Nvidia's Memory-Chip Alliance and Vera Rubin Rollout Give Bulls Fresh Ammunition

Nvidia stock rebounds 8.6% after earnings, with SK Hynix memory deal, Vera Rubin production, and new networking tech driving growth.

The whiplash was brief but telling. Nvidia's blowout quarterly numbers initially drew a muted after-hours response — a sign of just how stratospheric expectations had become — before the stock snapped back with a vengeance the following day. By Thursday, the shares had climbed roughly 8.6 percent to 195.60 euros, leaving the longest losing streak since 2022 firmly in the rear-view mirror.

What's powering the rebound goes beyond the earnings print itself. The company has been quietly fortifying its supply chain and product roadmap in ways that suggest the AI infrastructure buildout is only accelerating.

SK Hynix Deal Locks In Critical Memory Supply

At the center of that effort is a multi-year partnership with SK Hynix, confirmed this week, covering AI memory chips destined for Nvidia's next-generation hardware. The agreement spans the upcoming Vera Rubin systems, the Vera CPU, RTX Spark, and Jetson Thor — a clear signal that Nvidia is moving early to secure access to components that have become a bottleneck across the industry.

The timing is no coincidence. Nvidia has warned of "extreme pricing conditions" in the memory market, and those pressures are intensifying. The SK Hynix tie-up reads less as a routine technology collaboration and more as a strategic hedge against a supply chain that could otherwise throttle growth.

A New Networking Stack for the AI Era

Alongside the memory deal, Nvidia unveiled two technologies aimed at squeezing more performance out of future accelerators. The NVLink Fusion and NVHBM architectures promise 30 percent higher memory bandwidth and 15 percent lower power consumption compared to HBM4e, while freeing up 25 percent more die area for compute logic.

The company is leaning on a broader ecosystem to make this happen, partnering with Marvell, MediaTek, and Qualcomm rather than going it alone. That marks a deliberate shift toward embedding itself across the semiconductor value chain.

On the networking front, Nvidia introduced the Mellanox Switch 920-9B210-00FN-0D0, a component built for large-scale AI clusters. With 64 ports running at 400 gigabits per second each under the NDR standard, the switch delivers a total switching capacity of 51.2 terabits per second. Its integrated SHARPv3 technology cuts latency in collective communication by 40 percent — a metric that matters directly to data center operators whose GPU fleets depend on near-instantaneous data exchange.

Vera Rubin Reaches Full Production

The Vera Rubin platform, Nvidia's next major growth driver, has now hit full production and is expected to account for roughly a fifth of data center revenue in the current quarter. The supply chain for this generation is reportedly twice the size of the one built for Grace Blackwell — a telling indicator of how demand is shaping up for the quarters ahead.

That optimism is grounded in the numbers. Nvidia's fiscal second quarter delivered revenue of $96.2 billion, up 106 percent from $46.7 billion a year earlier. Earnings per share nearly doubled to $2.22, while gross margin improved to 75.0 percent. The data center segment, now 92 percent of total revenue, brought in around $89 billion — comfortably ahead of the $86.33 billion analysts had penciled in, and up 117 percent year over year.

For the current quarter, CEO Jensen Huang guided to $108 billion in revenue, again above consensus, and notably without any China data center revenue factored in. His message on the earnings call was characteristically direct: "AI has reached its tipping point. It does useful work. Its tokens are productive and profitable. Now compute equals revenue."

The demand picture extends well beyond Nvidia's own numbers. The five largest hyperscalers are expected to lift capital expenditures from $800 billion this year to $1.3 trillion next year, and Amazon Web Services has committed to ordering two million Nvidia GPUs along with the new Vera CPU platform.

What the Technicals Say

The stock's current level of 195.60 euros sits 3.4 percent below its 52-week high of 202.50 euros. More striking is the distance from the 200-day moving average of 168.64 euros — a 16 percent gap that underscores how far the uptrend has stretched in a relatively short period.

For investors, the question now shifts from whether Nvidia can deliver on its current generation to whether the promised supply chain capacity translates into revenue over the coming quarters. The company itself has flagged rising memory costs as a headwind, and the SK Hynix partnership is, in part, an answer to that challenge.

The broader takeaway is that Nvidia is no longer just selling GPUs. Between the Vera Rubin rollout, the new networking infrastructure, and the deepened memory partnership, the company is positioning itself to control the entire AI infrastructure stack — from storage and switching to the compute engines at its core. Whether that vertical integration pays off at the pace the market expects remains the open question, but the pieces are clearly being put in place.