Morgan Stanley has doubled down on Nvidia as its top semiconductor pick, reaffirming a $288 price target after a meeting with CEO Jensen Huang, as the chipmaker’s AI revenue rockets toward an annual run rate of $100 billion. The vote of confidence, paired with fresh details on the next-generation Vera Rubin architecture, signals that Nvidia’s grip on the AI infrastructure market is only tightening — with direct consequences for Windows users, from the AI features in your next laptop to the GPU you’ll need for gaming and content creation.

Morgan Stanley’s Bullish Bet

In a note to clients, Morgan Stanley analyst Joseph Moore called Nvidia “the most compelling story in semiconductors,” maintaining an Overweight rating. The $288 price target implies a significant upside from current levels, reflecting conviction that AI spending is far from peaking. Moore’s confidence was bolstered by the meeting with Huang, which he described as reinforcing the roadmap and demand pipeline. While specific financial projections weren’t disclosed in the public excerpt, the firm’s math now places Nvidia’s AI-related revenue on pace to surpass $100 billion annually — a staggering figure that would have been unthinkable just three years ago.

AI Revenue Nears $100 Billion

The $100 billion milestone is not an official Nvidia forecast but the brokerage’s extrapolation of current data center GPU sales, licensing, and networking revenue. At the heart of that surge is the H100/H200 and Blackwell product lines, which are being snapped up by hyperscalers including Microsoft, Amazon, and Google. Microsoft alone has poured billions into Nvidia hardware for Azure’s AI services, which underpin everything from GitHub Copilot to enterprise AI agents. As Windows users increasingly interact with AI tools — whether through Microsoft 365, Windows Copilot, or third-party apps — much of that back-end processing runs on Nvidia silicon in the cloud.

Vera Rubin: The Next Leap

During the meeting, Huang is said to have offered a glimpse of the Vera Rubin architecture, the successor to Blackwell, expected to debut in 2026. Named after the pioneering astronomer, Rubin is rumored to use a chiplet design with TSMC’s 3nm process and HBM4 memory, delivering a 2-3x performance uplift for AI training and inference. For local workloads, the consumer variant — likely branded as GeForce RTX 60-series — could bring substantial gains to Windows laptops and desktops. Key improvements under discussion include:

  • Double the transistor density compared to Blackwell, enabling more AI-specific accelerators on-die.
  • Upgraded NVLink interconnects for multi-GPU scaling, critical for data centers and high-end workstations.
  • PCIe 6.0 and DisplayPort 2.1 support, future-proofing for high-refresh 8K monitors.
  • An enhanced media engine capable of decoding 8K 60fps video with real-time AI upscaling.

While details remain speculative, Morgan Stanley’s note explicitly cites Rubin as a catalyst for long-term growth, suggesting that early customer engagement has been “unusually strong” even by Nvidia’s standards.

What It Means for Windows Users

For Home Users and Gamers

If you’re running Windows 11 and wondering whether your next GPU purchase will be future-proof, the Vera Rubin roadmap matters. Current RTX 40-series cards already accelerate features like DLSS 3.5 and Windows Studio Effects (background blur, eye contact) on NPU-capable Copilot+ PCs, but Rubin promises to push AI-assisted rendering and productivity to new levels. Gamers can expect even more realistic ray tracing and frame generation, while creators will see shorter export times in tools like Adobe Premiere Pro and DaVinci Resolve, which lean heavily on GPU compute.

That said, the runaway demand for AI silicon has kept GPU prices elevated. The H100’s profitability for Nvidia means consumer cards often get less capacity allocation. If history repeats, the RTX 60-series could face supply constraints at launch. Morgan Stanley’s target implies that Nvidia’s data center business will remain the star, so gamers and home users may need patience — or be prepared to pay a premium.

For IT Professionals and Developers

Admins managing Windows fleets or hybrid cloud environments will feel the ripples in two ways. First, Microsoft’s aggressive rollout of Copilot and other AI features in Office and Windows will increase backend demand for Nvidia GPUs in Azure, potentially affecting instance availability and pricing. Second, the arrival of Vera Rubin will likely prompt a refresh cycle in on-premises AI training clusters. If your organization is planning an AI workstation build for model fine-tuning or inference, waiting for the Rubin-based RTX 6000 Ada successor might be worth it — but that’s late 2026 at the earliest. In the meantime, Blackwell-based RTX 5000-series cards (expected later in 2025) will offer a more immediate step up.

Developers building Windows apps with AI inference on-device should note that Microsoft is heavily pushing the NPU in Copilot+ PCs. Nvidia’s GPUs, however, remain the go-to for training and for inference tasks that exceed the NPU’s abilities. The Windows AI toolkit and DirectML already support Nvidia acceleration, so Rubin’s raw throughput will directly translate to faster development cycles.

How We Got Here: From GeForce to AI Juggernaut

Nvidia’s transformation didn’t happen overnight. A decade ago, the company was primarily known for its GeForce gaming cards, with a budding enterprise business. The turning point was the 2016 launch of Pascal and the subsequent V100 accelerator, which made Nvidia the default choice for deep learning researchers. By the time OpenAI used thousands of Nvidia GPUs to train GPT-3, the company’s data center revenue had eclipsed gaming for the first time.

Microsoft’s relationship with Nvidia deepened as Azure became one of the first clouds to offer GPU instances at scale. Today, Microsoft is Nvidia’s second-largest customer, and the two companies collaborate closely on AI frameworks and hardware/software co-design. The release of Blackwell in 2024 and the upcoming Vera Rubin underscore a cadence that keeps competitors like AMD and Intel in a constant state of catch-up.

What to Do Now

  1. For gamers and creators: If you need a GPU now, the RTX 40-series Super cards offer strong value and will handle AI-augmented workloads in Windows for years. Waiting for Rubin means a 2026 purchase at best, with potential for limited supply. The RTX 50-series (Blackwell) will close the gap sooner — watch for official announcements at Computex 2025.
  2. For IT buyers: Start evaluating your AI infrastructure roadmaps in light of Rubin’s projected 2026 arrival. If your current hardware meets near-term needs, delaying a major refresh could yield significant performance-per-dollar gains. For immediate needs, the H100/H200 remains the workhorse; negotiate supply agreements early.
  3. For developers: Keep an eye on the Windows Copilot Runtime and DirectML updates. Nvidia’s CUDA ecosystem will remain dominant for the foreseeable future, but Microsoft’s push toward framework-agnostic AI means you may be able to target NPU, GPU, and cloud with a single codebase using ONNX Runtime or similar tools.
  4. For everyone: GPU pricing is unlikely to dip dramatically as long as AI demand outstrips supply. If you can wait, Rubin’s advanced process node should bring better energy efficiency and maybe a reset in the used market as enterprises upgrade.

Outlook: Rubin and Beyond

Morgan Stanley’s note signals that Wall Street sees Nvidia running well ahead of the pack. Vera Rubin is the next mile marker, but the road extends further: a 2028 “Feynman” architecture is already rumored. For Windows users, the practical outcome is clear: the AI features you use daily — from search to photo editing to code completion — will get faster and more capable on both local and cloud hardware. Whether you’re a home tinkerer or an enterprise admin, Nvidia’s $100 billion AI engine is powering the Windows experience of tomorrow. Buckle up.