure names: thesis is margin expansion from internal silicon economics plus cloud monetization, with lower downside than pure hardware plays.
• Initiate a small long AVGO position as a TPU commercialization surrogate over 6-12 months: asymmetric upside if external TPU demand scales, with limited fundamental dependence on one customer.
• Avoid chasing NVDA at stretched multiples into any supply-chain headline risk; use it as a source of funds rather than a core add until visibility improves on training-vs-inference mix.
• If you want a cleaner second-order expression, buy AMD Jan-2026 calls and finance by selling near-dated calls against a smaller spot position; this limits drawdown if AI spend pauses while preserving upside if rack mix shifts materially.