From AI Beta to Realizing Profits: How to Find New Ways to Make Money in Q3 U.S. Stocks?

Let’s get straight to the point: U.S. stocks still have support in Q3, but if you want to keep making money in the market, the way you trade may need to change.

In the recently concluded Q2, the market’s pricing of geopolitical shocks became temporarily dulled; combined with a rebound in AI infrastructure and a recovery in risk appetite, this jointly drove U.S. stocks to regain strength—especially under the leadership of mega-cap tech stocks and core AI assets. At one point, the market returned to a familiar trading pattern: as long as capital expenditures keep rising, as long as demand for computing power remains strong, valuations can keep moving higher.

But once Q3 begins, this logic is facing a higher bar for verification.

On one hand, inflation is still above the Fed’s target, and long-end yields and the policy path continue to restrict the expansion room for high-valuation assets. On the other hand, AI-related companies’ stock prices have already built in fairly optimistic growth expectations. What the market needs to see next is no longer just bigger CapEx numbers, but orders, deliveries, gross margins, cash flow, and return on capital.

Therefore, the MSX Maitan Research Institute maintains a neutral-to-positive view on U.S. stocks in Q3.

The index hasn’t entered a systemic bear cycle yet, but the source of returns is shifting from “valuation expansion” to “earnings delivery.” AI is still the most important industrial main theme, but the trading focus will move from broad AI Beta further down to areas that are easier to verify via earnings reports—storage, networking and optical interconnects, power, cooling, data center delivery, and edge computing and Physical AI built around real-world applications.

If you had to summarize the Q3 market environment in one sentence: inflation caps the valuation ceiling, earnings determine the index floor, AI realization decides structural Alpha, and market breadth determines the quality of the rally.

I. After the retreat of valuation expansion, earnings must catch the index

From Q2 to Q3, the market’s dominant contradiction has clearly changed.

In Q2, the trading chain was relatively clear: geopolitical tensions affected oil prices and inflation expectations; the rate path adjusted accordingly; after risk appetite repaired, capital flowed back into AI and large-cap tech stocks. The core of the market’s trading was valuation repair after macro pressure eased at the margin.

In Q3, the contradiction propagates further downstream—especially inflation constraining valuations, the Fed reducing forward guidance, earnings needing to support the index, and AI having to move from CapEx to real delivery.

This doesn’t mean the market is about to turn bearish. A more accurate description is the return threshold is rising.

1. Inflation is still the ceiling for high-valuation assets

Q3’s first layer of constraint still comes from inflation and the Fed.

U.S. inflation remains clearly above the 2% long-term policy target, meaning the foundation for “rapid rate cuts propping up valuations” is not solid. At the same time, under Chairman Waller’s leadership, the Fed’s communication style places even more emphasis on real-time data, price stability, and policy discipline, weakening the market’s reliance on forward guidance it has grown accustomed to over the long term.

This will bring three impacts:

• The familiar “Fed put” is thinning: investors can no longer simply assume that as long as market volatility increases, policymakers will quickly release easing signals.
• The market’s sensitivity to individual data will increase further: CPI, PCE, wages, employment, oil prices, consumer data, and even corporate earnings reports could trigger a repricing of the rate path and valuations.
• The margin for error for high-valuation growth stocks will decline noticeably: the AI industry trend remains intact, but “being directionally correct” is no longer enough to support sustained valuation expansion. The market needs more evidence—orders, revenue, profit margins, and cash flow—to show that current valuations are not built on distant imagination.

Therefore, Q3’s macro “base color” is not a typical recession trade. It is a high-valuation market that still has growth support, but where interest-rate constraints always remain.

In this environment, capital will favor two types of assets: one is companies with strong earnings certainty, high cash flow quality, and sound balance sheets; the other is themes with low duration, resource attributes, or inflation-hedging capabilities—including gold, resources, power, and some high-cash-flow financial assets.

2. The index can still rise, but it can’t rely solely on a higher P/E

The most important support for U.S. stocks in Q3 still comes from corporate earnings.

Multiple Wall Street institutions continued to raise their U.S. stock targets in their mid-year outlooks. The core basis is not that valuations can expand indefinitely, but that corporate EPS still has further room to be revised upward.

This distinction is extremely important.

