The Last Quarter Can Change the Story
Nine months can tell investors a lot about a year.
The final stretch can change the narrative.
By the fourth quarter, investors have already had months to digest earnings, economic data, interest-rate expectations, Washington policy and the investment themes that dominated much of the year.
Some stories have strengthened.
Some have become crowded.
Some companies have delivered impressive operating results but may face increasingly difficult expectations.
Others are approaching important tests that could determine how investors think about their businesses beyond 2026.
That is what this watchlist is designed to find.
Not seven stocks predicted to soar.
Not seven “must buys.”
And not a collection of whichever companies happen to have generated the biggest headlines.
Instead, this report looks for something more useful:
Companies entering the final stretch of 2026 with an important question still waiting to be answered.
Can the enormous AI infrastructure buildout continue translating into demand for the companies supplying the physical equipment?
Is the AI semiconductor story expanding beyond the names that dominated the first phase of the cycle?
Can America’s electricity infrastructure keep pace with extraordinary new demand?
Is commercial space becoming a larger defense and communications business?
Will autonomous vehicles weaken transportation platforms or make them more valuable?
Can a rapidly expanding restaurant concept maintain its economics while opening more locations?
And what can America’s largest retailer tell investors about the health of the consumer?
Those questions take us across seven very different companies:
Vertiv Holdings (NYSE: VRT) The AI Infrastructure Test
Marvell Technology (NASDAQ: MRVL) The AI Silicon Test
GE Vernova (NYSE: GEV) The Power Demand Test
Rocket Lab USA (NASDAQ: RKLB) The Space Economy Test
Uber Technologies (NYSE: UBER) The Autonomy Test
CAVA Group (NYSE: CAVA) The Consumer Growth Test
Walmart (NASDAQ: WMT) The American Consumer Test
Think of this report as a Q4 research board.
Every company has a thesis.
Every thesis has supporting evidence.
Every company also has risks that could challenge the story.
And each has specific developments worth monitoring through the final stretch of 2026.
Let’s put them on the board.
WATCH #1 | VERTIV HOLDINGS
The AI Infrastructure Test
THE Q4 QUESTION
Can AI infrastructure demand keep translating into physical equipment orders?
Artificial intelligence tends to be discussed in terms of chips and software.
But an AI data center is still a physical building filled with machines that consume enormous amounts of electricity and produce enormous amounts of heat.
Those servers need power.
They need cooling.
They need backup systems.
They need electrical distribution.
They need equipment capable of operating reliably around the clock.
That puts Vertiv in an interesting position within the AI investment cycle.
The company provides power-management, thermal-management and infrastructure systems used in data centers and other critical environments.
It does not manufacture the AI chip.
It helps create the environment in which those chips can operate.
That distinction makes Vertiv one of the more useful companies for investors trying to determine whether AI spending is spreading beyond semiconductor manufacturers and into the physical infrastructure underneath the technology.
WHY IT MADE THE BOARD
Vertiv entered 2026 with an estimated combined order backlog of approximately $15 billion, compared with $7.2 billion at the end of 2024.
The company cautions that backlog should not automatically be interpreted as future sales and that customer orders can be delayed or reduced under certain circumstances.
Still, the magnitude of the increase provides one measure of the demand being placed on data-center infrastructure suppliers.
Vertiv then reported second-quarter 2026 net sales of approximately $3.27 billion, 24% higher than the comparable quarter in 2025. Adjusted operating margin reached 22.6%, an improvement of 410 basis points from the prior-year period.
The important part for this watchlist is not simply that Vertiv has grown.
It is why.
The computing density inside AI data centers can create different power and cooling requirements than traditional computing environments.
Advanced accelerators are packed into increasingly powerful systems. More power consumed in a smaller physical space creates additional heat.
That creates an engineering problem.
And engineering problems can create infrastructure demand.
THE NUMBER THAT GOT OUR ATTENTION
$15 billion
Vertiv reported approximately $15 billion of combined order backlog at the end of 2025.
Again, backlog is not guaranteed revenue.
But comparing that figure with the $7.2 billion reported one year earlier provides a useful indication of how quickly customer commitments expanded during the AI infrastructure buildout.
THE BULL CASE TO RESEARCH
The favorable research case rests on a relatively simple idea:
AI infrastructure spending may be broader than chips.
If hyperscalers, cloud providers and enterprises continue building increasingly dense computing environments, suppliers of power and cooling equipment could continue participating in that spending cycle.
