Because the begin of the AI growth in 2023, there have been a number of moments the place the narrative across the buildout and long-term potential of this rising expertise has swung sharply from exuberance to doubt. Over the previous few weeks, we seem to have entered one other a type of doubt phases.
However it is vital to not lose sight of how far this theme has come, not simply during the last three years, however even during the last three months. In April, equities appeared like they have been getting into a broader correction as geopolitical tensions flared and danger urge for food deteriorated. But simply weeks later, shares discovered their footing and rallied aggressively.
That transfer was led by expertise, AI-adjacent shares, and, most notably, semiconductors. The SOXX semiconductor ETF greater than doubled from these lows, whereas a number of the largest winners within the group, corresponding to Micron Expertise (MU), rallied greater than 300% from depressed ranges.
That sort of transfer naturally invitations a reset.
Picture Supply: TradingView
The “Narrative Pendulum” is an idea I picked up from analyst Alex Barrow, and I believe it’s a helpful framework for understanding this market (detailed right here). The essential concept is that even when a strong secular development stays intact, the market’s notion of that development can swing dramatically between extremes. Within the case of AI, traders transfer from believing the chance is almost limitless to worrying that the whole buildout is extreme, wasteful, or unlikely to generate enough returns.
That’s the place we at the moment are. Issues round overspending, capital misallocation, falling LLM prices, hyperscaler margins, and the last word return on invested capital are starting to weigh on the AI commerce. These issues will not be essentially deadly to the long-term thesis. In reality, they’re in all probability wholesome. Intervals of doubt assist cool the sort of speculative enthusiasm that may drive costs virtually straight increased and create a extra sturdy base for the subsequent leg of the cycle.
I proceed to imagine the AI growth has room to run, however a pause or correction right here wouldn’t be shocking.
That brings me to Qualcomm (QCOM), a serious participant within the semiconductor trade that, till not too long ago, has been finest generally known as the dominant power in cell chips. That is still a core enterprise for the corporate, however smartphones at the moment are a mature market. Consequently, Qualcomm has more and more been considered as a slower-growth, extra cyclical, and considerably commoditized semiconductor firm, not in contrast to how Micron was considered within the reminiscence area a bit over a 12 months in the past.
That notion might now be altering.
A few weeks in the past, on the firm’s investor day occasion, Qualcomm administration introduced a big pivot within the firm’s strategic course. Whereas the corporate had been hinting at a bigger position in AI during the last a number of months, the investor day made that shift much more express. Administration unveiled a broader slate of AI-related enterprise verticals, main hyperscaler relationships, and a way more bold imaginative and prescient for Qualcomm’s position within the AI infrastructure stack.
The important thing takeaway is that Qualcomm is just not merely making an attempt to enter the AI sector with one product. It’s making an attempt to place itself as a broader AI infrastructure platform.
That would embrace chips, connectivity, edge AI, inference capabilities, customized silicon alternatives, and information heart acceleration. In different phrases, Qualcomm seems to be transferring from being primarily considered as a mobile-chip firm to one thing nearer to an “AI manufacturing facility accelerator” — an organization that helps hyperscalers and enterprise clients construct, join, optimize, and scale the infrastructure required for AI workloads.
I’ve many ideas on this evolution, which I’ll element extra absolutely, however the timing of the announcement has been considerably unlucky within the quick to medium time period. Qualcomm unveiled this strategic pivot simply because the semiconductor narrative started to swing from exuberance again towards skepticism. The inventory initially reacted strongly to the information, however has since pale to multi-month lows.
In my opinion, that weak spot has much less to do with Qualcomm’s particular developments and extra to do with the broader trade pullback. The market is presently questioning the whole AI infrastructure commerce, and Qualcomm is being dragged into that reset regardless of probably having simply laid out one of many extra essential strategic transitions in its latest historical past.
If Qualcomm can efficiently execute on this pivot, the inventory might not need to commerce primarily as a mature mobile-chip firm. As a substitute, traders might start to revalue it as a broader AI infrastructure beneficiary with publicity to hyperscalers, edge AI, information heart acceleration, and next-generation compute demand.
