Nobody missed a quarter. Nobody cut guidance. Qualcomm fell roughly 7% on June 1, 2026 because Jensen Huang walked onto a stage in Taipei and announced a chip.

That is worth sitting with for a second. The RTX Spark – also called the N1X – is a Windows-on-Arm processor Nvidia built with Microsoft and MediaTek. It is not shipping yet. Fall 2026 is the target. And yet QCOM pre-market touched -10% before settling around -7.3% on the session. The S&P 500 was up 0.2% that same day. The Nasdaq edged up too. This was not a macro flush. It was a single competitive signal and a market that had priced QCOM at a 52-week high of $259.92 just days earlier.

When you run up 70%-plus in a year and then the dominant AI chip company announces it is entering your newest market, the air comes out fast.

The part people skip: QCOM’s AI PC positioning was always a growth story layered on top of a mature smartphone modem business. Snapdragon X Elite gave the company something new to sell – a reason to trade at a premium multiple. Microsoft’s Copilot+ PC initiative was built heavily around Snapdragon silicon, and an exclusivity window reportedly kept Nvidia and others out for a while. That window has reportedly closed. Dell, HP, ASUS, Lenovo, and MSI are already committed to RTX Spark devices. That is the exact OEM roster Snapdragon X has been relying on.

Nvidia does not just bring a faster chip. It brings CUDA. TensorRT. Years of GPU driver maturity and a developer ecosystem that Qualcomm has never come close to matching. Snapdragon X has posted solid battery life numbers and respectable performance benchmarks, but app and driver compatibility have been friction points since day one. Nvidia walks in without that problem.

What is interesting is that Qualcomm’s actual Q2 2026 results – reported April 29 – were not bad. Total revenue came in at $10.6 billion, down 3% year over year. The chip segment, QCT, posted $9.1 billion. Handsets were the soft spot at $6.0 billion, down 13% year over year, but management was explicit that China Android shipments were being deliberately held below end demand to normalize channel inventory. They called Q3 the expected bottom for that. Automotive came in at $1.33 billion for the quarter, up 38% year over year, a new record, with the annualized run rate exceeding $5 billion for the first time. IoT added $1.73 billion, up 9%. The licensing segment, QTL, generated $1.4 billion at a 72% EBT margin. Non-GAAP EPS was $2.65, at the high end of guidance. Free cash flow ran $11.2 billion in fiscal 2024 and $12.8 billion in fiscal 2025. The company returned $3.7 billion to shareholders in Q2 alone – $2.8 billion in buybacks and $945 million in dividends – and authorized a new $20 billion buyback. Q3 guidance came in at $9.2 to $10.0 billion in revenue with non-GAAP EPS of $2.10 to $2.30. None of that screams distress.

So then the question becomes: was QCOM actually cheap before Computex, or had the market already priced in the AI PC story?

Here is where it gets honest. The stock ran nearly 40% in May alone. Heading into June 1 the forward P/E was somewhere in the 17 to 20x range depending on methodology – still a real discount to the semiconductor sector median near 37x, but not the beaten-down value stock it looked like earlier in 2026. The EV/FCF was sitting around 18 to 19x. Reasonable. Not screaming. The selloff partially corrects a premium that was tied to Snapdragon X optimism, and that optimism just got complicated.

Slight tangent worth filing away: some of the June 1 commentary mixed in the Arm Holdings licensing dispute as another risk layered on top. That framing is outdated. Qualcomm won that case. A Delaware jury ruled in its favor in December 2024. A federal judge issued a final ruling cementing that in September 2025. Arm filed an appeal in October 2025 but the legal read is broadly that an overturn is unlikely. The license cancellation threat that was once a serious overhang on Snapdragon X is largely off the table. Still worth monitoring, but it should not be driving a selloff thesis in mid-2026.

Here is where I am at on the actual risk. RTX Spark does not ship until fall. A brand-new platform on a new architecture with a new software stack takes time to earn real volume. Premium pricing, a thin initial SKU count, and the natural friction of a new ecosystem all slow adoption. Qualcomm does not lose its Copilot+ design wins overnight. The question is what the OEM SKU mix looks like 12 to 18 months from now – how many Lenovo and HP AI PC configurations feature RTX Spark versus Snapdragon X at equivalent price points. That data does not exist yet.

Meanwhile the automotive story keeps compounding. Thirty-eight percent growth per year at $1.33 billion per quarter is a serious business. Management has guided toward a $6 billion annual run rate exit for fiscal 2026. If that holds, automotive alone increasingly justifies a meaningful chunk of the current valuation without any contribution from AI PCs at all. That part of the business barely gets mentioned in the RTX Spark coverage.

Qualcomm also launched Dragonfly at Computex – a new AI data-center brand that got completely buried under the RTX Spark headlines. Details were light and deferred to Investor Day on June 24. That is the next real event to watch. If Dragonfly is substantive – real silicon, real customers, real revenue line – it adds another diversification leg the market has not yet assigned value to. If it stays vague, that is useful information too.

The things worth tracking from here are straightforward: how many RTX Spark SKUs actually launch in fall 2026 across those committed OEMs and at what price versus Snapdragon X configurations; whether Microsoft weights RTX Spark and Snapdragon equally in Copilot+ marketing or begins tilting; what Q3 handset numbers look like to confirm the China channel inventory bottom; and what Dragonfly actually is when Qualcomm has to show its cards on June 24.

A 7% drop on an announcement – not a miss, not a guidance cut – deserves a measured read. The business generating $12 billion in free cash flow annually with automotive growing 38% year over year is not the same as the business that just lost its only growth driver.

But QCOM was not cheap going into this. That matters. And Nvidia is a serious competitor with real advantages. That matters more.

June 24 is the first real test of whether management has an answer.