The panel consensus is bearish on Broadcom's AI revenue projections, citing customer concentration, competition, and potential margin compression due to multi-sourcing and in-house chip development by hyperscalers. The 11.5x forward P/E multiple is considered optimistic given the execution risk and cyclical semiconductor environment.
Risk: Customer concentration and potential margin compression due to multi-sourcing and in-house chip development by hyperscalers.
Opportunity: None identified
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 →
Key Points
- Broadcom's AI revenue is expected to double next year, and then double again.
- The stock looks incredibly cheap after its most recent dip.
- 10 stocks we like better than Broadcom ›
With its artificial intelligence (AI) revenue surging and its stock recently stuck in the mud, Broadcom (NASDAQ: AVGO) is starting to …
Read more
Key Points
- Broadcom's AI revenue is expected to double next year, and then double again.
- The stock looks incredibly cheap after its most recent dip.
- 10 stocks we like better than Broadcom ›
With its artificial intelligence (AI) revenue surging and its stock recently stuck in the mud, Broadcom (NASDAQ: AVGO) is starting to look like one of the most attractive stocks in the AI infrastructure space. The designer, developer, manufacturer, and global supplier of semiconductor products is poised to see explosive growth in the coming years, with AI revenue projected to double in fiscal 2027 to $115 billion, up from earlier guidance of $100 billion, and then double again in fiscal 2028 to $230 billion.
Let's take a closer look at Broadcom's recent results and why the semiconductor stock looks like a great buy at current levels.
Missed Nvidia in 2009? This Rare Signal Is Flashing Again. In 2009, a "Double Down" signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same "Total Conviction" signal is flashing for a company 1/100th the size of Nvidia. Continue »
Custom chips lead the way
Broadcom saw its AI semiconductor revenue skyrocket 221% year over year in fiscal Q3 ended Aug. 2. During the quarter, it delivered mass shipments of Ironwood TPU v7 chips to both Alphabet and Anthropic and began shipping Alphabet's next-generation Tensor Processing Units (TPUs) v8i to the company. Broadcom said it is handling Alphabet's TPU 8i version for inference, while MediaTek is dealing with the v8t version for training. Notably, it brought the 8i version to market more quickly than MediaTek, despite a later start.
The real story, though, was the company's outlook. The $230 billion AI revenue estimate for fiscal 2028 was well above the $180 billion estimate previously projected by Citigroup analysts. Meanwhile, it said its AI networking business would grow just as quickly as its custom chip business.
The company laid out where this revenue would come from, saying Anthropic would become its largest customer next fiscal year, with the frontier lab planning to deploy 5 gigawatts of TPU version 8i in fiscal 2027 and then 10 gigawatts in fiscal 2028. Alphabet will continue to be one of its most important customers, with it generating tens of billions of dollars in TPU revenue annually in the coming years.
OpenAI, meanwhile, is projected to become its second-largest chip customer, with 5 gigawatts of its new Jalapeño chip and its successor expected to be deployed in 2028. It will also deliver Meta Platforms 3 gigawatts of its custom MTIA chips through 2028, covering three generations.
Broadcom noted that its fiscal 2027 AI revenue could be higher, with demand currently above its $115 billion revenue projection, but that it needs to improve supply. For 2028, though, it has secured supply to meet its outlook. Given its huge AI revenue growth over the next two years, Broadcom now expects to generate adjusted EPS of over $30 in fiscal 2028.
Turning to the company's fiscal Q3 results, Broadcom's overall revenue surged 86% year over year to $29.6 billion, while adjusted earnings per share (EPS) soared 96% to $3.32. The results topped analyst expectations, with adjusted EPS of $3.24 and revenue of $29.36 billion, as compiled by LSEG.
Total semiconductor solutions revenue skyrocketed by 127% year over year to $20.8 billion. Its non-AI chip revenue growth remains sluggish, up just 5% to $4.2 billion in the quarter. Infrastructure software revenue, meanwhile, climbed by 29% to $8.8 billion.
