On July 29, Qualcomm presented a report card that left the market uncertain how to grade it.

The financial report shows that for the third fiscal quarter of fiscal 2026 (ending June 28, 2026), Qualcomm's revenue was $9.95 billion, down 4% year-over-year, slightly exceeding Wall Street's expectation of $9.62 billion. Non-GAAP earnings per share were $2.21, down 20% year-over-year and 1 cent below expectations. After-hours trading saw the stock initially fall before narrowing, still down over 4%.

The paradox of this report lies in three contrasting pictures: looking at mobile phones, revenue plummeted 20% year-over-year to just over $5 billion, marking the worst quarter in recent years; looking at automotive, a record 61% growth was achieved; and looking at data centers, the company confirmed that custom chips will begin shipping by the end of this year. Three entirely different report cards from the same company—which one should be trusted?

Mobile Phones: Qualcomm's Old Foundation is Cracking, but Not Collapsed

Mobile chip revenue was $5.086 billion, down 20% year-over-year. From the peak of $7.6 billion in the first quarter of fiscal 2025, it has dropped by one-third over three quarters. The root cause is not on the demand side. Global smartphone shipments are indeed declining—IDC data shows a 6.7% year-over-year drop in the second quarter, with two consecutive quarters of decline—but not a cliff-like drop. What truly constrains Qualcomm is memory chips. Memory chip costs surged about 300% year-over-year this quarter. AI data centers have absorbed HBM and DRAM production capacity, leaving mobile phone OEMs unable to secure supply or afford prices. The result is reduced production and inventory clearing, along with cuts to Qualcomm's chip orders. Qualcomm CEO Cristiano Amon was quite direct on the earnings call: consumers are starting to buy lower-tier versions within high-end models, or simply opting for last year's models. Even Qualcomm's most stable high-end Android base is loosening. However, the situation is not so linear. Management clearly stated that orders from Chinese OEMs bottomed in the third quarter and will return to double-digit sequential growth in the fourth quarter. Samsung's Galaxy S26 series will still use Qualcomm's custom Snapdragon 8 Elite Gen 5. The high-end phone base has not collapsed; it's just temporarily suppressed by rising memory prices. Mobile phone revenue's share of QCT has dropped from 67% a year ago to 60%. According to the company's investor day roadmap, this figure will fall to one-third by fiscal 2029. Mobile phones won't disappear, but their weight at Qualcomm is being systematically diluted.

Automotive: Not Just a Proof of Concept, It's Actually Making Money

The financial report shows automotive chip revenue of $1.588 billion, up 61% year-over-year, setting records for several consecutive quarters. The annualized revenue has already reached $6.35 billion. Three months ago, the company set an annual target of $7 billion, which many analysts privately thought was optimistic. Now, it seems Amon's numbers are likely achievable. The logic is not complicated. The chip value per vehicle for a traditional fuel car is about $100, while a smart electric vehicle, incorporating digital cockpit and ADAS solutions, can range from $500 to $1,500 per vehicle. Qualcomm's Snapdragon Digital Chassis and Snapdragon Ride platform are perfectly positioned at the throat of this transition. This quarter, new customers include BMW and Stellantis. Previously, mass-production customers included Mercedes-Benz, General Motors, Ford, Honda, Hyundai, and BYD. The design win pipeline has reached $65 billion, offering high revenue visibility. The only slight concern is gross margin. Automotive chips have high customization and long verification cycles, resulting in gross margins 5 to 10 percentage points lower than mobile chips. As automotive revenue's share continues to rise from 17%, it will drag down QCT's overall gross margin. This is not a major issue but requires ongoing monitoring. Management has raised the fiscal 2029 automotive revenue target from $8 billion to $10 billion. Given the current growth rate and design win conversion pace, this is not an exaggeration.

