By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panel agrees that the shift towards custom silicon (ASICs) poses a threat to Nvidia's dominance, but there's no consensus on whether Broadcom's ASICs represent a significant risk or opportunity. Nvidia's software moat and pricing power are seen as enduring strengths, while Broadcom's diversified portfolio and networking dominance are viewed as potential advantages.
Risk: Margin compression due to customer-led vertical integration and potential supply chain or yield bottlenecks in Nvidia's next-gen cycles.
Opportunity: Broadcom's growth in AI revenue and its dominant position in optical networking.
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 →
Nvidia(NASDAQ: NVDA) has long been considered the ultimate artificial intelligence (AI) stock to buy. Why? Because the company is the maker of the key element needed for the most crucial AI tasks: the chip for the training and inference of large language models. Others also make AI chips, but Nvidia's graphics processing units (GPUs) offer the highest performance, making them the most sought-after on the market.
All of this has translated into surging revenue growth quarter after quarter. In the latest full year, Nvidia's revenue jumped 65% to $215 billion, a record level. And stock performance has followed, with the shares climbing 1,300% over the past five years.
Will AI create the world's first trillionaire? Our team just released a report on a little-known company, called an "Indispensable Monopoly," providing the critical technology Nvidia and Intel both need.
In spite of this stellar performance, you may be wondering if Nvidia is still the best stock to buy. After all, other tech companies have been innovating in the field and growing their products and services. One in particular, a chip and networking innovator, stands out. So right now is a great time to ask the following question: Is Nvidia still the best AI stock to buy in 2026 -- or is this challenger?
A networking giant
The challenger I'm talking about is Broadcom(NASDAQ: AVGO). The company is a networking giant, with about 99% of internet traffic touching at least one Broadcom chip. And Broadcom's range of products is present in a variety of settings, from your smartphone to major data centers.
In recent years, the company, like many other tech giants, has turned its attention to AI, and this has led to explosive growth for its switches and routers -- and Broadcom also has made a splash with compute designed to accelerate certain AI tasks.
How is the company faring when it comes to competing with market giant Nvidia? The great news is that Broadcom doesn't compete directly with the chip behemoth. What I mean by this is their chips are different, so a customer probably isn't comparing Nvidia directly with Broadcom. While Nvidia's chips are general-purpose, Broadcom makes custom chips meant for a specific task. Because of this, both of these companies will likely continue to grow, undisturbed by each other.
Like Nvidia, Broadcom has seen fantastic growth in recent quarters thanks to demand for its AI products. In the latest period, AI revenue climbed more than 100% to $8.4 billion. And the company announced a forecast for incredible growth in AI chips: It predicts AI revenue from chips may reach more than $100 billion in 2027.
A deep relationship with hyperscalers
And the company said its relationship with its six hyperscaler customers "is deep, strategic, and multiyear." So we could see significant growth from Broadcom this year and in the years to come, particularly considering tech giants' major investment in infrastructure. Some of the biggest tech companies have pledged to spend nearly $700 billion this year alone.
Meanwhile, Nvidia, too, continues to deliver solid growth -- and the company's annual chip update is just ahead, with Vera Rubin availability on track for later this year. So far, the company's new releases have met with explosive demand, often surpassing supply. All of this means we should expect more growth from Nvidia in the coming quarters, too.
Now, let's return to our question: Is Nvidia still theAI stock to buy, or is it Broadcom? As mentioned, both of these players are on track to benefit from the current and next stages of AI growth -- and given that their chips are different, they aren't likely to upset each other's market position.
But one thing in particular stands out right now, and that's valuation. While Broadcom still is a reasonable buy, Nvidia stock is dirt cheap in relation to forward earnings estimates.
Nvidia's valuation has dropped significantly, even as earnings continue to march higher and demand remains strong. I see this as a fantastic buying opportunity -- and others may too, which could result in a pop in the stock in the months to come.
So, while Broadcom makes a compelling addition to any AI portfolio, thanks to valuation at the moment and growth prospects, Nvidia remains the best AI stock to buy in 2026.
Should you buy stock in Nvidia right now?
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 $471,827!* Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,319,291!*
Now, it’s worth noting Stock Advisor’s total average return is 986% — a market-crushing outperformance compared to 207% 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.
Adria Cimino has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Broadcom and Nvidia. The Motley Fool has a disclosure policy.
Four leading AI models discuss this article
"The transition from general-purpose GPU training to custom ASIC inference will compress Nvidia's long-term gross margins, regardless of their current valuation metrics."
The article frames this as a choice between NVDA and AVGO, but it ignores the fundamental shift from 'general-purpose' GPU dominance to 'custom silicon' (ASIC) proliferation. While Nvidia currently enjoys a moat via CUDA software, Broadcom’s deep integration with hyperscalers like Google and Meta for custom AI accelerators represents a structural threat to Nvidia's margin profile. By 2026, as hyperscalers optimize for cost-per-inference rather than raw training power, we may see a pivot toward specialized silicon. Nvidia remains a 'must-own' for training, but the article’s claim that NVDA is 'dirt cheap' ignores the massive execution risk if their next-gen Blackwell or Rubin cycles encounter even minor supply-chain or yield bottlenecks.
The bull case for Nvidia rests on the assumption that software lock-in (CUDA) is insurmountable, potentially rendering the move toward custom ASICs a secondary concern for years to come.
"Broadcom's custom AI chips and networking dominance position it to outgrow Nvidia through 2027 as hyperscalers prioritize efficient inference infrastructure."
