By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panel consensus is neutral, with concerns about stretched valuations, competition in custom ASICs, and the risk of disappointing AI project ROI. While AI demand remains a long-term tailwind, investors should monitor risks and be realistic about multiples.
Risk: The shift from 'must-have' training compute to 'cost-optimized' inference hardware, which will inevitably erode Nvidia's pricing power and compress multiples.
Opportunity: Long-term infrastructure demand driven by hyperscalers' capex.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
There are several strong artificial intelligence (AI) stock picks available in the market right now. The AI infrastructure build-out is expected to last through at least 2030, so scooping up shares now with a long-term investing mindset is a smart way to approach the current market environment. These five in particular look like solid buys right now.
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
Nvidia (NASDAQ: NVDA) has been the top AI stock pick since 2023 for a good reason: Its products sit at the core of the AI build-out. Nvidia makes GPUs (graphics processing units), which are the primary computing units deployed in data centers to handle AI workloads. Though it's already the world's largest company by market cap, Nvidia continues to see incredible growth, with its revenue rising 85% in its most recent quarter.
Its growth is far from over, given projections that annual data center capital expenditures will rise to the $3 trillion to $4 trillion range by 2030. That's a major, long-term investing opportunity. With Nvidia's chips likely to be at the center of that, it's well worth buying its shares now.
Because of the AI infrastructure build-out, demand for memory chips now far exceeds supply, and the companies that make those chips are profiting from the shortage. When the supply of any commodity lags behind rising demand, basic economics dictates that the commodity price will soar, and that's exactly why Sandisk (NASDAQ: SNDK) has done so well lately. It makes NAND memory for solid-state drives (SSDs) for long-term data storage in data centers. Its revenues and profits are undergoing monstrous growth, and even though the stock has risen by a tremendous amount over the past year, it doesn't appear to be stopping.
Wall Street analysts expect 336% growth during Q4 of its fiscal 2026 (which ends this month), and 122% in fiscal 2027. With the memory chip crunch expected to persist for years, that makes Sandisk a solid investment pick right now.
Microsoft (NASDAQ: MSFT) used to be one of the more popular investment options in the AI realm. However, the market has lost some faith in it, and the stock is down around 25% from its all-time high. Yet all that Microsoft has been doing is growing its two primary AI divisions.
Microsoft's annual recurring AI revenue (from products like Copilot) crossed $37 billion last quarter, up 123% year over year. Its cloud computing division, Azure, saw 40% revenue growth, reflecting the huge demand for AI computing resources. Microsoft looks like a bargain buy right now, and investors should scoop up shares of this proven winner before it returns to setting new all-time highs.
Meta Platforms (NASDAQ: META) is probably the biggest wildcard among the four AI hyperscalers. It doesn't rent out its computing capacity to others, as CEO Mark Zuckerberg claims it's using it all. So, all of its AI spending has gone into boosting its own capabilities, which has worked out well for it on the advertising front.
Meta Platforms operates the social media platforms Instagram, Facebook, Threads, and WhatsApp, and advertising on these platforms generates nearly all of Meta's revenue. Meta has used its AI investments to improve the effectiveness of its ad platform, which has led to solid 33% revenue growth. However, investors want more.
Meta is working on "more" with some products like AI glasses and a personal superintelligence model. If either of these two is a hit, Meta's stock could be primed for a major upside. Even if they don't pan out, Meta's ad business is still a solid reason to buy and hold the stock.
Although many may focus on Amazon's (NASDAQ: AMZN) e-commerce business, as an investor, I prefer to look at its cloud computing unit, Amazon Web Services (AWS). AWS provides more than half of Amazon's operating profits, so it's one of its most important business units. In Q1, it grew revenue by 28% year over year -- its best pace in nearly four years. With demand for cloud computing capacity booming and Amazon spending $200 billion on data center capital expenditures this year alone, the growth rate for AWS will likely explode in the next few years.
Given that AWS' profit margins are substantially better than those of the e-commerce segment, this should lead to outsize growth on the bottom line, which is why I expect Amazon to be one of the best-performing stocks over the next few years.
Before you buy stock in Nvidia, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Nvidia wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $438,283! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,257,427!
Now, it’s worth noting Stock Advisor’s total average return is 938% — a market-crushing outperformance compared to 206% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
**Stock Advisor returns as of June 12, 2026. *
Keithen Drury has positions in Amazon, Meta Platforms, Microsoft, and Nvidia. The Motley Fool has positions in and recommends Amazon, Meta Platforms, Microsoft, and Nvidia. The Motley Fool has a disclosure policy.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
Four leading AI models discuss this article
"The Sandisk ticker error signals sloppy research that undercuts in the entire list of recommended AI holdings."
The article pushes five AI names but immediately loses credibility by recommending Sandisk (SNDK), a ticker that ceased trading after Western Digital's 2016 acquisition. NVDA still sits at the center of data-center GPU demand, while MSFT, AMZN, and META monetize cloud and ad platforms, yet the piece ignores stretched valuations, rising competition in custom ASICs, and the risk that 2026-2030 capex forecasts prove too optimistic if ROI on AI projects disappoints. Memory pricing cycles also remain volatile beyond the current shortage.
Even if SNDK is an error, the revenue beats at Azure and AWS plus Nvidia's 85% growth show the infrastructure build-out is real and could extend further than skeptics expect.