Vertiv could also benefit if next-generation computing systems require more sophisticated cooling technologies.
Traditional air cooling may not always be sufficient for extremely dense systems.
Liquid-cooling technology has therefore become increasingly important to the data-center discussion.
That could make thermal management almost as important to AI infrastructure research as processors themselves.
THE BEAR CASE TO RESEARCH
Strong demand creates another problem:
Expectations.
When investors already expect rapid growth, good results may not be enough.
Data-center construction could also slow if customers adjust capital spending, AI economics disappoint, power availability becomes constrained or projects are delayed.
Backlog introduces additional risk.
Vertiv explicitly warns investors that customers may cancel, reduce or defer orders in certain circumstances.
A large backlog therefore provides evidence of demand, but it should not be treated as guaranteed future revenue.
Competition, manufacturing capacity, supply-chain execution and pricing also deserve attention.
Q4 TRIPWIRES
Researchers may want to follow:
- New orders relative to completed shipments
- Backlog conversion
- Data-center capital spending
- Demand for liquid-cooling systems
- Operating-margin trends
- Capacity expansion
- Any indication that major customers are slowing AI infrastructure projects
WATCHLIST VERDICT
RESEARCH PRIORITY: HIGH
Vertiv represents one of the clearest ways to examine whether the AI investment cycle continues moving from silicon into physical infrastructure.
The question isn’t whether AI remains popular.
The more useful question is whether customers continue committing real capital to the power and cooling equipment required to run it.
WATCH #2 | MARVELL TECHNOLOGY
The AI Silicon Test
THE Q4 QUESTION
Is AI semiconductor demand broadening beyond the most obvious chip companies?
The AI semiconductor story is often reduced to one product:
The graphics processor.
But building an enormous AI data center requires considerably more than accelerators.
Data has to move between processors.
Servers need to communicate.
Networks have to operate at extraordinary speeds.
Optical connections become increasingly important.
Cloud companies may also want custom chips designed around their specific workloads.
That is where Marvell becomes interesting.
Marvell develops semiconductor technology used across data centers, networking, storage and communications infrastructure.
Its AI exposure therefore sits in a different part of the computing stack.
WHY IT MADE THE BOARD
Marvell reported approximately $8.2 billion in fiscal 2026 revenue, an increase of 42% from the prior fiscal year.
Data-center revenue exceeded $6.1 billion and represented roughly 74% of total company revenue.
The company attributed much of the growth to AI-related demand across custom silicon and electro-optics.
The trend continued into fiscal 2027.
For the three months ended August 1, 2026, Marvell reported approximately $2.74 billion in revenue, 37% higher than the comparable period one year earlier.
Sales into the data-center end market increased 46% year over year during that quarter, with the company again pointing to AI-related demand.
That makes Marvell a useful test of whether AI spending continues spreading through the networking and connectivity layer of the data center.
THE NUMBER THAT GOT OUR ATTENTION
74%
Approximately 74% of Marvell’s fiscal 2026 revenue came from its data-center end market.
A decade earlier, the company looked very different.
That concentration shows how dramatically Marvell’s business mix has shifted toward data infrastructure.
THE TECHNOLOGY TO UNDERSTAND
AI systems don’t operate as isolated chips.
Thousands of processors may need to work together.
That requires enormous amounts of data to move quickly between computing systems.
The larger AI clusters become, the more important networking and optical interconnect can become.
Marvell also participates in custom silicon.
Instead of purchasing only standardized processors, large cloud companies can design chips optimized for particular workloads.
That creates an interesting long-term question for the semiconductor industry.
Does AI computing remain dominated by general-purpose accelerators?
Or does a larger portion of the market shift toward specialized chips and customized infrastructure?
Marvell gives investors exposure to that debate.
THE BULL CASE TO RESEARCH
The favorable scenario is that AI infrastructure becomes increasingly heterogeneous.
Different workloads may require different processors.
Networking requirements may continue increasing.
Optical connections may become more important as computing clusters scale.
Hyperscalers may continue investing in custom silicon.
Under that scenario, AI semiconductor spending could broaden rather than remain concentrated in a small number of accelerator suppliers.
THE BEAR CASE TO RESEARCH
Marvell’s data-center concentration cuts both ways.
Strong AI demand has driven a significant portion of its growth.
That means a slowdown in hyperscaler capital spending could matter considerably.
Customer concentration is another issue to understand.
Custom semiconductor programs can depend on a relatively small number of very large customers.
