Qualcomm Targets $15 Billion in AI Chip Sales by Fiscal 2029

Qualcomm has put a $15 billion number on its data center ambitions. The chipmaker told investors on Wednesday it expects $15 billion a year in data center chip sales by fiscal 2029, sending shares up 15% in after-hours trading to a $197.41 close. The forecast came at Qualcomm’s Investor Day in New York and landed alongside a stack of new targets: $40 billion in non-handset revenue by the same year, $10 billion from automotive chips, and adjusted earnings of more than $18 a share.

The pivot is the signal of the day. Handset chips still made up two-thirds of Qualcomm’s product revenue in the quarter ended in March. By fiscal 2029 the company says handsets will be a third of its chip revenue, with the rest split across data center, automotive, and other categories. ‘We will be truly diversified,’ CFO Akash Palkhiwala said on stage.

The Targets Qualcomm Put on the Table

The headline number is the data center target. Qualcomm said it expects $15 billion a year from data center chips by fiscal 2029, a figure that did not appear on any prior forecast and is roughly triple the $5 billion it expects the segment to bring in for fiscal 2027. The $5 billion figure for fiscal 2027 includes $1 billion from new custom-chip customers, the company said.

The $40 billion non-handset target is also a jump from the $22 billion Qualcomm had previously guided for. A second forecast covers automotive. Qualcomm said it had expanded its automotive design-win pipeline to $65 billion and raised its automotive chip revenue target to $10 billion by fiscal 2029. The third number is the bottom line, with Qualcomm guiding to GAAP diluted earnings per share of more than $14.50 and adjusted (non-GAAP) diluted EPS of more than $18.00 by fiscal 2029.

Analysts polled by LSEG have an EPS estimate of $15.26 for fiscal 2029, leaving the chipmaker’s adjusted target above consensus. The Investor Day release detailing the fiscal 2029 EPS targets is on Qualcomm’s investor site.

• Data center chip revenue: more than $15 billion by fiscal 2029 (vs. $5 billion guided for fiscal 2027)
• Non-handset chip revenue: $40 billion by fiscal 2029 (up from prior $22 billion)
• Automotive chip revenue: $10 billion by fiscal 2029; design-win pipeline $65 billion
• Non-GAAP diluted EPS: more than $18.00 by fiscal 2029 (LSEG consensus $15.26)

Who Bought In on Day One

Qualcomm announced two named customers and two unnamed ones for its new data center silicon. Meta has signed a multi-generation deal to use the Dragonfly C1000, a CPU Qualcomm built specifically for AI data centers, with production starting in 2028. Microsoft will use a different chip category Qualcomm is calling ‘High Bandwidth Compute’ or HBC, which relies on cheaper memory chips found in smartphones and laptops rather than the high-bandwidth memory used by Nvidia. Qualcomm also disclosed two more hyperscaler design wins for custom chips, with revenue starting before the end of this calendar year.

Qualcomm already has relationships with nearly every major hyperscaler through its smartphone chips, CFO Akash Palkhiwala said. The Dragonfly C1000 was designed for agentic AI workloads and built for power efficiency, according to Qualcomm’s presentation. The Investor Day announcement and Meta multi-generation deal are detailed in coverage of the event.

I have not had to push my way into hyperscale customers; they’ve been pulling us in.

Where the Forecast Meets the Skeptics

The forecast is a bet Qualcomm is making on three chip categories, all at once. At its Investor Day and in April, the company said it is working with customers on central processing units, inference accelerators, and custom ASICs. The Dragonfly C1000 is the CPU, the HBC chip a second lane, and custom ASICs for hyperscalers the third. That third lane has its own competitive history, with rivals Broadcom and Marvell running large ASIC businesses around similar deals.

The earlier ByteDance ASIC deal showed the lane was open to a new entrant. Qualcomm is also buying, not just building, with the Investor Day acquisition of Modular. The deal price was not disclosed.

Modular’s software lets AI applications run on a broad range of chip architectures. Qualcomm described the startup as an equivalent to Nvidia’s CUDA, the dominant programming layer for AI workloads. The chipmaker is also weighing a much larger AI chip move, with the in-progress Tenstorrent acquisition talks valued at $8 billion to $10 billion. That deal would add a fourth lane to the data center portfolio.

The biggest bet is the size of the number. By 2029, Bank of America’s analysts expect Qualcomm’s data center push to deliver modest revenue of $2 billion to $5 billion annually, while Qualcomm’s own target is three to seven times higher. Amon told investors the right framing on a late entrance is scale, execution, engineering, and supply chain, with the full data center targets and analyst skepticism laid out in coverage of the same event.

How Crowded the Field Already Is

Qualcomm’s pitch is that power efficiency, the company’s strength in phones, becomes a selling point in AI data centers where electricity is the bottleneck. The market is already crowded with much larger incumbents.

