New York, Feb 18, 2026, 07:04 (EST) — Premarket
- Nvidia picked up around 1.2% before the bell, with shares moving higher after the company revealed a multiyear supply deal with Meta.
- The agreement spans Nvidia’s existing Blackwell chips, upcoming Rubin systems, plus the Grace and Vera processors.
- Investors are watching for the Fed minutes set for release at 2 p.m. EST, while Nvidia’s results land on Feb. 25.
Nvidia (NVDA) climbed 1.2% to $184.97 ahead of the U.S. open Wednesday, after announcing a multiyear pact with Meta Platforms to supply millions of its existing and upcoming AI chips. The agreement spans Nvidia’s Blackwell chips, the upcoming Rubin lineup, and also bundles in Grace and Vera CPUs — those compete squarely with Intel and Advanced Micro Devices. Meta stands out as one of the four customers accounting for 61% of Nvidia’s most recent quarter revenue. The iShares Semiconductor ETF barely budged; AMD dropped 2%, Intel sagged 1.2%, but Broadcom tacked on about 2%. Nvidia’s Ian Buck described Meta’s initial tests of the Vera processor as “very promising.” (Reuters)
Meta’s pledge comes right before Wall Street turns its focus to Nvidia’s earnings next Wednesday, Feb. 25. That’s when investors will get a sharper sense of just how aggressively Big Tech is scaling up AI data center spending. Nvidia’s set to break down its fourth quarter—and its fiscal 2026 numbers, which wrapped up Jan. 25—on a conference call at 2 p.m. PT after U.S. markets close. (NVIDIA Newsroom)
Latest 13F filings reveal several heavyweight investors pulled back on Nvidia during Q4, as concerns mounted over sky-high valuations for the “Magnificent Seven” AI names. SoftBank, for its part, exited its Nvidia position entirely, saying the move frees up cash for more bets on OpenAI. Analysts are penciling in a 67% surge in Nvidia’s upcoming revenue report, per LSEG figures. (Reuters)
Nvidia said its partnership with Meta covers on-premises, cloud, and AI infrastructure, signaling a massive rollout of Nvidia Grace CPUs and “millions of” Blackwell and Rubin GPUs as Meta pushes to build data centers specifically tuned for both training and inference workloads. “No one deploys AI at Meta’s scale,” Nvidia CEO Jensen Huang said. Mark Zuckerberg, Meta’s CEO, called the Vera Rubin platform a way to “deliver personal superintelligence to everyone in the world.” The firms added that Meta is using Nvidia confidential computing for private processing on WhatsApp. (NVIDIA Newsroom)
Wednesday’s not done throwing curveballs: the Fed’s January meeting minutes land at 2 p.m. EST, a potential spark for shifting rate-cut bets. Big tech has been jumpy on rate moves—chip names usually echo those swings. (Reuters)
Spending isn’t all one-directional. The Financial Times says Meta has inked a multibillion-dollar deal for millions of Nvidia chips, chasing AI infrastructure spending that could top $135 billion by 2026. That’s even as Meta pushes ahead with its own hardware and chip projects. (Financial Times)
Nvidia faces a pivotal week, with guidance in the spotlight alongside revenue. Investors are watching for signs that demand is tilting from training to inference-heavy workloads—implications there could reshape the company’s product mix and pricing strategies. Watch, too, for details around customer concentration and the transition from Blackwell to Rubin; those could swing Nvidia’s shares.
Nvidia has its fourth-quarter FY26 earnings call on the books for Feb. 25, kicking off at 2 p.m. PT, or 5 p.m. Eastern. The webcast is listed on its investor relations calendar. (Nvidia)