Competition is also intense.
Large semiconductor companies, specialized chip designers and even cloud providers themselves are developing technology for many of the same markets.
Q4 TRIPWIRES
Researchers may want to follow:
- Data-center revenue growth
- Custom silicon design wins
- Optical interconnect demand
- Hyperscaler capital spending
- Customer concentration
- Gross-margin trends
- Competitive developments in AI networking
WATCHLIST VERDICT
RESEARCH PRIORITY: HIGH
Marvell offers a way to ask a broader question about the AI trade.
Not whether artificial intelligence continues expanding.
But which parts of the semiconductor ecosystem capture the spending as AI infrastructure becomes larger and more complex.
WATCH #3 | GE VERNOVA
The Power Demand Test
THE Q4 QUESTION
Can the power system keep up with the computing boom?
Artificial intelligence has created an unusual market connection.
A software revolution has become an electricity story.
AI data centers can require enormous amounts of power.
At the same time, manufacturing investment, electrification and aging infrastructure are adding additional pressure to the grid.
Electricity demand therefore connects several investment themes that might otherwise appear unrelated.
Artificial intelligence.
Natural gas.
Nuclear power.
Transmission.
Transformers.
Grid modernization.
GE Vernova operates across much of that system.
Its businesses include gas power, nuclear services, grid equipment, wind technology, electrification and power conversion.
That makes the company less a bet on one source of electricity and more a case study in the infrastructure required to produce and move power.
WHY IT MADE THE BOARD
GE Vernova reported second-quarter 2026 orders of $24.2 billion, an 88% organic increase from the comparable period.
The company’s backlog increased by approximately $13 billion sequentially during the quarter.
Its Electrification segment reported $6.3 billion of orders, with management citing strong demand for grid equipment.
Electrification equipment backlog reached approximately $40.6 billion, 69% above the comparable year-earlier level.
Those numbers provide evidence that the electricity problem is not theoretical.
Utilities, developers and other customers are committing capital.
THE NUMBER THAT GOT OUR ATTENTION
$40.6 billion
That was the equipment backlog reported in GE Vernova’s Electrification segment for the second quarter of 2026.
The segment includes technology used in power transmission and grid systems.
The figure was 69% above the comparable year-earlier amount.
THE TECHNOLOGY TO UNDERSTAND
Generating electricity is only half the problem.
Power also has to travel from where it is produced to where it is needed.
That requires transformers, switchgear, substations, transmission lines and increasingly sophisticated grid-management technology.
Data centers complicate the equation because they can require unusually large and concentrated amounts of electricity.
A large project cannot simply be connected anywhere.
Transmission capacity matters.
Grid stability matters.
Generation availability matters.
Permitting matters.
Equipment availability matters.
That means the AI power story can become a bottleneck story.
THE BULL CASE TO RESEARCH
GE Vernova could potentially participate across several different approaches to solving the electricity problem.
More natural-gas generation can create turbine demand.
More nuclear generation can support service activity.
More renewable generation can require grid connections.
More electricity moving across the system can require additional transmission equipment.
This diversity may be important.
The political debate can change over which energy sources deserve the most support.
The need to move electricity remains.
THE BEAR CASE TO RESEARCH
GE Vernova’s businesses do not all share the same economics.
Its Wind segment, for example, reported weaker order and revenue trends during the second quarter of 2026 than Power and Electrification.
Large infrastructure projects can also face delays.
Permitting, financing, labor, supply chains and customer decision-making can slow deployment.
And strong order growth can create its own execution challenge if manufacturing capacity struggles to keep pace.
Q4 TRIPWIRES
Researchers may want to follow:
- Electrification orders
- Grid-equipment backlog
- Gas-turbine orders
- Data-center-related demand
- Manufacturing capacity
- Wind profitability
- Utility capital spending
WATCHLIST VERDICT
RESEARCH PRIORITY: HIGH
GE Vernova may be one of the most useful companies for investors trying to understand the connection between AI and electricity.
The Q4 research question is no longer simply how much computing companies want to build.
It is whether the power infrastructure can support it.
Q4 CHECKPOINT
Three companies into the watchlist, one theme is already clear.
AI is no longer only a semiconductor story.
VRT: The data center needs power and cooling.
MRVL: The processors need networking and connectivity.
GEV: The entire system needs electricity.
That creates a useful research chain:
More computing → More data movement → More heat → More electricity → More infrastructure
But Q4 is not only about AI.