Data Centers: Ambitious Vision, but the Window Hasn't Been Broken Through Yet

At the investor day on June 24, Qualcomm raised its non-mobile phone revenue target for fiscal 2029 from $22 billion directly to $40 billion, with data centers accounting for $15 billion. This breakdown involves three steps: First, custom chips. Multi-generation orders from two hyperscale cloud vendors have been signed, with the first product shipping before year-end. Management calls this a "multi-year partnership," not a one-off deal. Second, standard chips. The Dragonfly C1000 server CPU was unveiled at the investor day, featuring the self-developed Oryon architecture and focusing on energy efficiency. Meta is the first customer, but mass production won't begin until 2028. Third, software. The acquisition of Modular (valued at $3.9 billion) brings its MAX inference platform and Mojo language to compete with Nvidia's CUDA. The acquisition of Alphawave Semi at the end of last year ($2.4 billion) fills the gap in high-speed interconnect. The scope is large, but timing is an issue. Connectivity chips can contribute revenue this year, custom chips start next year, universal AI accelerators won't arrive until the second half of next year, and server CPUs won't come until the second half of the following year. Goldman Sachs gave a "neutral" rating after the investor day, with a 12-month target price of $145, lower than the current stock price. Its comment: there is still a long way between early partnerships and significant revenue. Frankly, this judgment is not conservative. The market's valuation of the data center business bets on fiscal 2028 to 2029, with a two-year execution window in between where any misstep in any part will lead to discounts.

Apple: This Headwind is Coming Faster Than Expected

Among all the headaches for Qualcomm, the progress of Apple's self-developed baseband chip is the most underestimated by the market. In this financial report, Qualcomm issued its first clear warning: revenue from Apple products is accelerating its decline. Management stated on the earnings call that Qualcomm's share of baseband chips in the upcoming iPhone will be "far below the previously expected 20%," and Apple revenue may be halved sequentially in the fourth quarter. For the past five years, discussions about "when Apple will ditch Qualcomm's baseband" have been ongoing, but each time, technical difficulties have stalled progress. Now, this process is materially accelerating. Let's do the math: Apple contributes between $1.5 billion and $2 billion in chip revenue (baseband plus RF front-end) to Qualcomm each quarter. A drop in share from about 20% to below 10% means a loss of $6 billion to $8 billion in high-margin revenue annually. Amon's response: non-mobile QCT revenue growth will exceed 60% in fiscal 2027, directly covering the impact of Apple's loss. The automotive, IoT, and data center lines must run simultaneously. Whether you believe this depends on your view of the growth rates of these three businesses. Personally, I think the automotive line is the most stable, while data centers carry the most uncertainty.

Price Increases: A Test with Little Room for Retreat

Qualcomm announced a company-wide price increase effective September 1, with a double-digit percentage increase. Amon's wording: "Costs have gone up, so prices must follow." This is not Qualcomm's first price hike, but it feels different this time. Previously, price increases relied on process node upgrades, making it natural to charge more for new products. This time, it's broad-based inflation in wafers, packaging, memory, and materials, which the company can no longer absorb internally. Qualcomm does have pricing power in Android flagship chips. Samsung, Xiaomi, OPPO, vivo, and Honor have almost no alternative to Snapdragon for their flagship models. This group of customers is likely to accept the price increase and pass it on to consumers. However, there is a reflexive problem. A 300% increase in memory, a double-digit increase in chips, and another round of price increases at the terminal level could push mid-to-low-end phones up by 300 to 500 RMB. The mid-to-low-end market is already price-sensitive; will volume further decline? Volume and price are pulling against each other. The financial report shows that gross margin in the third quarter of fiscal 2026 was 53.77%, down from 55.56% a year ago. The effect of the price increase will take two quarters to become clear, with answers coming no earlier than the first quarter of fiscal 2027. Qualcomm currently has a market cap of about $180 billion. Mobile phones set the floor, automotive and IoT provide medium-term certainty, and data centers determine how high the valuation can bounce. These three businesses are in completely different stages, but all face a concentrated verification window from the end of this year to early next year. In the December quarter, whether mobile phones will rebound as management said, and whether custom chips can ship on time—the answers to these two questions will carry more weight than any target number from the investor day.