Article claims Nvidia's latest full-year revenue hit $215B (65% growth), but that's incorrect—FY2025 ended at $130.5B (114% YoY). Broadcom's AI revenue doubled to $8.4B last quarter, forecasting >$100B by 2027 via custom ASICs and networking (99% internet traffic), complementing Nvidia GPUs but capturing hyperscaler inference/custom needs. Both ride $700B capex wave, yet article downplays ASIC shift eroding Nvidia's 80%+ GPU share. AVGO's diversified portfolio (smartphones to data centers) offers lower risk; Nvidia's 'dirt cheap' claim ignores ~40x fwd P/E vs AVGO's ~30x with dividend.
Nvidia's annual chip cadence (Rubin NVL144 imminent) and CUDA ecosystem lock-in could keep GPU demand insatiable, outpacing Broadcom's hyperscaler concentration risk.
"NVDA's valuation advantage over AVGO is real but modest, and the article conflates lower-than-peak multiples with true undervaluation while underestimating hyperscaler vertical integration risk to both players."
The article's core thesis — that NVDA is 'dirt cheap' on forward earnings — needs scrutiny. NVDA trades ~30x forward P/E; Broadcom ~25x. Neither is cheap by historical standards. The article conflates 'lower than peak valuation' with 'undervalued.' More critically: AVGO's $100B AI chip revenue forecast for 2027 is aspirational, not guaranteed. The 'deep relationships with hyperscalers' claim obscures a real risk: custom chips lock customers in, but also concentrate revenue among 6 players who have enormous negotiating power and can threaten vertical integration. NVDA's moat is broader. The article also omits that AVGO faces cyclical capex volatility — hyperscaler spending surges then plateaus. Finally, the 'not competing' framing is misleading; both benefit from AI infrastructure, but NVDA's pricing power and ecosystem lock-in are structurally superior.
If hyperscalers successfully develop in-house AI accelerators (as Google, Meta, and Amazon are doing), both NVDA and AVGO lose pricing power simultaneously — and AVGO's custom-chip model becomes a liability, not an asset, because customers can threaten to build their own custom silicon.
"NVIDIA remains the best AI stock for 2026, but the upside hinges on a persistent AI capex cycle and favorable regulatory conditions."
The article leans on Nvidia’s AI leadership and Broadcom’s non-core threat, but it glosses over key headwinds. AI demand is cyclical; hyperscalers may throttle capex if margins compress or funding rates change. Nvidia’s forward multiple hinges on sustained data-center growth and memory supply, which could disappoint if GPU supply surges or competition narrows pricing power. Regulatory/export controls on advanced chips, plus geopolitics with China, could curb growth or cap earnings upside. Broadcom’s growth is real but not a substitute for Nvidia’s end-market moat; margins and ecosystem leverage will matter as AI workloads diversify.
Nonetheless, the strongest counterpoint is that AI capex may stall if cloud customers optimize for efficiency or face macro headwinds, which would blunt Nvidia’s growth even as Broadcom scales its own AI infra. Also, we can't assume Nvidia's moat is untouchable; rivals and open-source accelerators could erode pricing and accelerate migration away from CUDA.
"Broadcom's custom ASIC model risks becoming a low-margin foundry service if hyperscalers move toward vertical integration."
Claude, your point on hyperscaler negotiating power is the missing link. If Google and Meta transition from 'buying' to 'designing' their own silicon, Broadcom’s custom ASIC model isn't a moat—it's a commoditization trap. They become a high-end foundry service rather than a product leader. While Grok highlights AVGO's diversification, that same breadth is a drag on operating margins compared to NVDA’s pure-play software-hardware vertical integration. The real risk is margin compression via customer-led vertical integration.
"AVGO's networking leadership provides a durable, high-margin moat in AI infrastructure scaling that complements NVDA rather than competes."
Gemini, hyperscaler vertical integration risks AVGO but ignores their optical networking dominance—essential for million-GPU AI clusters facing power bottlenecks. Jericho3-AI (1.6Tbps Ethernet) and Tomahawk5 switches capture 40%+ of AI data center traffic growth, with 75% gross margins. Nobody flags this: compute (NVDA) needs connectivity (AVGO) to scale, creating symbiotic demand not zero-sum erosion.
"Broadcom's networking dominance is temporary—hyperscalers will custom-design interconnects just as they're doing for AI accelerators, eroding AVGO's margin advantage."
Grok's networking moat is real but overstated. Yes, Jericho3-AI captures traffic growth—but that's a *scaling problem*, not a competitive advantage. As hyperscalers optimize clusters, they'll demand custom switching silicon too. Google's TPU ecosystem already includes custom interconnects. AVGO's 75% margins on switches compress if customers verticalize networking like they're doing compute. The symbiosis Grok describes becomes a liability when your customer controls both ends.
"In-house silicon shifts are not a guaranteed threat to NVDA/AVGO moats; software ecosystems and cyclical capex dynamics matter more than a straightforward erosion of pricing power."
Responding to Claude: the in-house silicon thesis isn’t guaranteed to strip NVDA’s or AVGO’s moats. Hyperscalers face huge software, verification and ecosystem costs to replace CUDA-like tooling, and multi-tenant workloads may resist standardization. NVDA’s software edge could persist even if hardware shifts, while AVGO’s appeal shifts toward end-to-end data-center fabrics rather than standalone ASIC revenue. The risk is more cyclical capex reliance than a clean, durable competitive erosion.
The panel agrees that the shift towards custom silicon (ASICs) poses a threat to Nvidia's dominance, but there's no consensus on whether Broadcom's ASICs represent a significant risk or opportunity. Nvidia's software moat and pricing power are seen as enduring strengths, while Broadcom's diversified portfolio and networking dominance are viewed as potential advantages.
Broadcom's growth in AI revenue and its dominant position in optical networking.
Margin compression due to customer-led vertical integration and potential supply chain or yield bottlenecks in Nvidia's next-gen cycles.