Bill Ackman’s Three AI Stock Picks Reveal a Major Portfolio Rotation

Bill Ackman owns three prominent AI stocks, but the Google News headline misses the most important signal: his fund has radically changed its exposure.

Pershing Square increased its Amazon position and established a major Microsoft stake during the first quarter of 2026. It also retained Uber as a top holding. Meanwhile, the fund reduced its Alphabet position by roughly 95%.

That combination creates a more useful story than a simple list of stocks to copy. Ackman appears to be choosing established businesses that can fund AI development through existing operations. He is also willing to rotate between industry leaders when their valuations and risk profiles change.

The three names that best reflect that strategy are Amazon, Microsoft, and Uber. Amazon and Microsoft sell essential computing infrastructure. Uber offers a distribution network for autonomous transportation, without requiring one robotaxi developer to win every market.

These companies do not represent identical AI bets. They occupy different layers of the commercial stack, from computing capacity to business software and physical-world transportation.

That diversity matters because the central question surrounding AI stocks has changed. Investors are no longer judging companies only by model quality or product announcements. They increasingly want evidence that large capital investments can produce durable revenue and operating income.

Ackman’s activity should not be treated as a personal recommendation. A Form 13F is a delayed snapshot, and it excludes several asset classes. It also says nothing about an investor’s risk tolerance, purchase price, or complete strategy.

Still, the filing offers a window into how one concentrated investor is navigating an expensive AI market. The useful lesson is not to copy three ticker symbols. It is to understand why those businesses survived Pershing Square’s increasingly selective allocation process.

What the Google News Headline Leaves Out

Pershing Square did not simply identify three promising AI companies. It reallocated billions in reported equity value between competing AI platforms.

The fund’s quarterly filing covered holdings as of March 31, 2026. It reported 11 positions with an aggregate market value of approximately $13.7 billion.

Amazon represented about 17.4% of that reported portfolio. Uber accounted for approximately 15.7%. Microsoft, a new position during the quarter, represented about 15.3%.

Together, those three companies comprised almost half of the reported portfolio. That degree of concentration gives each selection far more significance than a small experimental position would carry.

Pershing Square added about 1.8 million Amazon shares during the quarter, bringing its disclosed position to roughly 11.5 million shares. It also acquired approximately 5.7 million Microsoft shares.

Uber remained one of the fund’s largest investments, although Pershing Square modestly reduced its share count. The holding still represented more than one-seventh of the disclosed portfolio.

The other side of the transaction is equally revealing. Pershing Square sold most of its Alphabet holdings, reducing both share classes by approximately 95%.

Ackman said the sale was not a negative judgment on Alphabet. He described it as a source of capital for the Microsoft investment, given Pershing Square’s finite resources.

That distinction matters. A concentrated fund cannot own every attractive company at a full position size. Each investment must compete with every other opportunity for capital.

The Google News framing turns that allocation problem into a shopping list. The filing presents a harder question: which company offers the best combination of business quality, AI exposure, and expected return at a specific valuation?

Pershing Square’s answer changed within a single quarter. Alphabet moved from a major position to a small residual holding. Microsoft became a core investment. Amazon grew larger, while Uber remained central.

That rotation also shows why following a famous investor can be difficult. Public filings arrive weeks after the quarter closes. The manager might have changed the portfolio again before readers see the disclosure.

Investors also lack Pershing Square’s complete decision model. They do not know its internal valuation assumptions, intended holding periods, hedges, or position limits.

A headline can accurately identify stocks that Ackman owns while obscuring the active judgment behind them. The important event is not ownership alone. It is the movement of capital toward companies that already control major distribution channels.

Amazon controls cloud infrastructure, online commerce, advertising, and a large logistics network. Microsoft controls enterprise software, developer tools, and one of the world’s largest cloud platforms. Uber controls a global marketplace connecting transportation demand with available supply.

Each company can embed AI into an existing customer relationship. That creates a different investment proposition from funding a stand-alone model provider that must build distribution after developing its technology.

