Nvidia's record buyback shows chipmaker's stock is too cheap for CEO Huang to resist

Nvidia authorized an additional $150 billion in share repurchases, supplementing an $80 billion plan announced in May, as management argues the stock is undervalued relative to its earnings trajectory. Its fiscal-2028 forward P/E is 14.5 versus a five-year average current P/E of 62.9, while analysts forecast nearly $385 billion in fiscal-2028 net income, up 60% year over year. Nvidia has guided to 70% fiscal-2028 sales growth and said full use of the buyback authorization could reduce its share count by roughly 4%; UBS estimates the stepped-up repurchases could add $0.08 to 2027 EPS of $17.16.

Analysis

The repurchase is economically immaterial to EPS near term, but strategically important: allocating incremental cash to stock rather than capacity, acquisitions, or customer financing indicates Nvidia sees diminishing returns on reinvestment at the margin. That shifts the debate from supply availability to durability of hyperscaler AI spend. A 14-15x forward multiple only remains attractive if the earnings base is sustained; even a 15% reduction in fiscal-2028 estimates would remove most of the apparent valuation discount.

The second-order negative is for AMD and Intel, where a lower NVDA valuation reduces the relative-multiple support for challengers without improving their product economics. AVGO is more nuanced: custom ASIC adoption is the principal mechanism that could compress Nvidia's inference margins, but a sustained Nvidia cash-return program also raises the hurdle for customers to fund alternative software ecosystems. Suppliers with broad AI exposure, particularly MU, benefit if total accelerator unit growth persists, but their earnings remain more exposed to memory pricing than to Nvidia-specific share gains.

Consensus is treating buybacks as a valuation catalyst rather than a signal that Nvidia's growth profile is transitioning from scarcity-driven expansion toward capital-return maturity. The stock can outperform over 1-3 months as estimate revisions and reduced float support sentiment, but the 6-18 month catalyst is customer ROI disclosure: any hyperscaler capex moderation, lower GPU utilization, or evidence that inference shifts to ASICs would drive a sharper de-rating than the repurchase can offset.

Key Decisions for Investors

• Maintain/add NVDA on weakness rather than chase the announcement; target a 1-3 month hold into next guidance and hyperscaler capex updates. Thesis requires fiscal-2028 revenue expectations to hold; reduce if consensus revenue falls more than 10% or management signals material gross-margin pressure.
• Express relative AI compute leadership via long NVDA / short AMD in equal dollar size for 3-6 months. AMD needs meaningful accelerator revenue conversion and software adoption to close the gap; cover the short if AMD raises data-center AI revenue guidance by more than 20% or NVDA's data-center growth decelerates below 30%.
• Avoid treating the authorization as a standalone EPS trade: the expected per-share benefit is too small to justify long-dated call premium. Use any post-event rally to monitor implied volatility and only add upside exposure if next-quarter guidance confirms demand beyond the current hyperscaler order cycle.
• Watch AVGO versus NVDA as the key structural hedge: initiate a small long AVGO/NVDA pair only if evidence emerges that custom-silicon programs are ramping faster than expected. The trigger is accelerated ASIC revenue commentary from major cloud customers; absent that, Nvidia's platform/software moat remains the cleaner exposure.