The next company takes us much farther from the data center.
WATCH #4 | ROCKET LAB USA
The Space Economy Test
THE Q4 QUESTION
Can Rocket Lab become much more than a launch company?
Rocket Lab is best known for Electron, its small orbital launch vehicle.
But describing Rocket Lab simply as a rocket company misses a large part of its strategy.
The company also manufactures satellite components and spacecraft systems.
It operates launch infrastructure.
It participates in government and defense programs.
And it has been working toward a broader model in which more pieces of the space supply chain sit inside one organization.
That makes Rocket Lab an interesting test of how the commercial space industry is evolving.
WHY IT MADE THE BOARD
Rocket Lab reported second-quarter 2026 revenue of $234 million, 62% above the comparable period one year earlier.
Its backlog for that quarter was approximately $2.36 billion, 137% higher year over year.
The company also reported more than 90 launches in its launch backlog when combining Electron, HASTE and Neutron commitments cited with its second-quarter business update.
Those numbers demonstrate that Rocket Lab’s story has grown beyond occasional launch announcements.
The company is attempting to build scale across launch and space systems.
THE NUMBER THAT GOT OUR ATTENTION
$2.36 billion
Rocket Lab reported that backlog for the second quarter of 2026.
Backlog can change and does not guarantee that every contracted dollar becomes revenue.
But it provides one indication of the amount of contracted work sitting behind the company’s expansion.
THE TECHNOLOGY TO UNDERSTAND
A space company can participate in several different layers of the market.
Launch: Put something into orbit.
Satellite components: Build pieces used inside spacecraft.
Spacecraft: Build the satellite itself.
Constellations: Operate networks of satellites.
Applications: Sell services using those networks.
Companies that control more layers may be able to capture more of the economics.
They also take on more complexity.
Rocket Lab’s strategic direction has increasingly moved toward vertical integration.
That makes execution particularly important.
THE BULL CASE TO RESEARCH
Government demand for space systems could remain important.
Space has become deeply connected to communications, navigation, defense, missile warning and intelligence.
Commercial operators are also building increasingly large satellite constellations.
If Rocket Lab successfully participates across both launch and satellite infrastructure, its addressable market could look different from that of a pure launch provider.
THE BEAR CASE TO RESEARCH
Space is capital intensive.
Development programs can be delayed.
Launch vehicles can fail.
Government awards can shift.
Acquisitions can create integration challenges.
Rocket Lab is also developing Neutron, a significantly larger launch vehicle than Electron.
Moving from development to dependable commercial operations represents a major technical and financial challenge.
High expectations can magnify the impact of delays.
Q4 TRIPWIRES
Researchers may want to follow:
- Launch cadence
- Space Systems revenue
- Government contract wins
- Backlog conversion
- Neutron development milestones
- Capital requirements
- Acquisition integration
WATCHLIST VERDICT
RESEARCH PRIORITY: ELEVATED / HIGH RISK
Rocket Lab combines several powerful themes:
Space.
Defense.
Satellite infrastructure.
Launch.
But those opportunities come with substantial execution risk.
This is exactly the kind of company a watchlist is designed for.
The story is interesting.
The outcome is not predetermined.
WATCH #5 | UBER TECHNOLOGIES
The Autonomy Test
THE Q4 QUESTION
Are autonomous vehicles a threat to Uber or the next expansion of its platform?
For years, autonomous vehicles were treated as an existential threat to ride-sharing companies.
The logic seemed straightforward.
If the driver disappears, why would an autonomous vehicle company need Uber?
But another possibility has emerged.
Autonomous-vehicle developers may be extremely good at robotics without wanting to build every part of a global transportation marketplace.
They still need customers.
Demand forecasting.
Payments.
Routing.
Fleet utilization.
Local market knowledge.
And a massive installed base of people already opening an app when they need a ride.
That could make Uber either an intermediary that autonomous vehicles eliminate or a distribution platform they increasingly use.
The answer remains one of the more interesting unresolved questions in transportation technology.
WHY IT MADE THE BOARD
Uber reported 3.9 billion trips during the second quarter of 2026, an 18% increase from the comparable period in 2025.
Gross bookings reached approximately $58 billion, while revenue totaled approximately $14.2 billion.
The company also reported $1.9 billion of GAAP operating income for the quarter.
Those numbers matter because Uber approaches autonomy from a very different position than it did during its earlier years.
It has a large existing marketplace generating substantial operating activity.
