Three AI Stocks Positioned for Growth: Alphabet, Nebius, Arista

Alphabet runs AI experiments from multiple angles—as a product feature in search, as a standalone consumer app in Gemini, and as an enterprise cloud offering. All three are funded by deeply profitable cash flow from an advertising business built over decades that does not depend on AI bets to keep growing.

Search revenue continued to grow after Alphabet rolled out AI overviews, disproving the bear case that chatbots would cannibalize search. In Q2 2026, overall revenue grew 24% year-over-year while search and advertising revenue grew 17%. Google Cloud revenue grew 82% year-over-year, reaching a $514 billion backlog of contracted future revenue.

Gemini reached 950 million monthly active users, with daily active users tripling over the past year. The Gemini Enterprise offering is now used by nearly 90% of the Fortune 100. The stock fell after recent earnings as investors focused on AI infrastructure spending relative to expectations.

Despite strong growth across cloud and AI products, Alphabet's ability to fund these bets without disrupting the core advertising engine remains a structural advantage. The company has been able to layer AI onto its existing business rather than cannibalize it.

Alphabet has overcome regulatory challenges time and time again throughout its history, but antitrust risk remains a factor investors must weigh. The company faces continued scrutiny from governments globally, particularly regarding its dominance in search and advertising.

For long-term investors, regulatory risk sits in the background as a variable to monitor but not necessarily a reason to avoid the stock. Alphabet's scale and legal resources have allowed it to navigate past pressures, though the outcome of current cases could reshape parts of its business.

Among AI stocks, Alphabet's valuation has trended toward being more favorable than much of the competition. This combination—genuine AI growth engines across three revenue streams, plus a more reasonable valuation—makes it an appealing choice for investors seeking exposure to artificial intelligence.

For analysts tracking the stock, Alphabet remains a compelling buy in the AI space, particularly for those who have held it through its long history as a core technology holding.

Google is a very profitable business at its core, even though the long-term question of search disruption remains open. Search now represents closer to 50% of Alphabet's total revenue, down from higher percentages in prior years. Google Cloud is growing much faster than search and becoming extremely profitable, which means that even if search encounters headwinds, Alphabet has diversified revenue sources working in its favor.

This shift in the revenue mix has already begun to hedge against any future search market disruption. The speed and profitability of the cloud business give Alphabet breathing room that younger, more narrowly-focused AI companies do not have.

When the team had to leave Russia, they relocated to Amsterdam and started Nebius from scratch, positioning it as a European success story. The company could become the first large European publicly traded company to reach a $100 billion market capitalization in more than a decade.

The core business is renting out AI compute power. Nebius is expanding beyond bare metal sales through acquisitions, building a sticky software platform to increase margins per GPU and lock in customers who might otherwise shop for cheaper capacity elsewhere.

Q2 revenue reached $582 million, up 454% year-over-year and up 46% from Q1. The AI cloud business alone grew 514% year-over-year. When calculated as annual recurring revenue—taking last month's revenue and multiplying by 12—the run rate hit $3 billion, up from $1.9 billion just three months prior.

On an adjusted basis, the company is profitable. Group adjusted EBITDA was $236 million with roughly 41% margin, demonstrating that Nebius is not burning capital to achieve its growth rates despite the high premium the stock commands at $60 billion market cap.

Nebius projects annualized recurring revenue to grow to between $7 billion and $9 billion. To reach that target requires substantial capital investment: capex is expected to be $20 billion to $25 billion over the projection period.

One analyst estimates 2027 capex will climb much higher than the $20-25 billion guidance, predicting that Nebius could reach $30 billion to $40 billion in ARR by the end of 2027. This aggressive expansion is only feasible because demand for AI compute capacity vastly outpaces available supply globally.

Nebius has around $40 billion in contracted backlog, with Microsoft and Meta accounting for major portions of it. This customer concentration provides revenue certainty and helps fund rapid buildout.

These large deals are not high-margin arrangements for Nebius, which is strategic. The company uses low-margin contracts to solidify its core business, secure long-term customer relationships, and fund even faster growth than competitors can achieve. This willingness to sacrifice margin for market share and capital efficiency is a deliberate trade-off.

