Earnings Preview | Revenue expected to increase by 27% but profits under pressure. Will MI308 licensing and AI price hikes be the engine for AMD's continued stock price rise?

According to the Zhitong Finance APP, AMD (AMD.US) will announce its second-quarter financial report after the US stock market closes on August 5 (morning of August 6 Beijing time). According to forecast data compiled by Visible Alpha, benefiting from strong sales in the data center business, AMD's second-quarter revenue is expected to reach $7.43 billion, a year-on-year increase of 27%. However, adjusted net profit may decline from $1.26 billion (earnings per share of $0.69) in the same period last year to $796.6 million (earnings per share of $0.48). The chip manufacturer's stock price has performed strongly recently, as the market expects that export restrictions to China may soon be eased. In May of this year, AMD warned that it would face a loss of about $800 million this quarter due to tightening restrictions on chip sales to China. However, Wall Street analysts remain cautiously optimistic about the stock. It is reported that AMD stated last month that it plans to resume exports of MI308 chips to China once it receives final approval from US authorities. UBS recently raised AMD's target price from $160 to $210, noting that the company is expected to obtain permission for sales of MI308 chips to China. Among the 10 analysts tracked by Visible Alpha, 6 rated it as "buy" or equivalent, while 5 recommended "hold," with a target price range of $111 to $210. The stock has risen over 40% so far this year, closing around $172 last Friday. Key Variables Due to the significant rise in AMD's stock price this quarter, there are various possibilities for its performance after the financial report, but based on an optimistic outlook, AMD's overall trend in the second half of the year is expected to continue upward. The financial report this quarter should focus on the following three aspects: MI350 Pricing Strategy Adjustment AMD plans to raise the price of the Instinct MI350 AI accelerator from $15,000 to $25,000, an increase of nearly 70%. The market interprets this as a signal of surging demand for AI products and may also mark the beginning of its gradual adjustment of low-price strategies. Even after the price increase, the MI350 still has a price advantage over NVIDIA's (NVDA.US) B200, which ranges from $30,000 to $40,000, which will help improve gross margins. Last quarter, due to declining sales, AMD's gross margin contracted by 40 basis points. For the upcoming mass-produced MI355X, the expected price is in the $30,000 range. Although management has not commented, this may indicate a resurgence in AI concept enthusiasm. HSBC pointed out: "The MI350 is already capable of competing with NVIDIA's Blackwell B200... AMD's AI chip sales next year are expected to rise significantly from the previous estimate of $9.6 billion to $15.1 billion." Erosion of Intel's Market Share Combined with Tariff Front-Loading Effect Although many parties denied it last quarter, the tariff front-loading effect does exist, and AMD is also difficult to remain unaffected. Intel's management has confirmed that the tariff front-loading has led to quarter-on-quarter growth, and AMD's financial report may show a similar impact. However, AMD's continued market share growth in the client and server CPU markets will offset some of the impact. Currently, AMD's revenue share in the server CPU market has reached one-third, while Intel has fallen from a near-monopoly position to 63%. IDC and Mercury Research expect AMD's server CPU share to reach 36% this year, with some forecasts suggesting it will exceed 40%, potentially reaching 60% by 2026 Although the situation is favorable, AMD faces strong challenges from the Arm architecture. According to Mercury Research, Arm's market share in the x86 market has surpassed double digits, reaching 11.9%. The report also notes that the surge in demand for AMD's client CPUs has driven this increase in market share. It can be said that all manufacturers, except Intel, have made gains. Overall, the momentum of market share growth resonating with the pre-tariff effect will boost AMD's client business performance. Bank of America has raised AMD's target price from $175 to $200, noting that if Intel's (INTC.US) market share continues to decline, AMD could capture over 30% of the CPU market by 2026, a significant increase from less than 20% in 2023. Restart of MI308 Chip Export Licenses to China The U.S. ban in April led to the suspension of AMD's MI308 and NVIDIA's H20 AI GPUs in China, resulting in AMD recording an $800 million impairment. However, the suspension of the ban this month has driven AMD's stock price up by 6%, outpacing NVIDIA's gains. The market expects that the $800 million inventory will be digested at "nearly zero cost, with high margins," but this expectation may be overly optimistic. AMD's Chief Financial Officer Jean X. Hu candidly stated during a Bank of America conference call: "There were indeed many noises and uncertainties in the second quarter. The MI308 export license issue had a significant impact on AMD, with the data center GPU business alone losing $700 million in revenue. At that time, there was a surge in GPU demand in China due to companies like DeepSeek." Although the market expects a rebound after the obstacles are removed, this is more about reinforcing AMD's AI narrative rather than validating short-term performance. Notably, AMD successfully attracted OpenAI CEO to its AI Day event in June and announced that seven of the top ten AI giants are using Instinct chips, including Tesla (TSLA.US). Oracle (ORCL.US) has even signed a multi-billion dollar contract to deploy a cluster of 30,000 MI355X GPUs. CEO Lisa Su revealed during the Q1 earnings call: "This quarter, we added several first-tier cloud service and enterprise customers, including a top large model developer that has adopted Instinct GPUs for most of its daily inference tasks." In terms of technical indicators, the MI355X has a memory bandwidth of 22.1TB/s, three times that of the B200; its FP16 computing power of 3000-4000 TFLOPS (dense) is gradually approaching NVIDIA's 5000 TFLOPS (sparse) level. Although NVIDIA remains the industry favorite, AMD's performance is commendable