By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panel is mixed on AI infrastructure stocks Alphabet, Meta, and Broadcom, with concerns about the sustainability of capex cycles, competition in chip technology, and energy constraints for data centers.
Risk: The energy bottleneck for data centers and the potential weakening of chip technology moats.
Opportunity: The potential for durable earnings growth driven by AI infrastructure and ad-tech monetization.
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 →
As the most complete AI company, Alphabet looks well positioned to be a long-term winner.
Meta Platforms has been one of the best companies at using AI to drive revenue growth.
While its stock slid after its recent quarterly report, Broadcom still has huge potential to grow its custom chip design business.
Billionaire hedge fund manager Daniel Loeb of Third Point Management was busy making adjustments to his portfolio in Q1, trimming some positions while adding new ones. Among the stocks he bought during the period were three of my favorite artificial intelligence (AI) stocks, and they're still attractive buys right now.
It has been a couple of years since Loeb had a stake in Alphabet (NASDAQ: GOOGL) (NASDAQ: GOOG), but he opened a new one in Q1, and it's easy to see what he may like about the stock. While it's best known for Google Search, Alphabet is the most complete AI company, having developed both top-tier AI chips and models.
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 »
The company's biggest edge is arguably its Tensor Processing Units (TPUs), custom chips it began developing more than a decade ago, and which it optimized its entire ecosystem around. TPUs give it a big cost advantage when training its AI models and running inference workloads. They are also becoming a big new revenue source for Alphabet, as it is leasing them to Google Cloud clients and also sells them directly to select customers.
By having its own AI models, it also saves costs within its cloud computing unit by not always having to pay for third-party LLMs, although it also offers models other than its own. It also uses its Gemini models to power the AI features across Google Search and other products, which has been driving growth. Additionally, Alphabet has a big distribution edge (through its ownership of Chrome, Android, and its revenue-sharing deal with Apple) and one of the world's largest digital ad networks.
Add it all up, and this is a stock poised to be a long-term winner.
During the first quarter, Loeb also reopened a position in Meta Platforms (NASDAQ: META). (He'd sold out of his last position in Q1 2025.) Meta has been one of the best companies at harnessing the power of AI and incorporating it into its core business to drive growth. It's using AI to continually improve its recommendation algorithm, which is keeping users on its sites much longer. At the same time, it's using AI to help advertisers better target users, automate bidding, and make better creative campaigns.
The results have been impressive. The company grew its revenue by 33% last quarter as ad impressions jumped 19%. Meanwhile, because its ads are more effective, it was able to raise ad prices by 12%. While investors have fretted over Meta's heavy spending on AI infrastructure, the company has been seeing great returns on its investments. It also has a nice opportunity as it is now just starting to serve ads on the popular messaging platform WhatsApp and is still building out its newest platform, Threads.
Trading at a forward P/E ratio of just 17.5, the stock is too cheap for its growth, positioning, and prospects.
Another AI stock Loeb picked up in Q1 was Broadcom (NASDAQ: AVGO). While the stock is up from the prices it was trading at during that period, investors do have a nice opportunity to get into it after its sizable post-earnings dip.
That slide in its shares came despite the fact that Broadcom is seeing strong growth driven by its custom chip business. The company helped Alphabet develop its TPUs and is benefiting from its huge data center infrastructure spending. Broadcom expects AI semiconductor revenue of $56 billion this fiscal year, and management forecasts that figure will rise to well over $100 billion next year. In addition to Alphabet, it also has large commitments from Anthropic, OpenAI, and Meta. Its custom chip business also feeds into its data center networking business, which is seeing strong growth.
As hyperscalers continue to develop their own custom chips and diversify their compute power away from Nvidia's graphics processing units (GPUs), especially for inference workloads, Broadcom is poised for tremendous growth in the coming years. That makes the recent dip in its shares a great opportunity to buy the stock.
Before you buy stock in Broadcom, 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 Broadcom 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 $443,191! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,258,838!
Now, it’s worth noting Stock Advisor’s total average return is 941% — a market-crushing outperformance compared to 211% 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 6, 2026. *
Geoffrey Seiler has positions in Alphabet, Broadcom, and Meta Platforms. The Motley Fool has positions in and recommends Alphabet, Apple, Broadcom, and Meta Platforms. 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
"Near-term upside hinges on AI-driven revenue growth and margin expansion being durable enough to justify current multiples, which may fail if AI capex slows or ad headwinds intensify."
Loeb's Q1 moves spotlight Alphabet, Meta, and Broadcom as AI beneficiaries. The article paints them as long-term winners thanks to in-house AI ecosystems (Alphabet's TPUs and Gemini, Chrome/Android reach), Meta's AI-driven ad targeting, and Broadcom's role in hyperscaler chips. Yet the upside may already be priced in, and AI demand could be episodic rather than structural if Nvidia remains dominant for GPUs and cloud capex slows. Meta's ad growth hinges on privacy and regulatory stability; Alphabet faces ad-click headwinds and competition; Broadcom is exposed to a concentrated data-center cycle. In short, a constructive thesis needs a durable earnings path, not just hype.
Against my stance: if the AI infrastructure cycle accelerates and hyperscalers reduce Nvidia dependency, Alphabet, Meta, and Broadcom could re-rate faster than expected; Loeb's bets might pay off sooner than anticipated.
"The market is currently underestimating the risk that massive AI infrastructure spending will lead to margin compression before it produces meaningful enterprise revenue."
