NVIDIA’s upcoming earnings report faces intense market scrutiny over the Rubin architecture rollout timeline, rollover financing structures, and forward market share projections.
NVIDIA is scheduled to release its quarterly earnings report for the period ending in July next week. Given the pattern of recent quarters, the market has grown accustomed to “beat-and-raise” outcomes, where earnings exceed expectations and guidance is subsequently raised. However, investor focus has shifted beyond single-quarter performance to three critical long-term questions: whether the Rubin architecture can successfully succeed Blackwell, whether AI demand will continue to significantly outstrip supply, and the magnitude of financing and balance sheet commitments NVIDIA must undertake to secure future demand.
Fundamentals remain robust according to broker forecasts. According to the Zephyr Trading Desk, Jefferies released a preview report this week projecting NVIDIA’s revenue for the July quarter at $95 billion, surpassing the consensus estimate of approximately $91.9 billion. For the October quarter, Jefferies expects revenue guidance to reach $108 billion, also exceeding the market expectation of around $103.7 billion. Morgan Stanley offered a more conservative outlook, forecasting July quarter revenue at approximately $91.2 billion and the October quarter at about $102.3 billion. Nevertheless, it similarly anticipates that recent trends will persist, with quarterly revenue potentially continuing to increase by roughly $10 billion quarter-over-quarter. Consequently, the core of the current market debate centers on whether the growth trajectory for 2027 and 2028 can be revised upward following these strong results.
In the near term, demand for Blackwell remains robust. Microsoft and Amazon’s cloud divisions demonstrated notable improvement during the quarter, while long-term computing power commitments from entities such as OpenAI and SpaceX continue to underpin market confidence in NVIDIA’s data center revenue. Morgan Stanley noted that although Rubin has not yet made a substantial contribution, NVIDIA’s current growth momentum is already strong, with greater acceleration likely to materialize after the January quarter next year.
Rubin emerges as one of the most pivotal themes for the upcoming earnings conference call. Jefferies argues that the ramp-up speed for Rubin could significantly outpace Blackwell’s early stages. This accelerated adoption stems from Rubin’s continuation of the 72-GPU Oberon NVL72 rack form factor, which allows NVIDIA to leverage the liquid-cooling data center deployments and manufacturing infrastructure already established for GB200 and GB300. Furthermore, the new generation design drastically reduces manual cabling and pipeline installation. Management previously indicated that assembly time for compute trays is expected to plummet from approximately two hours to just five minutes. The underlying implication is that NVIDIA is systematically upgrading its AI chips into replicable, scalable “AI factory systems.”
Jefferies projects that VR/R200 will account for roughly 12% of NVIDIA’s GPU revenue in the third quarter of fiscal 2027, climb to over 40% by the fourth quarter, and become the dominant revenue source in the first quarter of fiscal 2028. If this trajectory holds, the market’s valuation logic for NVIDIA will fundamentally shift from asking “how long can the Blackwell cycle last” to evaluating “whether Rubin can unlock the next round of growth on a higher base.”
Beyond GPUs, the Vera CPU represents an incremental growth area warranting close attention in the financial report. Jefferies estimates that NVIDIA’s total CPU revenue in fiscal 2027 could reach $20 billion, encompassing both Grace and Vera architectures. Their breakdown indicates that CPU revenue tied directly to GPU sales would be approximately $8.65 billion, while standalone Vera CPU revenue would stand at around $11.35 billion, pushing the annualized run-rate toward $32 billion by the fourth quarter. While CPU revenue cannot yet rival NVIDIA’s AI GPU business in absolute terms, serving primarily as a complementary enhancement, its strategic significance is substantial. The Vera CPU fortifies NVIDIA’s dominance in AI server and AI factory architectures through low-power LPDDR5X memory and NVLink coherent connectivity within the Rubin ecosystem. This integration will inevitably exert fresh competitive pressure on AMD and Intel, particularly as enterprise customers increasingly prefer purchasing complete racks or integrated systems. NVIDIA is effectively transitioning to selling a comprehensive computing platform that spans GPUs, CPUs, networking hardware, and software.
Of greater concern to the market is NVIDIA’s long-term computing capacity arrangement with OpenAI. According to Jefferies, NVIDIA has entered into an agreement with SB Energy, under which SB Energy will build, own, and operate the PORTS-Pike Technology Campus in Ohio. The facility will be leased to OpenAI for 20 years, with NVIDIA designated as the exclusive compute provider. NVIDIA will provide credit support for the initial 4.25 gigawatts (GW) of IT capacity and holds options for the remaining 3.75 GW, bringing the total potential capacity to 8 GW. Additionally, NVIDIA is investing $1.5 billion in SB Energy. The company stated that each generation of system deployment at the campus could correspond to approximately 1.5 million GPUs, generating between $150 billion and $200 billion in revenue. OpenAI’s existing and planned commitments total approximately 12 GW, expandable to 16 GW, which translates to roughly $600 billion in NVIDIA compute procurement by 2030.
This arrangement undoubtedly provides a clearer anchor for NVIDIA’s long-term order book. However, the market will inevitably scrutinize the nature of this demand: how much reflects customers’ voluntary capital expenditure, and how much requires NVIDIA to actively drive through credit support, direct investment, or revenue-sharing structures? Morgan Stanley believes that financing, market share, and gross margin a