New York, March 26, 2026, 07:19 EDT
Shares of NVIDIA Corporation (NVDA) drifted just below $179 early Thursday, sticking close to their previous session’s 2% gain. That uptick kept the AI chipmaker’s market cap steady near $4.53 trillion. The stock tracked Wednesday’s recovery in chip names.
Strong results haven’t been enough to budge the stock so far. Investors dumped Nvidia right after February’s earnings—they weren’t convinced the company’s aggressive AI spending would deliver returns quickly. At GTC this month, CEO Jensen Huang tried to settle the nerves, pitching Blackwell and Rubin as a massive $1 trillion revenue opportunity by 2027. He declared, “the inference inflection has arrived”—that’s the phase where trained models actually get put to work, answering questions or carrying out tasks. Reuters
Wednesday brought an early trial: SLB announced an expanded partnership with Nvidia, aiming to develop AI infrastructure and models tailored for the energy industry. Vladimir Troy of Nvidia described the initiative as a way to transform massive stores of energy data into “actionable insights,” targeting faster, more efficient systems. Reuters
The startup pipeline is showing fresh signs of investor hunger. According to The Wall Street Journal, Nvidia-backed Reflection AI is reportedly in discussions to secure $2.5 billion in funding, targeting a lofty $25 billion valuation, though Reuters was unable to confirm this independently.
Rivals aren’t standing still. Arm claimed its latest data-center CPU—a chip built for AI jobs—might pull in close to $15 billion a year within five years. That helped drive Arm shares up 20% on Wednesday, while AMD and Intel added more than 5%.
This is significant because inference tasks aren’t as picky about hardware as model training — they can run on a wider range of chips. “More competition compared to a year ago” is coming for Nvidia, said KinNgai Chan at Summit Insights, pointing to big customers like Meta and OpenAI stepping up their own chip development. Over at Third Bridge, William McGonigle noted CPUs are “back in focus,” with AI agents — software capable of tackling multi-step tasks with minimal human guidance — starting to pick up more of the workload. Reuters
Nvidia’s numbers keep soaring. Fourth-quarter revenue landed at $68.1 billion, which is a 73% jump year-on-year, while the full-year tally reached $215.9 billion, up 65%. Adjusted earnings came in at $1.62 a share. The current-quarter sales outlook? That figure came in higher than what Wall Street had penciled in.
One order hasn’t faded from view: Last week, Nvidia put forward its plan to move 1 million chips to Amazon Web Services by the close of 2027, as part of a wide-reaching cloud agreement that ropes in Spectrum networking hardware and more of Nvidia’s own tech.
The tone has soured again. U.S. stock futures slipped Thursday—Nasdaq 100 futures were off 0.73%—with markets retreating from Federal Reserve rate cut bets as Middle East jitters kept oil prices stubbornly high. Reuters Breakingviews notes that big tech players are set to spend around $630 billion on data centers and AI chips in 2024. But it’s the basics—power, permits, transformers and cooling equipment—that are now in the way, slowing down how fast Nvidia chips get plugged in and put to use.