When market valuations are already at historically high levels, whether the index can move higher afterward no longer depends on whether investors are willing to pay higher multiples. Instead, it depends on whether corporate earnings can continue to grow above expectations. Goldman Sachs has raised its S&P 500 target for end-2026 to 8,000, and raised its EPS forecasts for 2026 and 2027 to 340 and 385 dollars, respectively.

Meanwhile, it expects forward valuations of U.S. stocks to remain roughly around 21x—already within the historical high range of the past 40 years.

In other words, the index’s upside ahead relies more on EPS than on further expansion of valuation multiples. If earnings seasons push EPS higher continuously, U.S. stocks still have a foundation for choppy upward movement. But if earnings upgrades start to slow down, while inflation or long-end yields rise again, the market could quickly switch from “earnings-driven” to “valuation compression.”

So, the most critical question in Q3 is not whether the index can go up, but whether, at current valuations, earnings can continue to support the index.

This also implies that investment thinking should not stop at passively chasing the index; it should shift more toward areas that can be verified by orders and earnings—including AI infrastructure, storage, power, data center infrastructure, industrials, financials, platform advertising, and consumer leaders with stable cash flow.

3. Market breadth will determine whether the rally is healthy

Besides index levels, Q3 also needs close attention to market breadth.

If U.S. stocks keep rising but the rally remains highly dependent on a small number of AI giants, market concentration will increase further. Any earnings miss could trigger more violent volatility.

A healthier market structure should be: AI continues to maintain the main theme, while industrials, financials, platform advertising, and parts of consumer sectors take over in turn.

In other words, Q3 cannot only be about whether Nvidia, the semiconductor index, or the Nasdaq hits new highs. It also needs to watch equal-weight indices, the number of advancing stocks, and whether earnings expectations for non-AI sectors improve in parallel.

AI determines how high the market can go, while market breadth determines how far this rally can run.

II. AI CapEx 2.0: From computing power scarcity to delivery realization

AI is still the most important industrial main theme in Q3, but the trading logic has shifted from “expectations” to “verification.”

In Q2, the market mainly priced computing power scarcity, upward CapEx revisions, and supply-chain expansion. As long as tech giants continue raising CapEx, and as long as GPUs remain in short supply, the industry chain can be revalued around ever-increasing demand.

But in Q3, the market will ask more directly:

• Can financing truly turn into GPUs and data centers?
• Can GPUs be turned into computing power that can be delivered on schedule?
• Can computing power form long-term stable revenue?
• Can revenue cover depreciation, financing costs, and equity dilution?
• Will it ultimately generate positive free cash flow and reasonable ROIC?

This is what people call AI CapEx 2.0. It is no longer a bet on a single chip, nor is it simply chasing a particular optical module. Instead, along the entire data center construction chain, it searches for the links that can truly realize orders and profits—such as chips and platforms → networking and optical interconnects → storage → power and cooling → server and system delivery → computing operations → edge and real-world applications.

1. Chips are still the entry point, but not the only answer

Among these links, chips remain the most important entry point for the AI industry.

NVDA is still the pricing anchor for global AI assets; AVGO corresponds to custom ASICs and networking platforms; MRVL benefits from both custom chips and optical interconnects; and TSM corresponds to advanced processes, advanced packaging, and the entire AI semiconductor manufacturing ecosystem.

But in Q3, the assessment of the chip layer will be stricter than before.

The market will not only care about chip performance—it will continue to ask whether orders can stay above expectations, whether advanced packaging and capacity bottlenecks can ease, whether the customer structure is healthy enough, whether gross margins can remain high, and whether inference, AI PCs, enterprise AI, and Edge AI can form new growth curves.

INTC needs to be understood under a different framework. It is not a direct substitute for NVDA, but more like a combined option on U.S. semiconductor security, server CPUs, AI PCs, Edge AI, and foundry business. Its logic is whether low-position assets can resonate with policy, industry, and fundamental repair.

2. The bigger the cluster, the more important networking and optical interconnects

The larger the GPU cluster, the more important the interconnects become.

In Q2, the market already priced optical modules, switches, and high-speed interconnects thoroughly. The focus in Q3 will shift from pure industry cycle sentiment to more detailed delivery quality—for example, whether 800G and 1.6T demand continues to be revised upward, whether order visibility is high enough, whether customer concentration is manageable, whether expansion plans and yields can keep up with demand, and whether silicon photonics, upstream materials, and specialty processes become new bottlenecks.

