Caterpillar vs. Corning: Which Stock Is a Better Buy in 2026?

Caterpillar generated FY2025 revenue of $67.6B (+4.3%) and $7.5B of free cash flow, though net income fell to $8.9B from $10.8B as net margin declined 360bps to 13.1%. Corning delivered stronger growth, with revenue up 19.1% to $15.6B and net income rising to $1.6B from $506M, aided by AI data-center and optical-communications demand. Caterpillar trades at 28.7x forward P/E and 5.3x sales versus Corning at 43.0x and 7.8x, leading the analysis to favor Caterpillar as the steadier long-term choice despite its cyclical exposure.

Analysis

The relevant distinction is not “cyclical industrial” versus “technology,” but where each sits in the capex stack. CAT is a late-cycle beneficiary of mining, power-generation and infrastructure spend, with dealer inventories and used-equipment values providing early signals on end-demand. Its financing arm also makes headline leverage less informative than it appears; the more investable risk is a dealer inventory correction, which can turn a modest end-market slowdown into outsized production and margin pressure over 1-3 quarters.

GLW has greater operating leverage to AI-related connectivity, but its upside depends on volume growth converting into durable pricing rather than merely absorbing underutilized manufacturing capacity. The market is likely assigning a scarcity multiple to optical-fiber exposure despite customer concentration and the possibility that hyperscalers shift mix toward higher-value transceivers, silicon photonics or alternative suppliers. A single large customer’s display, handset, or data-center procurement reset could create an earnings revision cycle disproportionate to the underlying end-market slowdown.

The valuation spread favors CAT on current earnings, but that is not automatically a long signal: CAT’s multiple can compress quickly if order backlog, dealer inventory or mining equipment utilization rolls over. Conversely, GLW’s elevated multiple is defensible only if optical communications growth sustains through 2027 and management demonstrates incremental margins above the recent recovery level. This is low-immediacy commentary rather than a standalone catalyst; monitor estimate revisions and order indicators before adding directional risk.

Contrarian view: AI infrastructure spending does not necessarily accrue most to GLW. Networking vendors and optical-component suppliers with direct exposure to bandwidth upgrades may capture more of the profit pool, while GLW bears capacity, customer-concentration and pricing risks. CAT may be the cleaner relative value only if power demand and mining capex remain resilient, not simply because it trades at a lower stated multiple.

Key Decisions for Investors

• Maintain a watchlist pair rather than initiate immediately: long CAT / short GLW on equal-dollar basis if GLW forward EPS estimates flatten or decline while CAT dealer inventory and backlog data remain stable. Target a 10-15% relative return over 3-6 months; stop if GLW secures independently verifiable multiyear optical-volume commitments or CAT guides to weaker dealer retail sales.
• For existing CAT exposure, use a 3-6 month downside hedge through CAT puts or an XLI put spread around the next dealer-statistics and earnings window. The key falsifier is a sequential decline in dealer inventory absorption or a material cut to machinery segment margin guidance, not a change in reported debt-to-equity.
• Do not add GLW solely on AI narrative momentum. Upgrade to a tactical long only after segment-level optical sales growth and incremental margin validate that revenue is not being purchased through pricing concessions or capacity utilization; the missing data are hyperscaler customer concentration, contract duration, and optical-fiber pricing.
• For broader AI-connectivity exposure, compare GLW’s estimate revision trend with direct networking/optical beneficiaries before allocating incremental capital. If GLW continues to outperform without upward earnings revisions over 1-3 months, favor trimming or short overlays rather than chasing multiple expansion.