By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
Panelists express caution about Broadcom's AI ambitions, highlighting execution risks, customer concentration, and potential margin compression due to vertical integration or shifting AI workloads.
Risk: Obsolescence cycle and decline in networking demand due to shifting AI workloads (Gemini)
Opportunity: Potential growth in AI revenue despite reduced networking spend per accelerator (Claude)
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 →
Broadcom has a path to $100 billion in artificial intelligence (AI) revenue in 2027, thanks to some blockbuster partnerships for its XPU AI chips.
The company is well positioned as AI's technological needs shift from model training to real-world use.
Broadcom stock isn't cheap anymore, but there's enough value to consider buying and holding shares.
Up until recently, Broadcom (NASDAQ: AVGO) was having a poor 2026. The stock declined by roughly 15% from January through March, only to begin a surge in April that has shares sitting up 23% year to date. However, Broadcom stock has been a remarkable investment for years now, up a staggering 860% over the past five years alone.
Investors who passed on buying the recent dip may be kicking themselves, but it's not necessarily too late to buy the stock if you're willing to hold shares for a while. After all, buying the dip over the past five years has continued to work in your favor.
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 »
Here's why Broadcom is likely still worth buying, despite its recent run-up to yet another all-time high.
Broadcom's success isn't anything new. The company has been a fantastic stock to own for years preceding the artificial intelligence (AI) boom. Broadcom has long been a leader in networking chips and has diversified into enterprise infrastructure software.
What's new and exciting for Broadcom is its rise to the forefront of custom silicon for AI hyperscalers. Many of the largest AI companies investing in data centers have begun looking to alternatives to Nvidia's GPUs. As a result, Broadcom has leveraged its chip expertise to design XPUs for several high-profile customers, including Alphabet, Anthropic, OpenAI, and Meta Platforms. Broadcom customizes its XPUs for each client's needs, resulting in more efficient performance.
Nvidia isn't going away, as there's simply too much need for computing power in today's AI landscape. The pie is plenty large enough for everyone to eat. That said, it's never wise to put all the eggs in one metaphorical basket. These AI companies wisely don't want to give Nvidia too much control over their AI infrastructure.
Broadcom's XPU business is still in its early innings. These chips are especially useful for inference workloads, which involve applying AI models to real-world applications. The story had been about training AI models, but as actual AI adoption continues, there could be an increasing shift from training to inference.
The company earned $20 billion from AI last year, with much of that coming from networking chips. As these XPU deals begin to ramp up, look for AI revenue to skyrocket. CEO Hock Tan has hinted that AI revenue could surge past $100 billion by next year, a potential fivefold increase in Broadcom's AI sales.
Broadcom's total revenue in 2025 was $63.9 billion, so the XPU opportunity could more than double the company's size in relatively short order. That explosive growth is why Broadcom's valuation is more attractive than you might expect for a stock that's risen so much, so quickly.
Sure, it would have been better to have bought Broadcom stock a couple of months ago, when it was in a slump. The stock's recent run has pushed its price-to-earnings ratio from under 60 to over 80 in a matter of weeks. But there is still enough value here to justify hitting that buy button.
The consensus among Wall Street analysts is that Broadcom's earnings will grow at an annualized rate of 41% over the next three to five years. That makes sense, considering the anticipated windfall in AI revenue from Broadcom's XPU deals.
So, while Broadcom stock trades at a very high valuation, it has the growth to justify it. That's a PEG ratio of about 2, which isn't a bargain, but it's reasonable for investors willing to buy and hold shares, letting the business grow into that valuation over time.
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 $483,476! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,362,941!
Now, it’s worth noting Stock Advisor’s total average return is 998% — 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.
**Stock Advisor returns as of May 19, 2026. *
Justin Pope has positions in Alphabet and Meta Platforms. The Motley Fool has positions in and recommends Alphabet, Broadcom, Meta Platforms, 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
"High customer concentration and unproven XPU scale-up create more downside risk than the article acknowledges at current multiples."
Broadcom's custom XPU deals with Alphabet, Meta, OpenAI and Anthropic position it to capture inference demand as AI shifts from training. Yet the article underplays execution risk on these still-ramping programs and customer concentration among a handful of hyperscalers. At 80x forward earnings against 41% projected growth, the PEG near 2 leaves little margin if XPU revenue ramps slower than the $100 billion target or if Nvidia retains more share than expected. Networking chips drove most of last year's $20 billion AI revenue; new silicon contributions remain uncertain.
If the XPU pipeline converts faster than modeled and gross margins hold above 60%, the current multiple could compress toward 50x while EPS doubles, rewarding holders who ignore near-term valuation.
"Broadcom's valuation assumes flawless XPU execution and hyperscaler commitment over three years—a binary bet disguised as a growth story."