Autonomy does not have to rescue the business.
It can potentially become another supply source inside the platform.
THE NUMBER THAT GOT OUR ATTENTION
3.9 billion trips
Uber facilitated approximately 3.9 billion trips during the second quarter of 2026.
That scale is central to the autonomy thesis.
An autonomous-vehicle developer needs vehicles.
Uber already has demand.
The question is whether those two pieces become increasingly valuable to each other.
THE TECHNOLOGY TO UNDERSTAND
Autonomous transportation has two distinct problems.
The first is technological:
Can a vehicle safely drive itself?
The second is economic:
Can a fleet of autonomous vehicles stay sufficiently utilized to justify the capital required to build and operate it?
A robotaxi that sits unused generates no fare.
A transportation marketplace can potentially help match vehicle supply with riders.
That is where Uber’s platform becomes relevant.
THE BULL CASE TO RESEARCH
Uber could potentially become an aggregator of autonomous transportation rather than having to develop all of the underlying self-driving technology itself.
Different autonomous-vehicle operators could plug into Uber’s marketplace.
Uber could provide demand.
Partners could provide vehicles.
If that structure develops, autonomy may expand Uber’s supply network rather than destroy it.
THE BEAR CASE TO RESEARCH
The threat has not disappeared.
A highly successful autonomous-vehicle company could choose to own the customer relationship itself.
Large technology companies possess substantial capital and may decide that controlling both the vehicle and the marketplace creates better economics.
Autonomous rides could also carry very different margins than human-driven rides.
Regulatory requirements, insurance, fleet maintenance and local restrictions add additional complexity.
Q4 TRIPWIRES
Researchers may want to follow:
- Autonomous-vehicle partnerships
- AV deployments through Uber
- Mobility trip growth
- Gross bookings
- Operating cash generation
- Take rates and platform economics
- Evidence that AV developers prefer direct distribution over third-party platforms
WATCHLIST VERDICT
RESEARCH PRIORITY: HIGH
Uber is interesting precisely because autonomy creates both a threat and an opportunity.
That makes the Q4 question unusually clean:
Does self-driving technology weaken Uber’s moat, or deepen it?
WATCH #6 | CAVA GROUP
The Consumer Growth Test
THE Q4 QUESTION
Can rapid restaurant expansion continue without weakening the underlying business?
CAVA brings a completely different type of growth story to this watchlist.
There are no AI chips.
No rockets.
No power turbines.
No robotaxis.
The question here is much more familiar:
Can a restaurant concept successfully expand across the country?
That sounds simple.
It isn’t.
Restaurant growth can look impressive when a company is opening locations quickly.
The difficult part is maintaining traffic, restaurant economics and brand appeal as the store base grows.
CAVA gives investors an interesting case study because expansion and same-store performance have both contributed to its growth.
WHY IT MADE THE BOARD
CAVA reported second-quarter 2026 revenue of approximately $365 million, an increase of 31.3% from the comparable period in 2025.
Same-restaurant sales increased 9%.
Importantly, guest traffic accounted for 5.3 percentage points of that increase, while menu pricing and product mix accounted for 3.7 points.
The company opened 17 net new restaurants during the quarter and ended the period with 476 locations.
Restaurant-level profit margin was 25.7%.
Those details matter.
A restaurant can increase comparable sales by raising prices.
Growing guest traffic provides a different signal.
THE NUMBER THAT GOT OUR ATTENTION
5.3%
Guest traffic increased 5.3% during CAVA’s second quarter of 2026.
For a consumer watchlist, traffic is particularly useful because it helps separate actual customer growth from sales increases created primarily through higher prices.
THE BUSINESS MODEL TO UNDERSTAND
Restaurant growth has two engines.
Existing locations need to perform.
And:
New locations need to open successfully.
If both happen simultaneously, revenue can grow quickly.
But rapid expansion creates risk.
New locations may perform differently in unfamiliar markets.
Labor gets harder to manage.
Supply chains become more complex.
Real estate matters.
Opening costs rise.
Brand consistency becomes more difficult.
This is why restaurant count alone is not enough.
Researchers should follow the economics underneath the expansion.
THE BULL CASE TO RESEARCH
CAVA’s favorable research case revolves around portability.
Can a Mediterranean fast-casual concept work across many different U.S. markets?
If new restaurants continue attracting customers while mature locations maintain traffic, the company’s store base could have room to expand substantially.
Digital sales are another factor.
