Stock Market Today (LIVE): Google to Invest $40B in Anthropic; TSMC Hits Record High on Policy Shift

The real signal is not just “more AI spend,” but a widening moat around infrastructure control. Alphabet’s move hardens a three-way landlord model in AI: model developers, cloud landlords, and chip suppliers all need each other, but the bargaining power is shifting toward whoever can offer secured power, data-center real estate, and financing at scale. That argues for owning the enablers with pricing power, while being wary of pure model exposure that can be commoditized by a better capital stack.

Intel’s tape is more important than the single-quarter beat. A re-rating here can persist because the market is finally paying for optionality in server CPUs and foundry credibility together; those two narratives reinforce each other if capacity tightness keeps ASPs firm for the next 2-3 quarters. The second-order risk is that the stock becomes a crowded “turnaround + AI” expression, so any capex slip, yield hiccup, or softer forward guide could unwind the move quickly even if headline demand stays healthy.

TSMC remains the cleanest beneficiary of the ongoing compute arms race, but the local-flow change also matters for cross-market arbitrage: if domestic capital is now structurally more elastic, ADR discounts can compress faster than fundamentals alone would justify. That creates a near-term tailwind for Taiwan-listed shares and a relative underperformance risk for U.S.-listed ADRs if foreign investors don’t match the incremental bid. The semis complex is strong, but the VIX/call-skew backdrop says positioning is crowded enough that upside may be more durable in select laggards than in the highest-momentum names.

The contrarian miss is that “CPU renaissance” may be less about CPUs replacing GPUs and more about a second wave of AI infrastructure demand after the GPU buildout: inference, orchestration, post-training, and agent workloads all need cheaper, more abundant compute. That broadens the beneficiary set to Intel, AMD, TSMC, and even custom-silicon ecosystems, while lowering the odds that Nvidia’s leadership translates into perpetual outperformance. In other words, the trade is no longer just AI accelerators; it’s the entire industrial supply chain behind deployed AI.