Amazon Turns AI Spending Into a Full-Stack Business

Amazon combines AI infrastructure, custom chips, cloud services, retail automation, and consumer distribution inside one operating company.

The most direct part of the thesis is Amazon Web Services. AWS provides computing, storage, databases, model access, and tools that companies use to build and operate AI applications.

Amazon reported that AWS sales increased 28% year over year in the first quarter of 2026. The segment generated $37.6 billion in quarterly sales, according to the company’s first-quarter results.

AWS growth matters because training and running large models require substantial computing capacity. Inference, the process of using a trained model to generate an answer or prediction, can create recurring demand after deployment.

Amazon is trying to capture that demand at several layers. It rents Nvidia-based infrastructure, develops its own Trainium and Inferentia chips, and provides managed model access through Amazon Bedrock.

Custom silicon gives Amazon another path to control costs. The company said its chip business had exceeded a $20 billion annualized revenue run rate and was growing at a triple-digit pace.

Those figures are company-reported and do not isolate every economic contribution from AI. They nevertheless show that Amazon’s infrastructure strategy has advanced beyond pilot projects.

The opportunity also extends outside AWS. Amazon uses machine learning for inventory placement, demand forecasting, product recommendations, advertising, delivery routing, and warehouse automation.

These deployments offer a practical advantage. Amazon can test AI against measurable operating outcomes, such as delivery times, advertising conversion, robot movement, or the cost of serving a cloud workload.

That feedback loop makes Amazon more than an infrastructure supplier. It is also a large internal customer for its own technology.

Pershing Square’s interest appears tied to this combination of growth and operational improvement. Amazon can generate AI-related revenue through AWS while using similar technology to improve its retail margins.

The risk is capital intensity. Data centers, chips, networking equipment, and energy capacity require large commitments before customer demand becomes certain.

AI infrastructure can also become less differentiated. Microsoft, Google, Oracle, and specialized cloud providers compete for many of the same workloads. Customers can distribute applications across multiple providers to reduce dependence on one platform.

Custom chips introduce execution risk as well. Amazon must persuade developers that its hardware and software tools offer sufficient performance, availability, and compatibility.

Nvidia’s broad software ecosystem remains an important competitive advantage. Switching hardware can create engineering work that outweighs a lower computing cost.

Amazon’s diversified operations partly reduce that risk. Weakness in one AI product does not eliminate the value of its cloud relationships, logistics network, marketplace, or advertising business.

However, diversification can obscure results. Investors may struggle to determine which improvements came from AI and which came from pricing, scale, cost reductions, or broader economic conditions.

The clearest evidence will come from AWS growth and operating income. Rising demand accompanied by stable economics would support the infrastructure thesis. Growth driven by aggressive spending or lower returns would weaken it.

Ackman’s increased position suggests Pershing Square believes Amazon can convert capital expenditure into durable cash generation. That remains a judgment about future execution, not a fact established by recent growth alone.

Microsoft Is the Enterprise Distribution Bet

Microsoft’s advantage is not simply access to advanced AI models. It is the ability to place AI inside software that organizations already use.

Pershing Square made Microsoft a core position during the first quarter of 2026. The size of that purchase indicates a high-conviction allocation rather than a trial investment.

Ackman explained that Pershing Square established the position after Microsoft’s valuation fell closer to the broader market’s multiple. He also argued that concerns about AI disrupting traditional software had become excessive.

That logic contains a significant tension. Generative AI can strengthen Microsoft’s products, but it can also reduce the value of familiar software interfaces.

An AI agent is software that can plan and execute multiple steps toward a goal. If agents begin performing work across applications, customers might spend less time inside individual productivity tools.

Microsoft is responding by inserting its own assistants and agents across Microsoft 365, GitHub, Dynamics, security products, and Azure. It wants to own the layer through which employees instruct software.