This layer is also one of the directions in which funds are easiest to spread from core AI leaders to second-tier assets.

When order visibility improves, optical communication, silicon photonics, and specialty materials companies often have both earnings elasticity and valuation repair room. Compared with companies that rely only on grand narratives, these firms are undoubtedly easier to verify through orders, capacity utilization, and earnings guidance.

ANET.M, CRDO.M, LITE.M, COHR.M, AAOI.M, FN.M, AXTI.M, and TSEM.M are important observation assets in this direction.

GLW.M is also worth including. It is not the purest optical module play, but its optical fiber, glass, and data center basic materials businesses allow it to benefit from rising data center connection density and infrastructure investment.

3. Storage is moving from an AI side theme to a core bottleneck

Storage is still the direction that Q3 needs to increase weighting on.

In the past, when the market talked about AI, the first thing it thought of was GPUs and networking. But as model parameters, inference calls, and data scale keep growing, AI consumption of HBM, DRAM, NAND, enterprise SSDs, and HDDs is also rising continuously.

Storage is no longer a side theme for the AI industry—it is becoming a core link in data center construction that is increasingly hard to bypass.

Micron’s recent earnings and guidance strengthened the view that “AI storage is entering the realization period.” In particular, the company’s revenue for fiscal Q3 2026 reached 414.56 billion dollars, Non-GAAP gross margin rose to 84.9%, and adjusted free cash flow was about 183 billion dollars. For the fourth fiscal quarter, the company provided revenue guidance of around 50 billion dollars with a fluctuation of plus or minus 1 billion dollars, and gross margin guidance of about 86%.

These figures show that the pull of AI on storage is no longer just about order expectations; it is starting to realize in sync as revenue, profit margins, and cash flow.

But storage trading in Q3 cannot continue to be understood simply as a “MU single-spot trade.” A more reasonable structure is to split storage into three tiers:

• The first tier is NAND, SSD, and HDD diffusion, including WDC.M, STX.M, and SNDK.M. They benefit from AI data volume growth, improved enterprise storage demand, and recovery in the traditional storage cycle, with relatively limited direct competitive overlap with HBM leaders.
• The second tier is MU.M, and Micron remains one of the most core storage assets in the U.S. stock market. It benefits from improvements across HBM, DRAM, and NAND as well. However, with SK hynix ADR’s plan progressing, MU’s premium as a “U.S. HBM scarcity proxy” may face partial diversion (extended reading: “During the Day, Watch Hynix; at Night, Trade U.S. Stocks: A New ‘Asia Session Barometer’ for Global AI?”).
• The third tier is controller and second-tier elasticity assets such as SIMO.M, which benefit from enterprise SSD, AI PC, and Edge AI storage diffusion. But their certainty and priority are temporarily lower than storage manufacturers and the HDD/NAND main line.

For the entire storage sector, SK hynix ADR is a classic double-edged sword.

On the positive side, it will reinforce the public-market pricing of the global HBM leader and increase investors’ attention to the entire storage industry. On the negative side, once U.S. stock investors have a more direct investment channel into the HBM leader, MU’s original scarcity-proxy premium could be partially weakened.

Therefore, Q3’s storage logic will move from “single-point scarcity” to “diffusion across the whole industry chain” gradually.

4. Data center infrastructure must be grouped separately

AI bottlenecks are expanding from “whether there are GPUs” to “whether there is power, whether there is a server room, whether there is cooling, and whether it can connect to the grid.”

This layer should no longer be simply classified under industrials or utilities. As AI capital expenditures gradually enter the real construction phase, power, thermal management, electrical equipment, construction delivery, and high-reliability components have already become part of the AI CapEx trade.

Data center infrastructure can be split into at least five layers:

• Power and thermal management: VRT.M;
• Electrical equipment and power distribution: ETN.M;
• Grid engineering and interconnection construction: PWR.M;
• Power generation and grid equipment: GEV.M;
• System delivery, PCB, connectors, and materials: DELL.M, SMCI.M, TTMI.M, APH.M, GLW.M.