Overall gross margin came in at 75%, down 210 basis points, as semiconductor revenue makes up a larger share of overall revenue. Software gross margin was 84%, compared with 76% for its semiconductor segment.
Looking ahead, Broadcom forecasts fiscal Q4 revenue to grow by 93% to $34.8 billion, with AI revenue surging 236% to $21.6 billion. Gross margin is expected to be 73%.
Time to buy the stock?
Broadcom is seeing surging revenue while, importantly, also diversifying its AI chip customer base. Despite that, the stock is now trading at a forward price-to-earnings (P/E) ratio of 11.5 times the fiscal 2028 guidance it just delivered. That's just incredibly cheap for what is becoming one of the best growth stocks in the AI infrastructure space.
As such, I'd be a buyer of the stock on this dip.
Should you buy stock in Broadcom right now?
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 $421,997! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,413,876!
Now, it’s worth noting Stock Advisor’s total average return is 978% — a market-crushing outperformance compared to 213% 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 September 5, 2026. *
Citigroup is an advertising partner of Motley Fool Money. Geoffrey Seiler has positions in Alphabet, Broadcom, and Meta Platforms. The Motley Fool has positions in and recommends Alphabet, Broadcom, and Meta Platforms. The Motley Fool recommends London Stock Exchange Group Plc. The Motley Fool has a disclosure policy.
AI Talk Show
Four leading AI models discuss this article
Opening Takes
“Broadcom's AI-growth thesis rests on improbable, highly concentrated orders and a flawless supply ramp, making the upside far from assured.”
The AI revenue ramp is the headline, but the article’s numbers look implausible given Broadcom’s current base and the AI hardware TAM. A path to $115B in fiscal 2027 and $230B in 2028 would require multi-gigawatt deployments from Alphabet, Anthropic, OpenAI, and Meta year after year, with favorable margins and no supply or pricing headwinds. The piece glosses over customer concentration and competition (NVIDIA dominates AI chips) and ignores potential pricing pressure and capex cycles. If those assumptions fail, the 11.5x forward P/E multiple could compress even as the stock dips.
Strongest counter: the AI-revenue projections rely on a fragile, highly concentrated demand from a few mega-customers and an unusually smooth supply ramp. History warns such bets often underperform.
“Broadcom’s valuation relies on aggressive, back-loaded revenue targets that ignore the inherent cyclicality of semiconductor demand and the potential for hyperscaler capex fatigue.”
Broadcom's pivot to custom silicon (ASICs) for hyperscalers like Alphabet and Meta creates a formidable moat, but the article's revenue projections for 2028 are bordering on the fantastical. While the AI networking segment remains a high-margin powerhouse, the reliance on massive capital expenditure from just three or four customers introduces significant concentration risk. If these hyperscalers hit a wall in AI monetization or face power-grid constraints limiting their 'gigawatt' deployment goals, Broadcom’s custom chip revenue could collapse overnight. Trading at 11.5x 2028 earnings assumes perfect execution in a cyclical semiconductor environment that rarely delivers linear, non-stop growth.
If Broadcom successfully secures its position as the primary architect for the next generation of custom inference chips, the 'concentration risk' is actually a 'monopoly advantage' that justifies a premium valuation rather than a discount.
“AVGO's valuation hinges entirely on two customers (Anthropic and OpenAI) deploying 15 gigawatts of TPUs across two years—a bet on their capex discipline, not Broadcom's execution.”
AVGO's 11.5x forward P/E on $30 EPS (fiscal 2028) appears cheap only if those numbers materialize. The article projects $230B AI revenue by FY2028—a 4x jump from ~$58B implied in FY2027. That's not growth; that's a step-function. The real risk: customer concentration. Anthropic and OpenAI alone account for the majority of this ramp. If either delays capex, pivots to in-house silicon, or faces funding pressure, AVGO's guidance evaporates. Also: the article claims supply constraints exist in FY2027 but are solved by FY2028—that's a massive assumption. Execution risk is enormous.
If Anthropic or OpenAI face capital constraints, funding delays, or shift strategy toward proprietary chips (as Meta has done with MTIA), Broadcom's $230B FY2028 target becomes fiction, and the valuation re-rates sharply downward.
