Supply and demand improve together + strong competitive outlook, NVIDIA's pre-earnings target price raised by Morgan Stanley to $206

According to the Zhitong Finance APP, artificial intelligence (AI) chip giant NVIDIA (NVDA.US) will announce its second-quarter results for fiscal year 2026 on August 27, Eastern Time. Morgan Stanley released a research report stating that NVIDIA's demand, supply, and competitive outlook for the next 12 months remain very favorable. Morgan Stanley maintains an "Overweight" rating on NVIDIA, raising the target price from $200 to $206. This target price represents an upside of nearly 13% compared to the stock's closing price on Monday.

Morgan Stanley pointed out that market expectations have already risen ahead of NVIDIA's earnings report, which is reasonable. The firm expects NVIDIA's second-quarter results and guidance to be strong. However, Morgan Stanley added that its optimism about NVIDIA is more focused on the future. The firm stated that while demand continues to improve, supply remains a limiting factor, and NVIDIA's revenue forecast depends on the production speed of the Blackwell chips. The firm expects NVIDIA's third-quarter revenue to benefit from accelerated shipments of Blackwell chips.

Morgan Stanley raised its revenue forecast for NVIDIA's second quarter from $45.2 billion to $46.6 billion, and its third-quarter revenue forecast from $51.3 billion to $52.5 billion; the full-year revenue forecast for fiscal year 2026 was raised from $264.6 billion to $273.2 billion, and the Non-GAAP earnings per share forecast was raised from $6.28 to $6.51.

Morgan Stanley noted that on the demand side, NVIDIA's customers used terms like "amazing," "unmet," and "huge" to describe demand during the earnings call. The tone of comments from hyperscale customers has changed: while they emphasized supply constraints at the end of last year, they now emphasize the surge in inference demand and the difficulty in keeping up with capacity deployment. This is a positive signal for the sustainability of NVIDIA's revenue growth, even as Blackwell chips begin to ship in large quantities.

The demand growth mainly comes from NVIDIA's four major hyperscale customers. In addition, the demand from second-tier cloud service providers and sovereign customers is also an underestimated force. For example, CoreWeave's capital expenditures are mostly concentrated in the second half of the year, with about 50% occurring in the fourth quarter. NVIDIA's customer base is expanding, and strong demand is no longer limited to top-tier customers.

On the supply side, several factors are improving. Rack assembly is accelerating, and Hon Hai expects its rack shipments to triple quarter-on-quarter in the third quarter. Morgan Stanley's Greater China hardware team expects the top four ODMs' rack shipments to double in the third quarter. For the year, rack shipments are expected to reach 34,000 units, corresponding to about 2.4 million GPUs, which means approximately $90 billion in revenue for NVIDIA. Moreover, this only includes the top four ODMs and does not account for other partners.

Bottlenecks in the testing phase are also easing. The delivery cycle for Advantest testing machines has shortened, helping to restore normal testing times. Morgan Stanley analyst Charlie Chan expects the testing volume for B200/B300 chips to increase from 1 million in the second quarter to 1.5 million in the third quarter. However, the predictions from the analysts who released this report are relatively conservative, forecasting that the testing volume for B200/B300 chips will reach 1.2 million by the third quarter and 1.42 million by the fourth quarter Morgan Stanley added that in the second quarter, supply was not only related to Blackwell chips. Currently, the production of Hopper chips has been halted, and although demand for these GPUs remains strong, it is difficult to determine whether there will be new purchases in July. The firm predicts that revenue from Hopper chips will decline by 50% quarter-on-quarter to around $2.8 billion in the second quarter. Additionally, the increase in shipments of B200 servers has compensated for some of the shortfall. Overall, the increase in rack shipment ratios and the introduction of B300 may bring slight benefits.

It is worth mentioning that in terms of the Chinese market, Morgan Stanley expects that NVIDIA's guidance for third-quarter performance will assume minimal contribution from the Chinese market. If license approvals accelerate, there is potential for revenue from the Chinese market to increase, but there remains uncertainty about whether large-scale customers in China will ultimately purchase H20 chips.

Morgan Stanley stated that three months ago, the firm was more optimistic than the market regarding improvements in NVIDIA's demand and supply. Despite the market's upward adjustments, the firm remains optimistic that NVIDIA will continue to increase its market share in 2025 and maintain about 85% market share in 2026, resisting competition from the ASIC sector and AMD (AMD.US).

Morgan Stanley noted that NVIDIA's market share has significantly increased this year, even among the largest ASIC users (such as Google), where spending on NVIDIA is expected to grow more than threefold, while spending on ASICs has only seen slight growth. With NVIDIA's R&D investment surpassing $15 billion and expanding into areas such as rack interconnect, software, and services, it is becoming increasingly difficult for competitors to develop ASICs that surpass NVIDIA in mainstream tasks. Although AMD plans to launch rack-level solutions in 2026, its interconnect technology UALink will be released later, by which time NVIDIA will have transitioned to the next-generation NVlink Rubin and provided customers with NVlink fusion technology. Therefore, while there may be a window for low-cost alternatives, the performance gap remains the biggest obstacle to competition