The biggest advantage of this direction is that as AI capital expenditure pushes further toward real construction, it becomes harder to bypass data center infrastructure.

After all, compared with assets that rely only on valuations and narratives, data center infrastructure companies often have clearer backlogs, order cycles, and delivery schedules, making it easier to verify industry trends through revenue and cash flow.

5. From single-hardware to AI Factory

When the market shifts from purchasing a single piece of hardware to building a complete AI system, the importance of AI Factory, server delivery, high-end PCBs, and enterprise AI infrastructure will also rise further.

The criteria for this layer include whether orders are sustainable, whether products can be delivered on schedule, whether gross margins are stable, whether customers expand from a single large customer to more enterprises, and whether enterprise AI deployments can form scalable revenue.

DELL.M and SMCI.M both belong to the system-delivery direction, but their nature is not completely the same. Compared with SMCI, DELL’s business structure is more focused on enterprise AI, servers, and full system delivery, and the revenue-verification path is relatively clearer. SMCI has higher earnings elasticity, but its volatility, governance, and risk from expectation gaps are also more prominent.

Other directions worth watching include PENG.M and HPE.M.

6. Computing operators have the most elasticity, but also the highest verification hurdle

Computing operators are the layer with the greatest elasticity in the AI main line—and also the layer with the highest risk.

Companies in this category have the most straightforward growth story: receive financing, purchase GPUs, build data centers, and then generate revenue through long-term computing contracts.

But what the capital market ultimately needs to verify is whether this business model can actually run. This includes whether GPUs truly arrive, whether power and server-room delivery can be completed on time, whether the quality of customers’ long-term contracts is high enough, whether computing utilization can continue improving, whether depreciation, debt, and financing costs will erode profits, and whether equity financing will cause sustained dilution.

Therefore, for computing operators, the key words are not simply an “AI concept”—it is financing, delivery, customers, and cash flow.

From this perspective, NBIS.M, IREN.M, CRWV.M, and APLD.M still have relatively large event and earnings elasticity, but investors also need to apply a higher risk discount (extended reading: “When Meta Prepares to Sell Computing Power, Are the ‘Ghost Stories’ of the AI Bull Market Coming?”).

7. AI begins to move from the cloud to the edge and the real world

In the second half of Q3, AI trading may continue expanding from training and cloud computing toward inference, edge computing, and real-world execution.

Edge AI’s core lies in low latency, low power consumption, privacy protection, and real-time response. True large-scale AI adoption cannot stay entirely in the cloud. Phones, PCs, cars, cameras, robots, and industrial equipment all require local inference capability.

QCOM.M and ARM.M are more mature mappings on the edge side. INTC.M corresponds to AI PCs and edge CPUs. NOK.M can be placed within a framework of AI-RAN, dedicated wireless networks, and industrial edge connectivity.

NOK is not a typical AI chip stock, but its network infrastructure, AI-RAN, and industrial connectivity businesses give it a repair pathway distinct from core computing assets.

Physical AI includes robotics, autonomous driving, drones, warehouse logistics, and industrial automation.

The core of this direction is not only the robot itself, but also perception, control, execution, simulation, and safety systems. OUST.M, BB.M, TER.M, ROK.M, SYM.M, MBLY.M, TSLA.M, and ISRG.M can all map to this trend from different links.

But it needs to be emphasized that Physical AI currently is still closer to “narrative heating up and early order verification,” and has not yet entered the stage of full profit realization. In Q3, it is more important to focus on real customers, orders, mass production, and revenue—not simply trade concepts.

III. Beyond AI, who can become the next diffusion direction?

Q3 cannot only focus on AI.

If the index keeps climbing but the market still has only one industrial main theme, the rally will become increasingly crowded and more fragile. So a healthier structure should be: AI continues to maintain the main theme, while industrials, financials, platform advertising, consumer, supply chain security, and commercial space begin to provide new earnings and event elasticity.

1. Industrials, power, and financials: core observation targets for market breadth

Industrials and electrical equipment are also beneficial directions for AI infrastructure expansion.

GE.M, ETN.M, PWR.M, HON.M, and RTX.M can benefit from manufacturing, power grids, and capital expenditures, while also offering a valuation duration that is relatively lower than pure tech stocks.

Financials are also worth continued tracking.