The timing could also be unlucky, however the setup is changing into more and more fascinating.
Scope of Qualcomm’s Endeavors
The monetary targets alone present how bold Qualcomm’s AI pivot has develop into. Administration is concentrating on $5 billion in information heart income by fiscal 2027 and $15 billion by fiscal 2029, with the early ramp anticipated to return largely from customized silicon and connectivity earlier than the corporate’s accelerators and server CPUs develop into greater contributors.
That may be a main shift for a corporation nonetheless principally considered by way of the lens of smartphones.
On the heart of the technique is Qualcomm Dragonfly, the corporate’s new information heart platform. Dragonfly is just not one product, however a layered portfolio that features connectivity silicon from the Alphawave acquisition, customized silicon for hyperscalers, AI inference accelerators, and ultimately Oryon-based server CPUs. Within the information heart, Qualcomm expects the sequence to start with connectivity, transfer into customized silicon in early fiscal 2027, then AI accelerators within the second half of fiscal 2027, adopted by Oryon server CPUs in fiscal 2028.
The strategic logic is constructed round a serious shift in AI workloads. The primary section of the AI growth was dominated by coaching giant fashions, the place Nvidia’s GPUs and CUDA software program stack stay the usual. However the subsequent section could also be more and more pushed by inference, particularly as agentic AI methods start chaining collectively dozens of mannequin calls to finish extra complicated duties. That dramatically will increase the variety of inference requests and makes energy effectivity, reminiscence bandwidth and value per token much more essential.
That is the place Qualcomm believes it has a gap.
The corporate’s most essential technical announcement was Excessive-Bandwidth Compute, or HBC. Somewhat than counting on the normal mannequin of pairing accelerators with stacks of high-bandwidth reminiscence, Qualcomm is pursuing a “reminiscence first” structure that locations compute extra immediately beneath the reminiscence stack. The objective is to scale back the space information has to journey, enhance effectivity, decrease energy consumption and tackle one of many largest bottlenecks in AI inference.
Simply as essential is the software program announcement. Qualcomm’s acquisition of Modular will be the key to creating the entire technique work. {Hardware} adoption in AI is closely depending on the developer ecosystem, and Nvidia’s CUDA moat has made it troublesome for rivals to realize significant share. Cristiano Amon has framed the Modular acquisition as a possible Android or Linux second for AI infrastructure, the place a extra open, hardware-agnostic software program layer might scale back dependence on any single vendor.
That may be a highly effective concept. Somewhat than asking clients to desert Nvidia in a single day, Qualcomm can provide a software program platform that runs throughout Nvidia, AMD and Qualcomm silicon, whereas nonetheless making a pure path towards its personal accelerators over time. If it really works, Modular offers Qualcomm a way more credible method to enter the AI infrastructure market than {hardware} alone.
The corporate additionally added buyer validation to the roadmap. Microsoft is predicted to deploy Qualcomm’s HBC expertise in Azure, whereas Meta has dedicated to a multigenerational settlement for Qualcomm CPUs in its information facilities. Qualcomm additionally strengthened the software program story by way of a partnership with Hugging Face, giving builders a path to deploy open fashions throughout Qualcomm platforms.
Lastly, Qualcomm’s connectivity experience could also be considered one of its most underappreciated benefits. AI information facilities are more and more constrained not solely by compute and reminiscence, however by the power to maneuver huge quantities of knowledge throughout racks and clusters. By way of Alphawave, Qualcomm now has high-speed connectivity belongings which are already producing income, giving Dragonfly a present income stream whereas the broader AI platform develops.
Execution danger stays important. Qualcomm is getting into a crowded market with highly effective incumbents, and several other of an important merchandise won’t attain industrial scale till fiscal 2027 or fiscal 2028. However the scope of the announcement is difficult to dismiss. Qualcomm is just not merely including AI publicity. It’s trying to construct a full information heart platform across the economics of inference, the place energy effectivity, reminiscence bandwidth, customized silicon, software program openness and connectivity might develop into more and more essential aggressive benefits.
