Up 377%, Is Intel Proving Why It Was a Mistake for Nvidia to Replace Intel in the Dow Jones Industrial Average?

Intel has rebounded more than 240% year to date and about 377% since being removed from the Dow, with a market cap now above $640 billion. The article argues that AI inference, CPUs, and custom ASICs have improved Intel’s role in AI infrastructure, while consensus sees EPS rising to $1.53 in fiscal 2027. Despite the turnaround, Intel still trades at 115x forward earnings versus 25.8x for Nvidia, and Nvidia remains the more deserving Dow component.

Analysis

The real signal here is not that one legacy CPU supplier re-rated; it is that the AI value chain is broadening from a GPU bottleneck into a systems market. That helps second-tier beneficiaries with attach rates into CPUs, memory, interconnect, and custom silicon, while diluting the “winner-take-most” premium once reserved for pure GPU exposure. The market is beginning to price in a multi-vendor AI stack, which is constructive for suppliers with design-in leverage but also means margins will likely migrate from accelerators toward whoever owns system architecture and customer integration.

Intel’s move looks less like a simple turnaround and more like a rerating of optionality: foundry credibility, inference relevance, and hyperscaler validation now matter more than historical share loss. The catch is that these are long-cycle wins; design-ins can take quarters to convert into revenue and even longer to prove durable. If the upcoming installation cycle shifts from pilot deployments to volume procurement, the upside can persist for 12–24 months, but any slip in execution or yield recovery would quickly unwind the multiple because the stock now discounts a cleaner comeback than the fundamentals have yet earned.

For Nvidia, the market seems to be underappreciating non-training demand elasticity. If physical AI, inference, and system-level racks expand as expected, Nvidia can preserve pricing power even as custom silicon proliferates, because the addressable market becomes broader rather than smaller. The risk is not immediate share loss but mix pressure: as hyperscalers internalize more logic, Nvidia may need to fund more of the ecosystem to defend sockets, which can cap incremental margin expansion despite strong top-line growth.

The contrarian view is that Intel may have become the crowded turnaround trade while the real mispricing is in enablers farther down the stack. Memory and networking names can capture incremental spend with less execution risk than a full manufacturing comeback, and they benefit whether workloads are served by Nvidia, Intel, AMD, or custom ASICs. In that framework, Intel is a valid tactical long, but not necessarily the best risk-adjusted expression of the AI infrastructure theme from here.

Key Decisions for Investors

• Long INTC vs short TSM on a 3-6 month horizon: expresses the market’s willingness to pay for domestic AI optionality and foundry narrative, but keep size modest because TSM remains the cleaner structural winner if execution reasserts itself.
• Pair long MU or SNDK vs short CRM/CSCO for 6-12 months: memory/storage beneficiaries should have a more direct earnings sensitivity to AI infrastructure spend than slower-moving enterprise software/networking incumbents.
• Add to NVDA on any 5-8% pullback over the next 1-2 months: the market may be underpricing inference and physical-AI upside; risk/reward improves if the stock compresses toward the low end of its recent range rather than chasing strength.
• Buy AVGO on weakness and hold 6-12 months: custom silicon and hyperscaler design wins are the cleanest second-order beneficiary of the shift toward inference and ASICs, with lower execution risk than Intel’s manufacturing turnaround.
• If INTC continues to gap higher, consider a medium-dated call spread instead of outright stock: upside remains real, but valuation now embeds a high degree of success, so options better define downside if the turnaround narrative stalls.