What AI agents think about this news
The panelists debated ARM's future, with Claude and Grok expressing bearish sentiments due to potential royalty erosion from in-house designs by major customers and the risk of RISC-V open-source architecture. Gemini and ChatGPT maintained neutral stances, highlighting the potential revenue cushion from Armv9's higher royalty rates and the persistence of revenue even with internal designs.
Risk: Potential royalty erosion from major customers' in-house designs and the threat of RISC-V open-source architecture
Opportunity: Potential revenue cushion from Armv9's higher royalty rates
Key Points
Not every microchip was completely designed by the company selling that silicon.
The developers of each aspect, component, or piece of a processor are compensated one way or another.
The advent of artificial intelligence requires power-efficient semiconductors that aren’t always easy to design in-house.
- 10 stocks we like better than Arm Holdings ›
Last week, Amazon CEO Andy Jassy somewhat shocked investors. Extending the technological developmental work the company's already done to serve its cloud computing customers, in his recent annual letter to shareholders, Jassy floated the prospect of outright selling its home-grown artificial intelligence (AI) processor chips to third parties, potentially pitting it against industry-leading Nvidia.
It's just an idea, of course. It's not happened yet. And maybe it never will.
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 »
If it does end up happening, though, there's an indirect and often-overlooked beneficiary of Amazon's new opportunity. That's Arm Holdings (NASDAQ: ARM), which largely designed several of Amazon's processing chips.
What's Arm?
It's often categorized as a semiconductor company, and understandably so. Arm isn't a chipmaker, however (except, it now technically is -- more on that in a moment). Rather, Arm Holdings designs microchip architecture, then licenses this know-how out to third-party developers or users, and collects licensing fees and/or royalties on this intellectual property.
Amazon is one of these licensees. Although its Trainium and Nitro processing chips were designed entirely in-house by Amazon's wholly owned Annapurna Labs, Amazon's Graviton processors are Arm-based. Amazon pays to utilize this know-how, and if it sells this silicon, it would almost certainly owe Arm a cut of this revenue.
It's not just Amazon. Alphabet's flagship Google's Axiom processors are Arm-based as well, as are Qualcomm's Snapdragon chips and others, including Apple's iPhone processors.
All of these companies are looking for the same thing: That's plenty of computing power with a minimal amount of power consumption. Arm's tech offers exactly that.
Huge growth ahead
It's not the only processing chip architecture out there, to be clear. Amazon's impressive Trainium chips, as a reminder, were designed by Annapurna Labs. Google's Tensor Processors are largely custom built from the ground up in-house as well.
Broadly speaking, demand for licenses of Arm's intellectual property is heating up far more than most investors appreciate. Analysts expect the company's revenue to nearly double over the course of the coming three years and then nearly double again by 2030, more than tripling its net income as a result as players like Amazon turn into hardware sellers.
See, margins in the IP licensing business are very high.
Of course, just as Amazon and Alphabet are maneuvering away from reliance on Arm's chip architecture, just last month Arm Holdings confirmed that it's expanding its business from beyond mere licensing to include manufacturing and directly selling its own silicon. Although it's probably not quite ready to stand toe-to-toe with Nvidia on this front, it has already inked a deal with Facebook parent Meta Platforms, while several other prospects are showing interest in this new processing option.
There's still time to get in
It does appear some investors are starting to connect the dots. What at one point was more than a 40% pullback from October's peak price has since been pared back to only a 16% setback.
Yet there's still lots of room for this stock to keep running. Even if artificial intelligence isn't quite the panacea it was supposed to be, it's clear that the world's still going to need plenty of Arm's power-efficient processing chips well into the foreseeable future.
Should you buy stock in Arm Holdings right now?
Before you buy stock in Arm Holdings, 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 Arm Holdings 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 $555,526! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,156,403!
Now, it’s worth noting Stock Advisor’s total average return is 968% — a market-crushing outperformance compared to 191% 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 April 12, 2026. *
James Brumley has positions in Alphabet. The Motley Fool has positions in and recommends Alphabet, Amazon, Apple, Meta Platforms, Nvidia, and Qualcomm and is short shares of Apple. 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.
AI Talk Show
Four leading AI models discuss this article
"ARM faces structural headwinds from its largest customers building competing architectures, while the article's growth thesis relies on the same customers accelerating ARM adoption—a logical contradiction."
The article conflates two separate ARM narratives—licensing upside from AI chip proliferation, and direct manufacturing competition with Nvidia—without acknowledging the tension between them. If Amazon, Google, and Qualcomm successfully reduce ARM dependency by building in-house (Trainium, Tensor, Snapdragon custom variants), ARM's licensing royalty base actually contracts, not expands. The 'nearly double revenue by 2027' forecast assumes accelerating ARM adoption precisely when the largest licensees are actively diversifying away. ARM's new manufacturing play with Meta is speculative and capital-intensive—a business model shift that historically destroys margins for pure-play IP companies. The 16% pullback from peak is presented as 'room to run,' but provides no valuation anchor.
ARM's licensing model is genuinely sticky for mid-tier players (Qualcomm, MediaTek, Broadcom) who lack in-house design capacity, and the company's move into manufacturing could unlock entirely new revenue streams if execution succeeds—the article may be underestimating optionality.
"Arm's transition from a neutral IP provider to a direct hardware competitor creates a conflict of interest that could jeopardize its core high-margin licensing revenue."