CAVA reported that digital channels represented 39% of revenue during the second quarter of 2026.
That provides another channel for customer engagement, although third-party delivery can also affect margins.
THE BEAR CASE TO RESEARCH
Restaurant companies can become victims of their own success.
Rapid expansion requires capital and management attention.
Food inflation can pressure margins.
Wages can rise.
New products can carry different economics.
CAVA’s restaurant-level margin decreased 60 basis points year over year during the second quarter despite strong revenue growth.
Management attributed the decline partly to input costs associated with a new menu item, higher third-party delivery mix and wage investment.
That illustrates why investors should examine more than headline growth.
Q4 TRIPWIRES
Researchers may want to follow:
- Guest traffic
- Same-restaurant sales
- New restaurant openings
- Restaurant-level margins
- Food and labor costs
- Digital sales mix
- Performance of newer geographic markets
WATCHLIST VERDICT
RESEARCH PRIORITY: ELEVATED
CAVA gives the watchlist exposure to a different question:
Can a high-growth consumer brand continue expanding while preserving the economics that made the expansion attractive in the first place?
WATCH #7 | WALMART
The American Consumer Test
THE Q4 QUESTION
What is America’s largest retailer telling us about the consumer?
Walmart is the least exotic company on this list.
That may be exactly why it belongs here.
Millions of people shop at Walmart.
Its stores sell groceries, clothing, electronics, household goods, pharmacy products and countless everyday necessities.
It serves customers across income levels.
Its digital business has also become much more important.
That gives Walmart unusual visibility into consumer behavior.
While CAVA can help researchers examine discretionary restaurant traffic, Walmart provides a much broader view of household spending.
WHY IT MADE THE BOARD
Walmart reported second-quarter fiscal 2027 revenue of approximately $187.9 billion, an increase of 5.9%.
Walmart U.S. comparable sales excluding fuel increased 2.6%.
Transactions increased 1.5%, while average ticket increased 1.1%.
Global eCommerce sales increased 23%, and Walmart U.S. eCommerce increased 24%.
The company’s global advertising business increased 38%.
Those numbers reveal why Walmart deserves to be studied as more than a retailer.
Its business model increasingly includes:
Retail.
Delivery.
Marketplace.
Advertising.
Membership.
Data.
And digital commerce.
THE NUMBER THAT GOT OUR ATTENTION
24%
Walmart U.S. eCommerce sales increased 24% in the company’s fiscal second quarter.
Store-fulfilled delivery was an important contributor, illustrating how Walmart’s enormous physical store footprint can also function as digital-commerce infrastructure.
THE BUSINESS SHIFT TO UNDERSTAND
For years, investors often framed retail as:
Amazon versus physical stores.
Walmart complicates that distinction.
A Walmart store can function simultaneously as:
A retail location.
A grocery store.
A pickup point.
A fulfillment center.
A delivery hub.
A marketplace gateway.
And a source of customer data that supports an advertising business.
That changes the economics of the store base.
The physical footprint can support the digital business instead of simply competing with it.
THE CONSUMER SIGNAL
Walmart’s scale makes it especially interesting during periods when investors are debating consumer strength.
What are shoppers buying?
Are they trading down?
Are higher-income households shopping at Walmart more often?
Are transaction counts increasing?
Are discretionary categories weakening relative to groceries?
Is eCommerce gaining share?
Those patterns can provide clues about household behavior that reach well beyond Walmart itself.
THE BULL CASE TO RESEARCH
Walmart’s scale provides several potential advantages.
Its purchasing power can support competitive pricing.
Its store network can enable rapid delivery.
Its advertising business can create higher-margin revenue.
Marketplace expansion can broaden assortment without requiring Walmart to own every piece of inventory.
Membership can deepen customer relationships.
Together, those businesses can potentially make Walmart structurally different from the retailer investors knew a decade ago.
THE BEAR CASE TO RESEARCH
Scale does not eliminate risk.
Consumers can become more cautious.
Tariffs and inflation can affect product costs.
Aggressive price investment can pressure margins.
Labor and healthcare expenses can increase.
Competition remains intense across physical and digital retail.
Walmart also reported inventory growth during its fiscal second quarter, making inventory management another area worth monitoring.
Q4 TRIPWIRES
Researchers may want to follow:
- U.S. transaction growth
- Average ticket
- Grocery versus discretionary sales
- eCommerce growth
- Advertising growth
- Inventory
- Consumer trade-down behavior
- Margin trends
WATCHLIST VERDICT
RESEARCH PRIORITY: HIGH
Walmart isn’t on the Q4 board because investors need another reason to recognize the world’s largest retailer.