Picture Supply: Qualcomm
Qualcomm’s Auto Execution Extrapolated
Full disclosure, going into Qualcomm’s Investor Day, I had my doubts in regards to the firm’s foray into the AI information heart buildout.
The technical functionality was by no means actually the query. Qualcomm has lengthy been probably the most subtle chip designers on this planet, with deep experience in energy effectivity, connectivity, system integration and edge computing. The larger query was whether or not the corporate was just too late. In a market already dominated by Nvidia, more and more focused by AMD and aggressively pursued by hyperscalers’ personal inside silicon groups, it was honest to wonder if Qualcomm might carve out a significant place.
However the extra I have a look at the technique, the extra compelling it turns into.
Qualcomm is just not making a single guess on one AI chip. It’s taking a multi-pronged method throughout connectivity, customized silicon, AI inference accelerators, server CPUs and software program. That offers the corporate a number of methods to win. Some items of the portfolio might lag expectations, and that will not be shocking given the dimensions of the endeavor. But when even one or two segments meaningfully outperform, the general alternative might nonetheless develop into materials.
I view the whole mission virtually as a strategic experiment. Qualcomm is placing a number of merchandise into the market, testing the place hyperscaler demand is strongest, and positioning itself across the areas the place AI infrastructure is almost certainly to evolve subsequent. Administration might not describe it that method explicitly, however I believe it’s the proper method. The AI information heart market remains to be younger, and the economics are altering rapidly. Somewhat than making an attempt to foretell the whole future with one product, Qualcomm is constructing a platform broad sufficient to adapt because the market develops.
That method turns into extra credible when considered by way of the lens of Qualcomm’s latest success in automotive.
Information from the automotive section can get misplaced when administration is asserting one thing as thrilling as AI information heart infrastructure, however the execution there could also be one of the best mannequin for what Qualcomm is making an attempt to do now. The automotive enterprise didn’t emerge in a single day. Qualcomm entered by way of connectivity, expanded into the digital cockpit, after which moved deeper into superior driver help and broader automobile compute.
That layered technique has labored. Automotive has rapidly grown into considered one of Qualcomm’s most essential non-handset companies, crossing a $5 billion annualized income run price in fiscal Q2 2026, with administration anticipating to exit fiscal 2026 above a $6 billion run price. That’s not a facet mission. It’s changing into an actual enterprise line and a significant proof level for Qualcomm’s diversification technique.
The parallel to AI infrastructure is essential. In automotive, Qualcomm didn’t have to personal the whole automobile to create worth. It wanted to determine the elements of the automobile the place compute, connectivity and software program have been changing into extra essential, then increase its content material over time. In information facilities, the identical logic might apply. Qualcomm doesn’t have to displace Nvidia throughout the complete AI stack to succeed. It wants to search out the areas the place its benefits matter most.
That’s the reason the inference-first focus is so essential. Qualcomm is just not making an attempt to win yesterday’s AI infrastructure battle. It’s making an attempt to place itself for the subsequent section of the market, the place energy effectivity, reminiscence bandwidth, connectivity and value per token develop into extra essential as AI workloads scale from coaching into large-scale inference. These are precisely the sorts of engineering issues Qualcomm has spent a long time fixing.
This doesn’t remove execution danger. The information heart market is bigger, quicker transferring and extra aggressive than automotive. Nvidia’s ecosystem is entrenched, hyperscalers are more and more constructing their very own chips, and Qualcomm nonetheless has to show that its roadmap can translate into industrial deployments at scale.
However automotive reveals that Qualcomm can execute the sort of transition. It might probably transfer past handsets, construct a platform in an adjoining market, increase its content material over time and convert lengthy design cycles into significant income. That doesn’t assure success in AI infrastructure, nevertheless it makes the plan far simpler to take severely.