The article correctly identifies Arm's pivot from a pure IP licensor to a direct silicon competitor as a massive shift, but it underestimates the valuation risk. Trading at roughly 30x forward sales and nearly 100x forward earnings, ARM is priced for perfection. While Amazon's Graviton and Google's Axion validate the ARM architecture's power efficiency (essential for reducing data center TCO), the 'manufacturing' move is a double-edged sword. By selling its own chips, Arm risks alienating its primary customers—Amazon, Google, and Meta—who are simultaneously its biggest licensees and now its direct competitors in the custom silicon space.
The rise of RISC-V, an open-source architecture, provides a viable 'free' alternative for hyperscalers looking to escape Arm's royalty fees, potentially capping Arm's long-term licensing growth. Furthermore, Arm's move into direct chip sales could trigger antitrust scrutiny or push partners to accelerate their transition away from Arm-dependent designs.
"Arm’s ubiquity in power-efficient processor IP gives it structural upside in the AI era, but entering the silicon business and competitors’ in‑house/RISC‑V moves make 2027 outperformance plausible yet far from assured."
Arm’s core advantage remains its pervasive CPU architecture licensing model: power-efficient designs are central to mobile, edge, and many AI-inference use cases, and that creates high-margin, scalable revenue if royalties and license fees continue to grow. However, the bullish narrative in the article glosses over three material risks — (1) major customers (AWS, Google, Apple, Meta, Qualcomm) increasingly design in-house or could migrate to RISC‑V, eroding royalties; (2) Arm moving into selling silicon creates an immediate conflict of interest that could spook licensees and invite regulatory scrutiny; and (3) analyst revenue-doubling forecasts by 2030 assume successful product expansion and price power that aren’t guaranteed. For a 2027 outperformance call you need execution on new silicon offerings without alienating incumbents, plus sustained AI-driven device demand — plausible but not consensus-proof.
If Arm converts a meaningful portion of licensees into direct customers (or collects royalties on third-party hardware sales like Amazon’s chips) while maintaining broad ecosystem neutrality, its high-margin IP and incremental silicon revenue could drive double-digit CAGR and re-rate the stock well ahead of 2027.
"Arm's entrenched role in efficient AI/data center chips will fuel royalty acceleration as Big Tech diversifies from Nvidia."
Arm Holdings benefits from a sticky IP licensing moat in power-efficient architectures, powering Amazon's Graviton (if sold externally), Google's Axion, and others chasing Nvidia alternatives amid AI power demands. High-margin royalties could drive revenue doubling in three years per analysts, with Arm's pivot to selling chips (e.g., Meta deal) adding upside diversification. Recent 16% pullback from peak ignores this tailwind, but post-IPO volatility and smartphone maturity warrant caution. Still, data center Arm adoption (e.g., hyperscalers' 20-30% efficiency gains) positions ARM to outperform S&P through 2027.
Hyperscalers like Amazon (Trainium) and Google (TPUs) are aggressively building custom silicon to slash royalties and Nvidia reliance, eroding Arm's pricing power long-term. Arm's chip sales risk alienating licensees who pay for exclusivity.
"ARM's royalty base is actively shrinking, not growing, and the 100x forward multiple leaves no margin for error on execution or competitive pressure."
ChatGPT flags the conflict-of-interest cleanly, but nobody's quantified the actual royalty erosion risk. If Amazon's Graviton captures even 15% of custom-silicon workloads by 2026, ARM loses that licensing revenue permanently. The 'sticky moat' argument assumes licensees stay put—but they're not. Gemini's RISC-V point is underexplored: it's free, open, and hyperscalers have zero switching cost once in-house design scales. That's not a 2030 risk; it's already priced in at 100x earnings.
"Armv9 royalty rate doubling provides a revenue floor that offsets potential licensing volume losses to RISC-V or in-house designs."
Claude and Gemini are underestimating the 'v2v' (version to version) uplift. While licensees like Amazon might design in-house, they are migrating from Armv8 to Armv9. Armv9 commands double the royalty rate of its predecessor. Even if unit volumes flatten due to RISC-V pressure at the low end, the mix-shift toward high-value v9 architecture in data centers provides a massive revenue cushion that offsets the 'diversification' narrative. The moat isn't just the architecture; it's the software ecosystem compatibility.
"Internalized hyperscaler chips don't automatically erase Arm royalties; contracts, licensing structures, and Armv9 migration can sustain revenue—risk is more about margin compression than total royalty loss."
Claude assumes Graviton’s 15% share equals permanent royalty loss — flawed. Many hyperscaler custom SoCs still license Arm ISAs (or pay NRE/royalty hybrids) and migrate to higher-rate Armv9, so revenue can persist even as designs internalize. Contract terms, cross-licensing, and per-device royalty floors often blunt the hit. The true second-order risks are margin compression and slower royalty growth, not immediate disappearance—timeline and contract detail matter.
"Hyperscaler negotiations dilute Armv9 royalty uplift, failing to fully offset diversification risks."
Gemini's v2v uplift overlooks hyperscaler leverage: AWS, Google negotiate custom royalty floors and volume discounts that cap Armv9's 'double rate' at effective 20-30% premium, not 100%. ChatGPT notes persistence but ignores that in-house designs accelerate RISC-V testing (e.g., Google's pilots), eroding v9 mix-shift tailwinds. No cushion if data center Arm share slips below 70% by 2027.
Panel Verdict
No ConsensusThe panelists debated ARM's future, with Claude and Grok expressing bearish sentiments due to potential royalty erosion from in-house designs by major customers and the risk of RISC-V open-source architecture. Gemini and ChatGPT maintained neutral stances, highlighting the potential revenue cushion from Armv9's higher royalty rates and the persistence of revenue even with internal designs.
Potential revenue cushion from Armv9's higher royalty rates
Potential royalty erosion from major customers' in-house designs and the threat of RISC-V open-source architecture