It is here because Walmart can act as a consumer dashboard.
If household behavior changes meaningfully, there is a good chance some evidence will appear in Walmart’s numbers.
THE Q4 CONTROL PANEL
Seven companies.
Seven different questions.
Here is the entire research board at a glance.
Note: “Research Risk” reflects how much this report’s own thesis depends on an unresolved question and how much could go wrong with it — not a rating, score, or recommendation on the security itself.
| Company | Primary Theme | Q4 Question | Research Risk |
|---|---|---|---|
| VRT | AI Infrastructure | Can equipment demand keep pace with AI expectations? | Elevated |
| MRVL | AI Semiconductors | Is AI spending broadening through the chip ecosystem? | Elevated |
| GEV | Power Infrastructure | Can electricity infrastructure keep up with demand? | Moderate |
| RKLB | Space & Defense | Can Rocket Lab scale across more of the space stack? | High |
| UBER | Autonomous Vehicles | Does autonomy threaten or strengthen the platform? | Moderate |
| CAVA | Consumer Growth | Can expansion continue without weakening economics? | Elevated |
| WMT | Consumer Health | What is household spending telling investors? | Lower |
These are research classifications describing how unsettled each company’s Q4 question is — not ratings, scores, or recommendations on the securities themselves.
THREE BIGGER THEMES TO CARRY THROUGH Q4
The seven companies may look unrelated.
Look closer and three larger market questions connect them.
THEME #1: AI HAS BECOME A PHYSICAL INFRASTRUCTURE STORY
AI started as a software story.
Then it became a semiconductor story.
Increasingly, it is becoming an infrastructure story.
Vertiv represents power and cooling.
Marvell represents connectivity and custom silicon.
GE Vernova represents the electricity system itself.
These three companies illustrate an important progression.
Building bigger AI models eventually requires building more physical infrastructure.
That means investors researching AI may need to spend as much time understanding electrical equipment, cooling and networking as they do understanding the models themselves.
The research question for Q4 is whether capital spending continues traveling through that chain.
THEME #2: TECHNOLOGY IS CHANGING OLD BUSINESS MODELS
Rocket Lab and Uber look like completely different companies.
But they share an important question.
Technology can change where value sits inside an industry.
Rocket Lab began with launch but is attempting to participate across more of the space stack.
Uber built a transportation marketplace around human drivers but is increasingly preparing for a world that could include autonomous fleets.
In both cases, investors need to ask:
Which part of the value chain matters most?
Owning the vehicle?
Owning the infrastructure?
Owning the customer?
Owning the network?
Owning several pieces at once?
The answer can determine which business models become more valuable as technology evolves.
THEME #3: DON’T FORGET THE CONSUMER
Technology can dominate headlines while households quietly determine the direction of enormous parts of the economy.
That is why CAVA and Walmart belong beside AI infrastructure and space.
CAVA gives researchers a focused view of restaurant traffic and expansion.
Walmart gives researchers a broad view across grocery, discretionary spending, digital commerce and household purchasing behavior.
If consumers become more selective, the effect will not appear identically at every company.
Some businesses can gain share during periods of caution.
Others can lose traffic.
The details matter.
HOW TO USE THIS WATCHLIST
This report is not designed to be read once and forgotten.
The most useful approach may be to revisit the seven research questions as new information arrives.
For every company, ask:
1. Is the Original Thesis Still Intact?
Do not confuse a moving share price with a changing business thesis.
The price can move without the underlying story changing.
The underlying story can also change before the price fully reflects it.
Keep the two questions separate.
2. Did the Key Metric Improve or Weaken?
For VRT, that might be orders or backlog.
For MRVL, data-center revenue.
For GEV, grid orders.
For RKLB, execution and backlog conversion.
For UBER, trips and autonomous partnerships.
For CAVA, guest traffic.
For Walmart, transactions and consumer mix.
3. Did Expectations Move Faster Than the Business?
A company can execute well and still disappoint investors if expectations become too aggressive.
This is particularly important with high-growth companies.
Strong fundamentals do not automatically mean an attractive setup at every valuation.
4. What Would Prove the Thesis Wrong?
This question deserves more attention than it usually gets.
If an investment thesis cannot be disproved, it is not much of a thesis.
Define the evidence that would cause the research case to change.