Loeb’s entry into GOOGL, META, and AVGO signals a pivot toward 'AI infrastructure winners' rather than pure-play hype. Alphabet’s vertical integration with TPUs provides a necessary margin buffer against rising compute costs, while Meta’s ad-tech efficiency is currently the most tangible ROI in the sector. Broadcom remains the critical 'picks and shovels' play for custom ASICs. However, the market is pricing these as if AI capex will scale linearly forever. We are approaching a point where hyperscalers must demonstrate that inference revenue actually offsets the massive depreciation costs of these data centers. If Q3/Q4 margins contract due to energy and hardware amortization, these valuations will face a sharp correction.
If the 'AI productivity' narrative fails to translate into significant enterprise software revenue, the massive capital expenditure cycles of these firms will be viewed as a sunk-cost margin disaster rather than a competitive moat.
"Loeb's buys validate AI themes but the article ignores how quickly custom-chip forecasts and ad-pricing power can be repriced lower if capex returns disappoint."
Loeb's Q1 reopenings in Alphabet, Meta, and Broadcom signal continued hedge-fund conviction in AI infrastructure and ad-tech monetization, yet the article underplays timing risk. Alphabet's TPU leasing and Meta's 33% revenue jump both hinge on sustained hyperscaler capex that could slow if ROI metrics disappoint in 2026. Broadcom's post-earnings slide after guiding $100B+ AI revenue highlights how quickly sentiment can shift when custom-chip ramp assumptions are tested. Forward multiples remain elevated relative to historical norms once Nvidia share-loss fears reemerge.
The strongest case against this caution is that Alphabet's distribution advantages and Meta's ad-efficiency gains have already produced durable 20%+ growth that easily supports 17-20x forward earnings even if AI spending pauses.
"While all three companies have genuine AI tailwinds, the article mistakes Loeb's Q1 purchases for a durable investment thesis and ignores that valuations have already repriced much of the AI upside, leaving limited margin of safety."
This article conflates three distinct narratives into one 'Loeb is buying, so should you' thesis—a common retail trap. Alphabet's TPU moat is real but faces intensifying competition (AMD, custom chips from Meta/OpenAI themselves). Meta's 33% revenue growth is impressive, but forward P/E of 17.5x assumes that growth sustains; any deceleration in ad pricing or AI ROI hits hard. Broadcom's $100B+ AI revenue forecast is aspirational, not guaranteed—hyperscalers are notoriously fickle customers who switch suppliers aggressively. The article also omits that Loeb's Q1 buys occurred before recent macro tightening and rising rates, which could pressure multiples regardless of fundamentals.
Billionaire portfolio moves are often tactical, not strategic—Loeb may have been front-running analyst upgrades or positioning for short-term volatility, not endorsing a multi-year hold. The article provides zero context on Loeb's exit strategy or position sizing, making his Q1 purchases nearly meaningless as a buy signal.
"The real risk is AI infra ROI durability: if Nvidia's dominance weakens or rival chip designs erode the TPU/ASIC moat, margins and valuations for Alphabet, Meta, and Broadcom could compress even if the near-term buys look convincing."
Claude, I’ll zero in on a risk you glossed: the cycle duration. Even if Loeb’s buys reflect conviction in AI infra, a durable ROI for hyperscalers is not guaranteed. If Nvidia’s dominance weakens due to new accelerators, or AMD/Meta/OpenAI chip designs erode TPU/ASIC moat, margins for Alphabet, Meta, and Broadcom could compress as depreciation and energy costs rise. The thesis hinges on a structural, long-lasting capex cycle, not just a hype-driven lift.
"The physical energy requirements for hyperscale data centers represent a hard ceiling on AI-driven margin growth that overrides software-based efficiency gains."
Claude, you’re right that retail investors often misinterpret hedge fund filings, but you’re ignoring the most critical risk: the 'energy bottleneck.' Even if hyperscalers solve the hardware supply chain, the grid capacity required to power these data centers is the real ceiling for Alphabet and Meta. If they cannot secure consistent, low-cost power, their AI-driven margin expansion thesis collapses regardless of chip efficiency or ad-tech ROI. This is a physical constraint, not a software one.
"Direct power sourcing deals can blunt the energy bottleneck for Alphabet and Meta more effectively than for Broadcom."
Gemini flags energy as the binding constraint, but this ignores Alphabet and Meta's direct nuclear/renewable PPAs already underway. Those contracts could cap power costs below grid rates, preserving the margin expansion thesis even if total capex stays elevated. Broadcom faces no such offset, so its ASIC ramp remains the more vulnerable link if deals close slower than expected.
"PPAs mitigate energy *cost* risk but not energy *capacity* risk; inference ROI remains the binding constraint for Broadcom's $100B forecast."
Grok's nuclear PPA argument is sound but incomplete. Alphabet and Meta's power deals lock in *current* capacity, not future growth. If AI inference demand accelerates beyond contracted MW, they face grid constraints anyway—PPAs don't scale infinitely. Gemini's energy ceiling stands. Broadcom's vulnerability isn't just deal timing; it's whether hyperscalers actually deploy ASIC volume at the rate Broadcom projects, which depends on proving inference ROI first.
The panel is mixed on AI infrastructure stocks Alphabet, Meta, and Broadcom, with concerns about the sustainability of capex cycles, competition in chip technology, and energy constraints for data centers.
The potential for durable earnings growth driven by AI infrastructure and ad-tech monetization.
The energy bottleneck for data centers and the potential weakening of chip technology moats.