The company also benefits when developers use Azure to train, customize, or run models. Microsoft therefore participates in both application revenue and infrastructure demand.

In its fiscal third-quarter 2026 materials, Microsoft said cloud and AI demand continued to drive results. Its investor disclosures provide the best evidence for whether that demand becomes sustained revenue rather than temporary experimentation.

Distribution is the central advantage. Many companies already manage employee identities, files, communications, source code, and business workflows through Microsoft products.

That existing position reduces the friction involved in offering an AI assistant. Microsoft does not need to persuade every customer to adopt an unfamiliar vendor before it can demonstrate a feature.

The same position creates governance advantages. Enterprise AI systems need permission controls, security monitoring, audit records, and access to approved data.

Microsoft already supplies many of those administrative layers. It can connect AI functions with a customer’s established identity and security policies.

Yet distribution does not guarantee valuable usage. An assistant bundled into common software might gain broad availability while receiving limited engagement.

Customers may also resist paying separately for features they consider incomplete. Organizations need evidence that AI improves output enough to justify deployment, training, security reviews, and process changes.

The strongest use cases are often narrow. A developer might use AI to explain unfamiliar code. A sales team might summarize a customer history. An analyst might compare documents and prepare a draft.

Knowledge workers still need reliable source material for those tasks. A well-maintained AI knowledge base can improve retrieval, but it cannot eliminate the need to verify outputs.

Microsoft also faces competition from Google, Amazon, Anthropic, OpenAI, Salesforce, ServiceNow, and many specialized software providers. Some competitors control their own models, while others focus on specific workflows.

Its relationship with OpenAI brings valuable technology and significant strategic complexity. Microsoft must support a major partner while continuing to develop its own models, infrastructure, and product identity.

The company is also diversifying its hardware. Microsoft announced an expansion of Azure infrastructure using advanced AMD accelerators alongside its internal silicon and other suppliers.

A heterogeneous approach can reduce reliance on one chip vendor. It also increases the engineering challenge of delivering predictable performance across different systems.

The Pershing Square thesis appears to treat these concerns as manageable. Microsoft’s installed customer base, cash generation, and cloud platform provide several ways to monetize AI.

The skeptical case is straightforward. AI agents might reduce switching costs, weaken traditional software bundles, or move value toward independent model providers.

Microsoft therefore must prove more than cloud growth. It must show that AI strengthens the economics of its software relationships instead of merely increasing the cost of defending them.

Uber Makes a Different Artificial Intelligence Stock Bet

Uber offers exposure to autonomous transportation without requiring Pershing Square to select one winning robotaxi manufacturer.

Uber is the least obvious member of this group. It does not operate a hyperscale cloud platform, train a leading general-purpose model, or sell enterprise productivity software.

Its AI opportunity comes from marketplace coordination. Uber matches riders, drivers, couriers, merchants, and available vehicles across a large network.

Machine learning already helps the platform estimate arrival times, predict demand, route trips, detect fraud, and balance prices with available supply. Autonomous vehicles could change who provides that supply.

Several companies are building robotaxis, including Alphabet’s Waymo, Tesla, and specialist developers. They face technical, regulatory, operational, and geographic constraints.

Even a capable autonomous vehicle needs customers, dispatching, payments, support, insurance processes, and integration with local transportation demand.

Uber argues that its marketplace can provide those functions. Instead of developing every autonomous driving system itself, it can connect multiple vehicle partners with riders.

This is a platform strategy. Uber wants to remain the demand aggregator even if human drivers gradually share the network with autonomous fleets.

That approach spreads technological risk across partners. Uber does not need one internal autonomy program to outperform every competing system.

It also allows fleet operators to use Uber’s existing consumer application. A developer entering a new city can access potential riders without building an equivalent marketplace from zero.

Ackman has described Uber as a high-quality business with substantial earnings potential. Pershing Square first disclosed the position in 2025 and kept it among its largest holdings in the March 2026 filing.