Demand for AI compute vastly exceeds supply across the board. Nebius could sell its entire 2027 capacity today on basic terms, but is deliberately holding back portions because it can command higher prices. A recent capacity auction for Nvidia Blackwell chips cleared 15% above the highest price the company has ever charged for that hardware.

Revenue per megawatt has climbed across the board since the auction. Previously, Nebius was generating $10 million to $15 million per megawatt; now the range is $20 million to $25 million, and some capacity is being sold for $40 million per megawatt. Because the AI infrastructure business is capital-intensive and requires continuous reinvestment to grow, accepting premium prices instead of clearing all inventory at lower rates allows Nebius to accelerate its expansion faster than peers.

Token factory is Nebius's managed inference platform. Rather than simply renting raw chips, companies pay to run AI models in production on Token factory, which handles the operational complexity. This shifts Nebius from a commodity chip-rental business to a higher-margin, stickier software business.

Nebius has made multiple acquisitions to expand Token factory's capabilities. Tavili, acquired earlier in 2026, gives AI agents live web search capability for applications that need fresh information. Agon AI was acquired to make models run faster on the same hardware, increasing answers generated per GPU per hour. Clarify optimizes job scheduling across thousands of GPUs. Infrise, acquired this week for around $150 million, reduces GPU cold start time from 20 minutes to 1 minute, dramatically improving hardware utilization rates. Each acquisition targets a different bottleneck in AI production, and higher GPU utilization allows Nebius to charge better prices and generate more revenue than competitors managing the same hardware.

Infrise was founded by engineers who previously built Infinigrade before Intel acquired and shut down that company. By reducing GPU cold start from 20 minutes to 1 minute, the Infrise technology saves customers significant time and improves the customer experience.

For Nebius, the benefit is equally concrete: reducing cold start time increases GPU utilization rates, meaning the same hardware generates more revenue and better amortizes the cost of the data center and power draw.

Permit delays, lawsuits, and regulatory obstacles are common challenges across the data center industry as companies race to build infrastructure for AI workloads. Nebius faces the same permitting headwinds as competitors.

The company mitigates this risk through geographic diversification. Nebius has a large presence in the United States, is growing in the Middle East, and is expanding in Europe. This spread reduces the impact of delays in any single jurisdiction. In a recent investor call, Nebius's chief revenue officer stated the company sees a path to reach $100 billion in annual revenue in the not-too-distant future, suggesting management confidence in clearing the execution hurdles.

The majority of Nebius engineers previously built Yandex, a large and successful company, which provides a significant competitive advantage in team coordination and capital deployment speed—both critical in a capital-intensive business where execution risk is high.

Arista Networks manufactures Ethernet switches used in AI data center infrastructure. The company's switches are built on top of its own proprietary software platform that runs across its entire product line, giving customers a single consistent operating environment regardless of which switch model they deploy.

This unified software approach creates customer lock-in. Once enterprises build their AI infrastructure on Arista's platform, switching to a competitor's hardware means rewriting network configurations and retraining teams—a significant friction point that favors Arista's continued expansion with existing customers.

AI-specific revenue target was raised to $3.6 billion for the full year. Arista now has over 100 cumulative AI fabric customers using its Ethernet switch platform, up from a handful of early adopters in 2024.

The total addressable market for Arista's Ethernet switch platform and AI infrastructure offerings is estimated at upwards of $105 billion, indicating significant room for the company to grow beyond its current revenue run rate.

Arista has built a hardware and software reputation over years of reliable products. The company's net promoter score of 89 ranks among the highest of any enterprise technology company, translating into genuine customer loyalty.

This high NPS makes Arista one of the trusted default choices for enterprises addressing bottlenecks in AI infrastructure buildout. When large enterprises evaluate which networking vendor to standardize on, Arista's reputation and customer satisfaction advantage become difficult for competitors to overcome.

Arista's last 12 months of revenue just exceeded $10 billion. If the market's valuation is justified, the company must double that revenue within two to two-and-a-half years, requiring sustained high growth even as the company reaches larger scale.