AI private placements, IPOs, pre-IPOs, bond issuance, underwriting, and a recovery in trading activity will improve the business cycle environment of capital markets and benefit GS.M, MS.M, JPM.M, BAC.M, and HOOD.M.

However, in Q3’s main theme ranking, financials are not the most core first tier. Compared with AI CapEx, data centers, and storage, it is more suitable as a verification direction for market breadth and risk appetite recovery.

2. Platform advertising and cash-flow-oriented consumer sectors fit better in a high-rate environment

Platform advertising, cloud computing, and subscription businesses still have strong earnings resilience.

GOOGL.M, META.M, and AMZN.M each have advantages in advertising, cloud computing, and platform ecosystems. NFLX.M corresponds to subscription revenue, expansion in the advertising tier, and operating leverage of content platforms.

Consumer sectors need to be more selective.

Keeping high interest rates for longer will suppress some discretionary consumption and financing-sensitive companies. From this angle, Q3 is more suitable for focusing on consumer leaders with strong cash flow, strong pricing power, or platform and network effects—for example, COST.M, WMT.M, BKNG.M, and MCD.M.

3. Supply chain security: from short-term events to long-term premium

Objectively speaking, as of the time of writing, geopolitical risks have not disappeared—they have evolved from short-term oil price shocks into long-term industrial fragmentation and security premiums.

The scope of supply chain security is no longer limited to semiconductors. It also includes defense, critical minerals, power systems, and energy security:

• For semiconductor self-sufficiency, watch INTC.M, TSM.M, GFS.M, ASML.M, AMAT.M, LRCX.M, and KLAC.M;
• For defense and security, watch LMT.M, NOC.M, RTX.M, LHX.M, KTOS.M, and AVAV.M;
• For critical resources, watch MP.M, LAC.M, FCX.M, and CCJ.M;
• For power grids and power, continue to watch GEV.M, ETN.M, PWR.M, and VRT.M.

These assets may not all rise fully at the same time. But together, they reflect a long-term change: companies and countries are beginning to be willing to pay higher costs for supply chain redundancy, energy security, and critical infrastructure.

4. Commercial space: after the leader anchors pricing, second-tier assets must prove Alpha

Commercial space is still an important non-AI growth direction, but it shouldn’t be written as a simple “sector-wide rally” logic.

Once industry leaders establish a pricing anchor in the public market, second-tier commercial space companies need to prove their independent value through orders, launch counts, satellite deployments, government contracts, and recurring revenue.

SPCX.M remains the pricing center for the whole sector. RKLB.M, ASTS.M, PL.M, LUNR.M, RDW.M, IRDM.M, GSAT.M, BKSY.M, and SATL.M need to rely on execution in their respective businesses to form differentiated Alpha.

In other words, the logic of commercial space is shifting from “industry imagination” to “who can truly form sustainable revenue.”

Based on the framework above, we divided relevant assets into four tiers of observation priority, centered on four clues: earnings realization, AI CapEx bottlenecks, data center infrastructure, and low-position repair with event elasticity. This tiering is mainly used to present the research framework and tracking order. It does not represent deterministic return judgments and does not constitute investment advice:

• A+｜Core priority: clear main line, direct verification pathway—core observation assets for Q3;
• A｜High priority: closely matches the main line, clear catalysts, with relatively better conditions for earnings or valuation repair;
• A-｜Mid-to-high priority: the industrial logic holds, but still needs more orders, earnings reports, or event verification;
• B+｜High-elasticity observation: higher event and price elasticity, but also greater realization uncertainty and volatility risk;

IV. Three Q3 scenarios: opportunities remain, but indiscriminate chasing is no longer suitable

Overall, the MSX Maitan Research Institute believes that, around inflation, earnings, and AI realization, Q3 can be broken down into three main scenarios.

1. What are the three scenarios?

First is the Base Case, i.e., neutral to positive.

In the base scenario, PCE and CPI no longer continue to spike meaningfully. The Fed maintains data dependence. The earnings season pushes corporate earnings expectations up modestly. AI CapEx orders and deliveries continue to be realized.

At the same time, market breadth starts to spread from core AI leaders into industrials, financials, platform advertising, and data center infrastructure.

