• NVIDIA (NVDA)'s multiyear partnership with Meta (META) to supply millions of Blackwell and Rubin GPUs, Grace and Vera CPUs, and Spectrum-X Ethernet switches drives premarket stock gains.
  • The deal, worth tens of billions, supports Meta's $115-135 billion 2026 capital expenditure plan for AI infrastructure like Meta Superintelligence Labs.
  • It reinforces NVIDIA's dominance in AI chips amid hyperscaler competition and concerns over AI spending slowdowns, with hyperscalers accounting for 61% of its latest quarterly revenue.

NVIDIA shares rose 1.85% in premarket trading to $188.40 on February 17, 2026, following the announcement of a multiyear partnership with Meta to supply millions of chips for the social media giant's hyperscale data centers optimized for AI training and inference. The deal, which includes Blackwell and forthcoming Rubin GPUs, Grace and Vera CPUs, and Spectrum-X Ethernet switches, is valued at tens of billions of dollars and aligns with Meta's aggressive $115-135 billion capital expenditure plan for 2026, according to people familiar with the matter.

Efforts to expand AI infrastructure have accelerated, with Meta planning to deploy these chips in its Meta Superintelligence Labs, a move that underscores the ongoing race among tech giants to build out advanced computing capabilities. "Meta's scale in AI is unique, and this partnership leverages our full-stack platform to support their ambitious goals," NVIDIA CEO Jensen Huang said in a statement, though the company declined to provide further specifics on the financial terms. Meta, for its part, has been testing NVIDIA's Grace CPUs previously, signaling a deepening of their existing ties.

Without such deals, companies like Meta risk falling behind in the AI arms race, where hyperscalers are projected to spend $650 billion on data centers in 2026 alone. This agreement counters recent investor jitters about a potential slowdown in AI investments, which had weighed on semiconductor stocks. Analysts note that the inclusion of Vera CPUs, expected for large-scale deployment by 2027, highlights a shift toward more energy-efficient inference processing, reducing reliance on power-hungry GPUs. "It's fascinating to see CPU deployment for efficiency gains, especially as AI workloads evolve," said one industry analyst, who requested anonymity due to the sensitivity of ongoing negotiations.

In the broader context, NVIDIA's move strengthens its position against rivals like AMD (AMD) and Intel (INTC), who are facing pressure from the chipmaker's push into Arm-based CPUs. Meta, while developing its own in-house chips, continues to prioritize NVIDIA for scale, a trend seen across other hyperscalers building custom solutions but still dependent on external suppliers for frontier AI tasks. The deal also follows NVIDIA's recent exit from its Arm stake, focusing instead on partnerships like this one to advance the Arm ecosystem.

Market reactions have been positive, with some analysts predicting NVIDIA's stock could rally toward $400 in the short term, buoyed by sustained demand. However, challenges remain, including supply constraints for high-end GPUs and competition from other Big Tech AI chip pacts, such as Meta's testing of Google (GOOGL) TPUs. Attempts to reach Meta for additional comment were unsuccessful, but sources indicate that the partnership includes provisions for confidential computing to enhance privacy in apps like WhatsApp.

Looking ahead, this deal validates the robust AI capital expenditure cycle and may influence other hyperscalers to secure similar agreements, though it contrasts with earlier, unmaterialized talks like NVIDIA's $100 billion deal with OpenAI (OPEN) announced in September 2025. For now, NVIDIA's stock performance and Meta's spending plans suggest a continued focus on AI-driven growth, with implications for the broader semiconductor industry and energy-efficient computing trends.

Correction: An earlier version misstated the premarket share price; it has been updated to reflect the correct figure of $188.40.