For traders, that could be the important thing level. Qualcomm’s AI information heart technique shouldn’t be judged solely as a late try to chase Nvidia. It ought to be considered as the subsequent take a look at of the identical diversification playbook that’s already working in automotive. If the corporate can repeat even a part of that success, the market could also be underestimating how completely different Qualcomm’s enterprise might look over the subsequent a number of years.
Qualcomm Inventory Breaks Down
The technical image in QCOM inventory gives a extra tactical view of the setup.
Again in Might, the inventory rerated considerably increased after the corporate teased a serious hyperscaler deal. From there, it constructed out a broad consolidation sample, however for the reason that full announcement, the inventory has traded decrease. During the last week, QCOM broke beneath a key stage of assist, mirroring the broader weak spot throughout the semiconductor sector.
Technical evaluation doesn’t present dependable forecasting skill by itself, however it could actually present the place giant orders have left footprints. That’s primarily what a “stage” represents: an space the place a significant quantity of shares have modified arms and the place patrons or sellers have beforehand proven up.
For now, QCOM stays beneath that breakdown stage, and the near-term downtrend seems intact. That makes the inventory harder for merchants on the lookout for a clear short-term entry. However at roughly 17x ahead earnings, and with a probably a lot bigger long-term AI infrastructure alternative starting to take form, the setup could also be extra enticing for traders on the lookout for an even bigger multi-year win moderately than merchants making an attempt to seize the subsequent short-term transfer.
The earnings revision image may helps a extra affected person view. Qualcomm presently has a Zacks Rank #3 (Maintain), reflecting earnings estimates which have been comparatively flat. Which means analysts will not be aggressively elevating expectations but, however in addition they will not be chopping estimates in a significant method. Within the context of a serious strategic pivot, that leaves room for upside if administration begins changing these bulletins into seen income alternatives.
If revisions begin to transfer increased, that might develop into an essential bullish catalyst. A pickup in estimate momentum would sign that analysts are starting to underwrite the AI information heart alternative extra immediately into their fashions, moderately than treating it as a longer-dated optionality story.
Finally, the subsequent main transfer in QCOM inventory seems closely tied to the broader semiconductor cycle. There should be draw back forward over the subsequent month if the group continues to unwind. However when the narrative pendulum lastly bottoms and the market begins to elevate the AI infrastructure theme once more, Qualcomm might emerge with a a lot stronger story than it had in prior cycles.
The inventory has damaged down technically, however the enterprise could also be breaking out strategically.
Picture Supply: TradingView
Backside Line on Qualcomm Inventory
Qualcomm’s AI information heart technique remains to be early, and execution danger stays excessive. The corporate is getting into a crowded market, a number of key merchandise are nonetheless years from scale, and the inventory stays caught within the broader semiconductor pullback.
However the announcement modifications the long-term story. Qualcomm is not only a mature mobile-chip firm on the lookout for incremental development. It’s trying to construct a broader AI infrastructure platform throughout inference, connectivity, customized silicon, software program and power-efficient compute.
For now, the technical setup is weak and earnings revisions stay flat, which helps the Zacks Rank #3 (Maintain). However that additionally leaves room for upside if analysts start elevating estimates as AI information heart income turns into extra seen.
Within the close to time period, QCOM should commerce with the broader semiconductor group. Over the subsequent a number of years, nevertheless, the larger query is whether or not Qualcomm can flip this roadmap into an actual second development engine.
The inventory is just not with out danger, however the setup is changing into rather more fascinating.
Past Nvidia: AI’s Second Wave Is Right here
The AI revolution has already minted millionaires. However the shares everybody is aware of about aren’t more likely to maintain delivering the most important income. AI’s second wave is transferring from infrastructure to implementation and these corporations are on the forefront of this transition, positioned to develop into what Amazon and Google have been to the web period.
QUALCOMM Included (QCOM) : Free Inventory Evaluation Report
Micron Expertise, Inc. (MU) : Free Inventory Evaluation Report
This text initially printed on Zacks Funding Analysis (zacks.com).
The views and opinions expressed herein are the views and opinions of the writer and don’t essentially replicate these of Nasdaq, Inc.