In this scenario, the index still has room for oscillating upside, but market style will shift from Mega-cap AI Beta further toward earnings realization, data center bottlenecks, and low-position elastic assets.

Second is the Bull Case, i.e., resonance between upward earnings revisions and falling inflation.

If oil prices fall, core services inflation cools, long-end yields fall in parallel, and AI, storage, power, and network orders continue to beat expectations, the market will enjoy a better risk-reward profile.

In this case, U.S. stock indices still have the possibility of making new highs. More importantly, the rally may no longer be limited to a handful of technology leaders. Data center infrastructure, industrials, cyclical growth, financials, and the capital markets chain could all see larger upside elasticity.

Last is the Bear Case, i.e., re-acceleration of inflation combined with AI realization coming in below expectations.

If oil prices, wages, rents, or hardware costs drive inflation to accelerate again, the market reprices to a higher rate path. Meanwhile, returns on AI CapEx start to face doubts. Earnings for storage, optical communications, or computing operators miss high expectations. High-valuation growth stocks will face a clear valuation compression.

In this scenario, volatility in high-elasticity small AI names could be significantly amplified. Market breadth narrows again, and funds flow back into high cash-flow, resource, and defensive assets.

2. Which indicators does Q3 most need to track?

To judge whether the Q3 baseline logic still holds, you can focus on the following types of signals:

• Macro data: PCE, CPI, core services inflation, wages, employment, retail sales, and oil prices;
• Rates and the Fed: 10-year Treasury yields, real yields, shifts in FOMC tone, and dot plot changes;
• Corporate earnings: S&P 500 EPS upgrade ratio, earnings Beat Rate, and profit margins;
• AI CapEx: tech giants’ capital expenditure guidance, GPU arrival, data center delivery, and power connection;
• Data centers: orders for power equipment, demand for liquid cooling, grid-connection cycles, server, PCB, and connector delivery;
• Storage cycle: HBM, DRAM, NAND prices, HDD demand, MU guidance, and SK hynix ADR progress;
• Optical interconnects: 800G, 1.6T, AEC, SerDes, and silicon photonics capacity;
• Market breadth: equal-weight indices, number of advancers, and profit expectation revisions in non-AI sectors;
• Capital markets: IPOs, secondary offerings, bond financing, trading volume, and investment-banking business recovery;
• Crowding: positions in semiconductors, AI infrastructure, and computing stocks, plus option-implied volatility.

But in terms of risk, it may not come from a single event. Instead, it could be that multiple variables shrink at the same time—for example, inflation rising again, the Fed taking a more hawkish stance, AI CapEx returns coming in below expectations, data center delivery delays, storage prices weakening, increased financing pressure for computing operators, and fund diversion caused by highly watched IPOs or ADRs from existing assets.

Meanwhile, crowding in semiconductors and AI infrastructure is already at relatively high levels. Once earnings reports only “meet expectations” rather than continue to massively beat them, stock prices may also experience significant volatility due to expectation gaps.

Therefore, the core of Q3 is not to avoid risk, but to raise the requirements for earnings realization quality.

Closing Thoughts

Q3 U.S. stocks are not short on opportunities. What truly changes is what kind of growth the market is willing to pay a premium for.

Over the past period, as long as you stood upstream of AI capital expenditures, and as long as you had enough scarce computing power, chips, or capacity, companies had opportunities for valuation re-rating. But when inflation keeps limiting valuation space and the interest-rate “safety cushion” gradually thins, the market ultimately needs to move from grand input numbers back to a set of financial statements.

Going forward, whether the index can keep moving higher depends on whether earnings can support current valuations. Whether the AI rally can continue depends on whether capital expenditures can, along the industrial chain, be sequentially transformed into orders, deliveries, revenue, free cash flow, and return on capital.

This also means that the AI main theme in Q3 has not ended—it is undergoing a more stringent internal filtering.

Chips are still the starting point, but no longer the only answer. Storage, optical interconnects, power, cooling, and system delivery are becoming the links that are easier to verify by earnings next. At the same time, whether industrials, platform advertising, financials, supply chain security, and commercial space can take over will determine whether this rally is merely an index boom driven by a few leaders—or a more sustainable diffusion of earnings.

From AI Beta to earnings realization, the main line has not ended.

A new round of repricing has already begun.