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

Intel’s AI-driven turnaround is gaining traction: the stock is up more than 240% year to date and about 377% since being removed from the Dow, with market cap now above $640 billion. Management’s restructuring, foundry growth, and hyperscaler deals are supporting expectations for EPS to reach $1.53 in fiscal 2027, though the stock trades at a rich 115x forward earnings versus Nvidia’s 25.8x. The piece argues Nvidia remains the more deserving Dow component, but Intel’s role in AI inferencing and custom chips has materially improved.

Analysis

The market is re-rating semis from a single-node GPU story to a systems story: CPU, memory, networking, interconnect, and custom silicon. That broadens the winner set and mechanically shifts bargaining power away from pure-play accelerators toward companies that sit at the control plane of the data center; the key second-order effect is that hyperscalers are now optimizing total cost per inference, not just peak training throughput. In that regime, Intel’s relevance improves even if it never regains leadership in discrete GPUs, because the highest-volume socket becomes the one attached to inference orchestration and general-purpose compute.

The bigger implication is that Nvidia’s moat is becoming more architectural than product-level. If racks increasingly bundle CPU, memory, and fabric with the GPU, Nvidia can monetize more of the bill of materials, but it also exposes the business to a longer product cycle and higher competition from vertically integrated buyers. That is why Broadcom, TSMC, and memory suppliers are the stealth beneficiaries: custom ASIC adoption pulls demand into design services, advanced packaging, and HBM/DRAM rather than leaving it concentrated in one SKU.

Consensus is likely underestimating how much of this is a capital cycle, not a secular straight line. The near-term risk for the whole AI complex is digestion: once hyperscaler capex is locked into custom inference racks, unit growth can slow even while revenue per rack rises, which usually pressures valuation multiples before earnings catch up. Intel’s move has likely gone too far versus fundamentals on a forward multiple basis, but the move is not invalidated unless execution slips or hyperscalers delay deployment; the more attractive setup may be a relative-value basket rather than a standalone directional bet.

Key Decisions for Investors

• Long INTC / short NVDA on a 3-6 month horizon as a mean-reversion trade: Intel has room for operational upside, but the market is already discounting a cleaner execution path than history supports. Best entered on any post-earnings pullback in INTC or after a broad AI tape fade; stop if hyperscaler deal announcements accelerate materially.
• Long AVGO vs short a basket of over-owned AI beta names for 2-4 quarters: custom silicon and networking capture the hidden spend that gets reclassified from GPU budget to platform infrastructure, with better margin durability than single-chip exposure.
• Pair long TSM / short INTC if you want to express the foundry-vs-integrated risk: TSM remains the toll collector for the entire AI supply chain, while Intel still needs flawless execution to justify its re-rating. Favor this if Intel volatility spikes on deal headlines.
• Buy MU and SNDK on pullbacks for a 6-12 month trade: inference-heavy deployments increase memory/storage intensity per rack, and this demand is less headline-sensitive than accelerator cycles. Use a staggered entry because these names can overshoot on positioning.
• Avoid chasing NVDA strength here; wait for a consolidation and re-enter only if evidence emerges that rack-level monetization is expanding faster than custom-ASIC substitution. If the stock re-rates on physical AI optionality, the upside remains, but near-term risk/reward is less compelling than the second-order infrastructure names.