The article conflates two separate narratives: Broadcom's proven infrastructure dominance (networking, software) versus speculative XPU upside. The $100B AI revenue claim by 2027 is CEO commentary, not guidance—a 5x jump in three years requires hyperscalers to actually deploy these chips at scale and for Broadcom to capture meaningful share against entrenched Nvidia. The PEG of ~2.0 looks reasonable only if 41% EPS CAGR materializes; if XPU adoption disappoints or hyperscalers vertically integrate further, that multiple compresses hard. The article also omits that Broadcom's core networking business is mature and cyclical—AI revenue is priced in, not a surprise.
If hyperscalers face inference cost pressures and Nvidia's software moat (CUDA, ecosystem) proves stickier than expected, or if custom silicon becomes a bargaining chip that erodes Broadcom's pricing power, the $100B thesis collapses and AVGO reverts to a 40-50x multiple on slower growth.
"Broadcom’s valuation has decoupled from its historical M&A-driven growth model, pricing in a level of AI-revenue certainty that ignores the inherent volatility of custom silicon partnerships."
Broadcom’s pivot from a networking hardware play to a custom silicon powerhouse is impressive, but the valuation expansion is concerning. Trading at an 80x P/E ratio is a massive premium for a company that historically relied on disciplined M&A to drive growth. While the $100 billion AI revenue target by 2027 is the bull case, it assumes flawless execution in the highly competitive custom ASIC market. Investors are essentially paying for perfection in an environment where hyperscalers like Meta and Google are notoriously fickle partners who may eventually bring more of this design work in-house to capture higher margins, compressing Broadcom's long-term profitability.
If Broadcom successfully secures its role as the 'TSMC of custom AI silicon' for the next decade, the current 80x P/E is merely a growth premium that will look cheap once their XPU margins scale.
"Broadcom can sustain a multi-year re-rating if XPUs ramp to $100B in AI revenue by 2027, driven by select hyperscalers and durable margins; otherwise the stock faces multiple compression."
The article frames Broadcom's XPUs as a near-term AI jackpot, with AI revenue potentially surpassing $100B by 2027 and a PEG near 2.0. That reframes Broadcom as not just a chips supplier but a critical AI infrastructure enabler alongside Nvidia. Yet the bullish thesis rests on aggressive execution: ramping custom XPUs with a few hyperscalers, maintaining margins amid compute hardware cycles, and avoiding competitive pressure from Nvidia/AMD, or customer exit risk. Valuation looks rich after miles of outperformance, leaving little cushion if AI capex slows, a key customer renegotiates, or a new competitor erodes pricing power.
The $100B AI-revenue path is highly uncertain and could prove overly optimistic. XPUs may remain a smaller, rate-limited contributor amid fierce competition and customer concentration, capping margins and upside.
"Successful XPU adoption might reduce per-unit networking demand, limiting total revenue upside."
Grok flags that networking chips drove most of last year's $20 billion AI revenue while questioning new silicon contributions. However, an overlooked risk is that successful XPU adoption could allow customers to redesign systems reducing overall networking spend per accelerator, potentially capping Broadcom's total AI revenue even if XPU volumes grow. This would pressure the 41% EPS growth assumption underpinning the current valuation.
"Cannibalization matters less than customer leverage over Broadcom's pricing power once XPU volumes scale."
Grok's cannibalization risk is sharp, but I'd push back: if XPUs reduce per-accelerator networking demand by 20% while XPU ASPs stay 3-4x higher, total AI revenue still grows. The real pressure isn't system redesign—it's whether hyperscalers accept Broadcom's pricing or force vertical integration faster. That's the margin cliff nobody's quantifying. How much pricing power evaporates if Meta or Google threatens to build XPUs in-house?
"Broadcom’s valuation relies on a networking-heavy AI architecture that may face structural obsolescence as inference workloads evolve away from massive bandwidth requirements."
Claude, your focus on vertical integration ignores the 'Broadcom Tax'—the sheer engineering complexity of high-speed SerDes and interconnects that hyperscalers struggle to replicate. Even if Google pulls design in-house, they remain tethered to Broadcom’s IP for connectivity. The real risk isn't just pricing power; it's the obsolescence cycle. If AI workloads shift from training to inference, the demand for Broadcom’s massive networking throughput may actually decline, regardless of XPU volume, making the 41% CAGR look like a fantasy.
"Hyperscaler vertical integration timing could erode Broadcom’s revenue and margins faster than the article’s $100B by 2027 thesis implies."
Responding to Claude: Vertical integration risk matters, but the more fragile hinge is timing and scale of XPUs. If any hyperscaler balks at Broadcom’s pricing and moves in-house, the revenue impact compounds fast because the base is a tight group of customers. The '100B by 2027' thesis collapses not only on XPU adoption misses but on how quickly networking budgets reallocate to internal IP, compressing Broadcom’s EBITDA if scale leverage erodes.
Panelists express caution about Broadcom's AI ambitions, highlighting execution risks, customer concentration, and potential margin compression due to vertical integration or shifting AI workloads.
Potential growth in AI revenue despite reduced networking spend per accelerator (Claude)
Obsolescence cycle and decline in networking demand due to shifting AI workloads (Gemini)