Nvidia dominates AI training and is pushing hard into inference. Cerebras Systems is a recent entrant. Amazon’s Graviton and Google’s Axion are already in production at hyperscalers. Broadcom and Marvell have grown large custom-silicon businesses for the same set of customers Qualcomm is now courting.

Why the push now? Smartphones, Qualcomm’s core business, is under pressure from two sides. A memory chip shortage, driven by surging AI infrastructure demand, is squeezing mid-range Android handset production, and major customers including Apple and Samsung are designing more of their own chips in-house. The smartphone market peaked in shipments in 2017, by industry estimates cited by CNBC, and the company’s $15 billion target bets that data center will carry the next leg of growth.

1. Nvidia: incumbent in AI training, now pushing into inference with its own CPU and accelerator lines
2. Cerebras Systems: inference-focused competitor that has raised its public profile in 2026
3. Amazon’s Graviton and Google’s Axion: in-production Arm-based server CPUs at the same hyperscalers Qualcomm is courting
4. Broadcom and Marvell: custom-silicon incumbents with established ASIC businesses at hyperscaler customers

What the Stock Is Asking Investors to Pay

Qualcomm’s stock has not sat still. Shares closed at $197.41 on Wednesday, up 15% for the year. The trailing P/E and GF Score breakdown for QCOM shows the price-to-earnings ratio at 21.45, higher than the five-year median of 17.46.

The data provider’s GF Value estimate for Qualcomm is $175.34, putting the close about 12.6% above that model. The GF Score 82/100 is a composite of five metrics. Profitability ranks 9 out of 10 and momentum 10 out of 10. Growth is the weakest score at 4 out of 10.

The earnings growth implied by Qualcomm’s own fiscal 2029 non-GAAP EPS target of more than $18 is roughly double the LSEG analyst consensus of $15.26 for the same year. Hitting the data center target alone would require landing multiple large customers, and the company has said two unnamed hyperscalers are already on board. The 15% after-hours jump came after a recent pullback from a record close. Whether the new forecast holds depends on whether the signed customers convert into shipping volume starting this year.

Insiders have sold $5.8 million worth of shares over the past three months, per GuruFocus’s insider data. The number is small relative to total volume but worth flagging on a stock where the momentum score just hit a 10.

• Trailing twelve months P/E: 21.45 (five-year median 17.46)
• GF Value estimate: $175.34 vs. $197.41 close, a 12.6% premium
• GF Score: 82/100; profitability 9/10, momentum 10/10, growth 4/10
• Insider selling: $5.8 million over the past three months

Frequently Asked Questions

What did Qualcomm announce on June 24, 2026?

At its Investor Day in New York on Wednesday, Qualcomm projected $15 billion a year in data center chip sales by fiscal 2029, a figure that did not appear on any prior forecast. The chipmaker also raised its non-handset revenue target to $40 billion for the same year and guided to non-GAAP EPS above $18. Qualcomm’s shares rose 15% in extended trading to a $197.41 close.

Who is using Qualcomm’s new data center chips?

Meta has signed a multi-generation deal to use Qualcomm’s Dragonfly C1000 CPU, with production starting in 2028. Microsoft will use Qualcomm’s new High Bandwidth Compute chips for AI workloads. Qualcomm also disclosed two unnamed hyperscaler customers for custom silicon, with revenue starting before the end of this calendar year. The Dragonfly C1000 was designed for agentic AI workloads and built for power efficiency.

How does Qualcomm’s $15 billion target compare with analyst estimates?

Bank of America analysts expect ‘modest revenue of roughly $2 billion to $5 billion annually’ from Qualcomm’s data center push by fiscal 2027-2028, well below the company’s $15 billion fiscal 2029 target. Analysts polled by LSEG have a fiscal 2029 EPS estimate of $15.26, below Qualcomm’s adjusted EPS target of more than $18.

Is Qualcomm stock overvalued after the rally?

GuruFocus’s GF Value model values Qualcomm at $175.34, putting the $197.41 close about 12.6% above that estimate. The trailing twelve months P/E ratio is 21.45, above the five-year median of 17.46. The GF Score is 82 out of 100, with momentum at a 10 and growth at a 4.

Who are Qualcomm’s main competitors in data center AI chips?

Nvidia is the incumbent in AI training and is moving into inference. Cerebras Systems, Amazon’s Graviton, Google’s Axion, and the custom-silicon businesses of Broadcom and Marvell all compete in the data center chip market Qualcomm is targeting. Arm Holdings, a Qualcomm supplier, also rose 5% on the news.

Disclaimer: This article is for informational purposes only and does not constitute financial advice. Investing in equities carries risk, including loss of principal. Consult a qualified professional before making investment decisions. Figures cited are accurate as of publication.