After suffering a drop of more than 11% over the past month, NVIDIA (NASDAQ:NVDA) found renewed buying interest on the last trading day of the quarter following a bullish assessment from semiconductor research shop SemiAnalysis. Shares were trading up 1.7% at noon New York Time.
SemiAnalysis projected that NVIDIA's data-center compute revenue in the second half of fiscal 2027 will be approximately 20% above consensus estimates from Wall Street, pointing to the easing of an HBM4 memory supply bottleneck and a faster-than-expected ramp of the company's Vera Rubin platform as the main drivers.
NVDA shares are currently trading at $198.37, up 1.74% intraday, with the SemiAnalysis note cited as a key reason behind the upward move.
"We are seeing a huge second half ramp for Nvidia this year. Our Accelerator Model estimate has Nvidia DC compute revenue 20% higher compared to consensus expectations for 2H FY2027. Rubin will have a big ramp after being pushed out due to earlier HBM4 issues that are now resolved and front end wafer supply is now built up. Unlike other sell-side analysts who like to model an easy benchmark for companies to beat numbers, SemiAnalysis estimates are informed by all the research from the supply chain - including but not limited to materials, fabrication, components, server integrators - all the way to what's happening at hyperscalers and frontier labs."
The crux of SemiAnalysis's bullish thesis is that HBM4 memory shortages that had delayed mass shipments of the Rubin platform have been addressed. With that bottleneck cleared and front-end wafer inventory built up, the firm anticipates a sharp increase in volumes in the second half of the fiscal year.
That supply-chain view was echoed in reporting from Digitimes on June 25, which indicated that the vendors tracked by its reporters see the Blackwell-to-Rubin product transition completing in Q2 2026, setting the stage for demand to accelerate from Q3 onward.
SemiAnalysis's note referenced the Vera Rubin platform entering full production on June 1, 2026, and said production shipments are slated for this fall to eight cloud partners: AWS, Google Cloud, Microsoft Azure, Oracle Cloud, CoreWeave, Lambda, Nebius, and Nscale, according to reporting by TechTimes. The Rubin platform's HBM4 memory was described as delivering roughly 22 TB/s of bandwidth per GPU - about three times the HBM3e bandwidth present in the Blackwell generation - which the research firm says translates to a meaningful performance-per-token improvement for hyperscalers.
SemiAnalysis did not present an entirely unqualified outlook. The firm flagged that NVIDIA's original four-chip Rubin Ultra configuration, which was introduced at GTC 2026, was canceled approximately three months after that announcement. It has been replaced by a version that is half the original physical size, with performance scaled down proportionally. NVIDIA has not publicly commented on that design change, and SemiAnalysis said the long-term revenue implications of the downgraded configuration relative to the initially announced specification remain unclear.
If SemiAnalysis's supply-chain read proves to be accurate, it would amount to a substantial upside revision to expectations heading into NVIDIA's Q2 FY2027 results. That earnings report is tentatively scheduled for August 26, 2026. Current consensus models project EPS of $2.07 on revenue of roughly $91.7 billion for the quarter; SemiAnalysis's findings imply that those consensus figures could be conservative.
Market context and implications
The SemiAnalysis projection has immediate relevance for markets sensitive to data-center hardware demand and memory supply dynamics, including investors focused on semiconductors, cloud-service providers, and the broader AI hardware ecosystem. A higher-than-expected ramp of Rubin units and the normalization of HBM4 supply would influence revenue trajectory for NVIDIA and could affect ordering patterns across server integrators and hyperscalers.
At the same time, uncertainty about the Rubin Ultra design change leaves open questions about the magnitude of future revenue tied to the highest-end configurations.