However, autonomy creates a genuine threat to Uber’s economics. A robotaxi operator with sufficient demand might bypass the platform and keep more of each fare.

Waymo’s relationship with Alphabet gives it access to capital, mapping expertise, and consumer distribution through Google services. Tesla can connect autonomy with a large installed base of vehicles and owners.

Local regulation could also limit Uber’s ability to scale autonomous partnerships. Approval processes, safety rules, insurance requirements, and reporting obligations vary between cities.

The technology remains uneven across environments. A system that works in a mapped, favorable operating area may struggle with weather, construction, emergency scenes, or unusual road behavior.

Uber cannot control the pace at which partners solve those problems. Its platform value rises only if autonomous fleets achieve dependable commercial service.

The company’s next quarterly disclosure is therefore important. Uber scheduled its second-quarter 2026 results for August 5, according to its earnings notice.

Investors should look beyond partnership counts. The more useful indicators include autonomous trip volume, geographic expansion, customer retention, and the economics shared between Uber and fleet providers.

A long list of agreements does not automatically create a profitable network. Some partners may remain in limited pilots for extended periods.

Uber must also preserve its human-driven marketplace during the transition. Drivers provide flexible capacity, especially during peak demand and in areas that autonomous vehicles cannot serve.

An abrupt shift could damage driver availability before robotic supply becomes dependable. A slow shift could delay the expected economic benefits of autonomy.

This makes Uber a different artificial intelligence stock from Amazon or Microsoft. Its upside depends on AI moving through the physical world under local constraints.

The investment case is not that Uber will build the best autonomous driving system. It is that a fragmented field of developers will need a common marketplace.

That thesis becomes stronger if multiple autonomous operators use Uber for commercial distribution. It becomes weaker if one provider establishes a dominant consumer network and treats Uber as unnecessary.

The Real Opponent Is Valuation, Not Alphabet

Pershing Square’s rotation suggests that good AI companies still compete against one another for scarce investment capital.

It would be tempting to frame Ackman’s move as Microsoft defeating Alphabet. His public explanation does not support that simple conclusion.

Ackman said Pershing Square remained bullish on Alphabet after selling most of the position. The fund used Alphabet as a source of capital because Microsoft appeared more attractive at the available valuation.

That is a relative decision. A company can retain excellent products and still become a less compelling investment after its share price rises.

Alphabet remains a formidable competitor. Google Search supplies global distribution, Google Cloud serves large organizations, and DeepMind develops leading AI systems.

The company also owns Waymo, which gives it direct exposure to autonomous transportation. Alphabet therefore competes with Microsoft in cloud and models, Amazon in cloud and advertising, and Uber in mobility.

Pershing Square’s sale does not erase those strengths. It shows that portfolio decisions incorporate expected return, not technological reputation alone.

This distinction is especially important for readers arriving through Google News. Headline language such as “brilliant stocks” compresses several uncertain judgments into one adjective.

A brilliant business can become an unattractive stock at an excessive valuation. A company facing real challenges can produce strong returns if those challenges are already reflected in expectations.

Form 13F data cannot reveal Pershing Square’s precise estimates. It only reveals the reported positions at one past date.

The filing also excludes short positions, private investments, cash, and some derivatives. Reported portfolio percentages should not be mistaken for a complete representation of the firm’s risk.

Investors face another issue when copying concentrated managers. Ackman can tolerate volatility differently from a household saving for a near-term expense.

Pershing Square can also communicate with company leadership, perform extensive research, and change positions at a scale unavailable to many individuals. None of that guarantees success, but it changes the context.

AI spending adds further uncertainty. Amazon, Microsoft, Alphabet, and other large technology companies are committing substantial resources to data centers and chips.

The demand is visible, yet long-term returns on that infrastructure remain unsettled. Capacity shortages can support pricing today while rapid expansion produces excess supply later.

Model efficiency creates another variable. If software requires less computing for the same task, customers may receive more value while infrastructure revenue grows more slowly.