5. Is the Company Still Worth Your Attention?
A watchlist does not need to remain static.
New information can make a company more interesting.
It can also make it less interesting.
The purpose of research is not to defend the original idea.
It is to improve it.
THE FINAL-STRETCH CHECKLIST
As 2026 moves through its final quarter, several broader developments could influence multiple companies on this list.
AI Capital Spending
Does spending continue expanding into infrastructure, networking and power?
That matters for VRT, MRVL and GEV.
Electricity Availability
Power is becoming a constraint for certain data-center projects.
Any acceleration or slowdown in grid investment could influence the AI infrastructure discussion.
Consumer Spending
Watch transaction counts, traffic and discretionary spending.
That matters most directly for CAVA and Walmart, but consumer conditions can influence the broader economy.
Autonomous Vehicle Deployment
The important development for Uber isn’t another impressive demonstration.
It is commercial deployment and evidence about how autonomous vehicles actually interact with transportation marketplaces.
Space and Defense Spending
Rocket Lab’s government exposure means contract activity and program execution remain important.
Expectations
Some of the companies on this list operate inside extremely popular market themes.
That can be an advantage operationally and a danger from an investment perspective.
The stronger the narrative becomes, the more important valuation and execution become.
FINAL PERSPECTIVE
A good watchlist does not tell investors what will happen.
It tells them where to look.
That’s the purpose of these seven companies.
Vertiv can help investors understand whether AI infrastructure spending continues reaching the physical data center.
Marvell can help show whether semiconductor demand is broadening through networking, optics and custom silicon.
GE Vernova sits directly inside the electricity challenge created by rising power demand.
Rocket Lab provides a window into the expansion of commercial and government space infrastructure.
Uber sits at the intersection of an established transportation marketplace and autonomous technology.
CAVA offers a test of whether a rapidly growing consumer concept can maintain traffic and restaurant economics while expanding.
And Walmart may provide one of the clearest windows into the financial behavior of the American consumer.
There is no guarantee that all seven stories strengthen.
Some may weaken.
Some expectations may prove too optimistic.
Unexpected competitors may emerge.
Economic conditions can change.
That uncertainty is precisely why the watchlist exists.
The objective isn’t to predict every move before it happens.
It is to identify the questions worth asking before the answers become obvious.
Through the final stretch of 2026, these seven companies give investors seven very different places to start.
Q4 WATCHLIST AT A GLANCE
VERTIV | VRT
Theme: AI infrastructure Watch: Orders, backlog, cooling demand Core Question: Is AI spending continuing to reach physical infrastructure? Research Priority: HIGH
MARVELL | MRVL
Theme: AI silicon and connectivity Watch: Data-center growth, custom silicon, optics Core Question: Is AI semiconductor spending broadening? Research Priority: HIGH
GE VERNOVA | GEV
Theme: Electricity infrastructure Watch: Grid orders, gas turbines, backlog Core Question: Can power infrastructure meet rising demand? Research Priority: HIGH
ROCKET LAB | RKLB
Theme: Commercial and defense space Watch: Launch execution, contracts, Space Systems Core Question: Can the company successfully expand across the space value chain? Research Priority: ELEVATED / HIGH RISK
UBER | UBER
Theme: Autonomous transportation Watch: AV partnerships, trips, platform economics Core Question: Does autonomy strengthen or threaten the marketplace? Research Priority: HIGH
CAVA | CAVA
Theme: Consumer expansion Watch: Traffic, openings, restaurant margins Core Question: Can rapid expansion maintain attractive unit economics? Research Priority: ELEVATED
WALMART | WMT
Theme: Consumer health and digital retail Watch: Transactions, eCommerce, advertising, spending mix Core Question: What is the consumer telling us? Research Priority: HIGH
Disclaimer
This report is for informational and educational purposes only and does not constitute financial, investment, legal or trading advice. The companies discussed are presented solely as examples for further independent research and should not be interpreted as recommendations to buy, sell or hold any security.
Investing and trading involve risk, including the possible loss of principal. Companies operating in rapidly changing industries may face additional risks related to competition, customer concentration, capital requirements, regulation, execution and changing market expectations.
Backlog, orders, revenue growth, customer commitments and other operating metrics discussed in this report do not guarantee future financial performance. Historical operating results should not be interpreted as predictions of future results.
Readers should conduct their own research, review company filings and other primary-source information, consider their own financial circumstances and consult with a qualified financial professional before making investment decisions.