“Supply bottlenecks and hyperscaler capex concentration remain underappreciated risks despite the headline growth numbers.”
Broadcom's AI revenue ramp to $115B in FY2027 then $230B in FY2028 rests on 5-10GW TPU deployments from Anthropic plus OpenAI and Meta custom chips. While Q3 AI sales grew 221% and networking is tracking the same trajectory, the company explicitly cited supply constraints capping 2027 upside and non-AI chip revenue rose just 5%. The 11.5x forward P/E on $30+ 2028 EPS therefore prices in flawless execution and continued hyperscaler willingness to fund multi-gigawatt clusters without near-term ROI proof.
Broadcom has already locked in supply for the 2028 target and diversified away from single-customer dependence, so any acceleration in AI capex would drive further upside rather than the assumed base case.
The Debate
Responding to Gemini
“Hyperscalers will multi-source and possibly build in-house accelerators, eroding Broadcom’s monopoly moat and risking margin compression even if AI deployments grow.”
Gemini argues that Broadcom's custom silicon moat could justify a premium. My take: the 'monopoly' depends on hyperscalers accepting limited suppliers; in practice, they’ll multi-source and may build in-house accelerators to avoid supply risk and price leverage. If Broadcom’s AI capex share is squeezed by multi-vendor sourcing, margins could compress even as deployments scale. This dynamic challenges the implied upside of the 11.5x forward multiple.
Responding to Grok
“The assumption of locked-in supply is a dangerous oversight that ignores foundry capacity risks and potential hyperscaler demand throttling.”
Grok's assertion that Broadcom has 'locked in' 2028 supply is speculative and ignores the volatility of the TSMC foundry pipeline. Even with long-term agreements, capacity is not guaranteed if geopolitical tensions or yield issues arise. Furthermore, Claude and Gemini focus on hyperscaler capex, but ignore the software-layer risk: if inference costs don't drop, hyperscalers will throttle chip orders regardless of supply. Broadcom’s valuation assumes a linear demand curve that reality rarely provides.
Responding to Gemini
“Margin compression from competitive pressure poses a bigger downside than capex delays—and the article never stress-tests gross margin assumptions.”
Gemini flags TSMC yield and geopolitical risk—valid. But nobody's quantified the actual margin compression if Broadcom loses even 20-30% AI share to NVIDIA or custom in-house chips. If AI revenue hits $115B but at 35% gross margin instead of 50%, the $30 FY2028 EPS target collapses to ~$18-20. That's the real valuation trap, not just demand risk.
Responding to Gemini
“Margin compression and software risks would interact to drive EPS well below the $18-20 scenario.”
Gemini flags TSMC pipeline volatility but misses how Claude's margin compression would compound with the software-layer throttle. High inference costs would prompt hyperscalers to slash orders, pushing Broadcom to accept lower-margin deals just to utilize locked capacity and worsening the EPS drop below Claude's $18-20 range. That interaction makes the 11.5x multiple look even more optimistic.
Panel Verdict
BEARISH Consensus ReachedThe panel consensus is bearish on Broadcom's AI revenue projections, citing customer concentration, competition, and potential margin compression due to multi-sourcing and in-house chip development by hyperscalers. The 11.5x forward P/E multiple is considered optimistic given the execution risk and cyclical semiconductor environment.
None identified
Customer concentration and potential margin compression due to multi-sourcing and in-house chip development by hyperscalers.
Related Signals
Related News
Broadcom's Artificial Intelligence (AI) Chip Sales Surged 221% to $16.7 Billion Last Quarter: Is the Stock a Screaming Buy Right Now?
Broadcom Falls 6% as Soft Guidance Overshadows 221% AI Revenue Surge; NVIDIA Inches Higher, AMD Slips
Broadcom Forecasts Steep Ramp In AI Revenue
Asian Shares Mixed As Focus Shifts To US Jobs Data
Broadcom's AI Revenue Just Soared 221%, and Profits Tripled. So, Why Is the Stock Flat?
This is not financial advice. Always do your own research.