The opposite scenario is also possible. Lower costs can encourage far more usage, increasing total demand even as each request becomes cheaper.

Nobody can establish that outcome from one quarter of cloud growth. Investors must watch utilization, pricing, margins, and customer behavior over several reporting periods.

The three selected companies also carry risks unrelated to AI. Amazon faces retail competition and labor costs. Microsoft faces cybersecurity and software-market scrutiny. Uber faces regulation, insurance exposure, and marketplace competition.

Those ordinary business risks matter because AI does not replace the underlying economics. It changes how each company serves customers and allocates capital.

A sound analysis should therefore separate three questions. Is the company using AI effectively? Can that use generate durable cash flow? Does the current valuation provide a reasonable margin for error?

Ackman’s portfolio rotation appears to answer all three with a preference for Amazon, Microsoft, and Uber. Other investors can examine the same evidence and reach a different conclusion.

The lesson is discipline, not imitation. A famous investor’s purchase provides a research prompt, not a substitute for independent judgment.

What to Watch After This Google News Stock Story

The next stage of the thesis depends on reported business performance, autonomous deployment, and Pershing Square’s subsequent disclosures.

The first signal is cloud growth paired with operating discipline. Amazon and Microsoft must show that AI demand produces revenue without permanently weakening cash generation.

For Amazon, watch AWS sales growth, segment operating income, and adoption of its custom chips. Growth across all three would support the argument that Amazon controls more of the AI infrastructure stack.

For Microsoft, watch Azure demand, data-center capacity, and paid usage across enterprise assistants. Broad availability matters less than sustained customer engagement.

The second signal is Uber’s autonomous trip activity. Investors should focus on commercial deployments instead of demonstrations or partnership announcements.

A useful update would identify growing trip volumes across several operators and cities. That result would strengthen Uber’s claim that its network can serve as shared distribution for autonomous fleets.

Exclusive networks would send the opposite signal. If leading robotaxi companies build direct customer relationships and avoid Uber, the marketplace thesis would weaken.

The third signal is Pershing Square’s next complete portfolio disclosure. The second-quarter Form 13F is due after the period covered by this article, so the March filing remains a delayed snapshot.

A larger Microsoft or Amazon position would reinforce the rotation. A significant reversal would show why investors should avoid treating any quarterly filing as permanent conviction.

Readers should also monitor Alphabet. Strong cloud growth, search resilience, or Waymo expansion could make Pershing Square’s reduced position appear overly cautious.

That would not invalidate Ackman’s allocation process. It would demonstrate that several strong companies can succeed while producing different returns from different starting valuations.

The phrase “three AI stocks” also deserves continued scrutiny. Amazon, Microsoft, and Uber earn most of their revenue from broader businesses.

That feature is part of their appeal. Existing operations can finance AI investment and provide immediate distribution.

It also makes their AI exposure difficult to measure. Revenue growth can reflect many factors, while management teams have incentives to associate new products with a popular technology cycle.

Investors should favor observable behavior over broad claims. Cloud consumption, software retention, autonomous trips, operating margins, and capital intensity provide more useful evidence than the number of AI announcements.

The same evidence-first approach helps knowledge workers evaluating AI products. Keeping source material organized and building a searchable knowledge base makes it easier to compare claims with later results.

The Google News headline is ultimately a starting point. It identifies three businesses connected to one prominent investor, but it does not establish that every reader should own them.

Amazon offers infrastructure and operational leverage. Microsoft offers enterprise distribution. Uber offers a marketplace for autonomous transportation.

Their shared advantage is not simply artificial intelligence. Each already controls a commercial channel that AI developers and customers need.

Their shared risk is equally clear. Investors expect those channels to convert costly AI investment into durable growth, leaving limited room for disappointment.

Watch the next earnings releases, Uber’s autonomous trip expansion, and Pershing Square’s next filing. Then ask a harder question than whether Bill Ackman owns the stocks: does the new evidence still support the reasons he appears to own them?