"Sustained data-center capex growth driven by AI demand is the key driver of Nvidia's upside; without it, the thesis unravels."
AI demand remains a long-term tailwind, particularly for Nvidia, but the article glosses several critical risks. The Sandisk pick appears outdated (SanDisk was acquired by Western Digital, leaving direct NAND exposure unclear). Nvidia's lofty multiple hinges on a durable data-center capex cycle and a relentless AI compute demand; any slowdown or competitive pressure could compress valuations. Cloud spend bears cyclicality: AWS/Azure growth could decelerate if enterprise budgets tighten or software efficiency reduces compute needs. Regulatory, export-control, and geopolitics add more volatility. Still, the thesis is constructive, but only with disciplined risk monitoring and realistic multiples.
Valuation risk is the strongest counterpoint: Nvidia's price already prices in an almost perpetual AI boom, so any moderation in data-center capex or a memory oversupply could hurt. If macro or policy shocks curb cloud spend, the upside may be far smaller than the article suggests.
"The article's inclusion of SanDisk as a standalone public ticker invalidates its credibility, signaling a need for extreme caution regarding the author's underlying data."
This article is dangerously outdated and factually compromised. Most notably, it lists SanDisk (SNDK) as a public investment, ignoring that Western Digital acquired it in 2016. Relying on this piece for actionable intelligence is a liability. While the thesis on hyperscaler CAPEX (Amazon, Microsoft) driving long-term infrastructure demand remains sound, the article ignores the 'AI ROI' gap—the reality that massive data center spending has yet to translate into proportional enterprise software revenue. Investors should focus on the transition from training to inference-based revenue, as the current valuation multiples for NVDA and AMZN assume a flawless execution of this pivot.
The massive, sustained capital expenditure from hyperscalers acts as a self-fulfilling prophecy, creating a baseline of demand that effectively de-risks the infrastructure build-out regardless of immediate software monetization.
"The article mistakes a secular AI trend for a buy signal at peak valuations, ignoring margin compression from competition and custom silicon cannibalization."
This article conflates AI infrastructure tailwinds with stock valuations—a dangerous gap. Yes, $3-4T capex by 2030 is real, but it's already priced into NVDA (forward P/E ~30x vs. 10-year avg ~20x) and MSFT (similar premium). The article ignores that hyperscalers are vertically integrating: Meta and Amazon build custom chips, Microsoft partners with AMD and Mobileye. Sandisk's 336% Q4 growth projection is unverified and wildly optimistic—memory cycles are notoriously volatile. The article also omits that AI capex growth is decelerating YoY (Nvidia's 85% revenue growth is already off the peak 200%+). Buying 'before it returns to all-time highs' is backward reasoning, not analysis.
The $3-4T capex thesis is real, demand for compute is structurally higher, and these five companies do control critical infrastructure—if execution continues, valuations may compress less than skeptics expect.
"Vertical integration will divert capex from Nvidia faster than the consensus $3-4T thesis assumes."
Claude correctly flags vertical integration but underplays how quickly custom ASICs from Meta and Amazon could shrink Nvidia's addressable market share inside the same $3-4T capex envelope. If hyperscalers internalize more inference workloads, the 85% growth rate he cites becomes a peak rather than a floor, pressuring multiples before any ROI disappointment materializes. Memory cycle volatility only amplifies this reallocation risk.
"Nvidia's ecosystem moat and broad demand across training and inference will limit hyperscalers' bespoke-ASIC erosion, making macro capex and ROI timing the bigger near-term risks to its valuation."
Grok flags hyperscaler custom ASICs as a threat to Nvidia's dominance, but the ecosystem moat (CUDA, libraries, tooling) and breadth of GPU demand across training and inference should limit share erosion. Even with some bespoke accelerators, Nvidia benefits from multi-vendor demand and upgradability. The bigger risks remain macro capex cycles, memory-price volatility, and AI ROI delays that could compress multiples before any ROI materializes.
"The transition from training to inference workloads will commoditize hardware and compress Nvidia's margins regardless of the CUDA moat."
ChatGPT, you underestimate the 'moat' argument. CUDA is a software lock-in, but hardware commoditization is inevitable as inference becomes the primary workload. Once models are trained, they run efficiently on cheaper, specialized silicon. By ignoring this, you miss that Nvidia’s margin profile is at its peak. The real risk isn't just macro; it's the shift from 'must-have' training compute to 'cost-optimized' inference hardware, which will inevitably erode Nvidia's pricing power and compress those lofty multiples.
"Nvidia's margin compression is real but delayed—training capex cycles buy 3-5 years before inference commoditization meaningfully erodes pricing power."
Gemini's inference-commoditization thesis is sound, but it assumes a false timeline. Training workloads—where Nvidia's margin and pricing power are highest—will remain the bottleneck through 2027-2028. Inference optimization happens *after* training scales. Hyperscalers won't cannibalize Nvidia's high-margin training revenue to save on inference until they've amortized trillion-dollar training clusters. That's the real moat: Nvidia captures the peak-margin phase before commoditization kicks in.
The panel consensus is neutral, with concerns about stretched valuations, competition in custom ASICs, and the risk of disappointing AI project ROI. While AI demand remains a long-term tailwind, investors should monitor risks and be realistic about multiples.
Long-term infrastructure demand driven by hyperscalers' capex.
The shift from 'must-have' training compute to 'cost-optimized' inference hardware, which will inevitably erode Nvidia's pricing power and compress multiples.