Broadcom Raises Its AI Revenue Target to $115 Billion, Yet Shares Still Fall: The Market Now Pays On

$Broadcom(AVGO)$ $Dell Technologies Inc.(DELL)$ $Palo Alto Networks(PANW)$

Broadcom delivered record results and raised its fiscal 2027 AI revenue target to $115 billion. However, its fourth-quarter revenue guidance fell about 0.7% short of market expectations—and that small gap still sent the stock down more than 6% at one point after hours.

After the market closed on September 2, Broadcom $AVGO reported its fiscal third-quarter 2026 results:

  • Revenue reached $29.591 billion, up 86% year over year and above the $29.36 billion consensus estimate;

  • Adjusted earnings per share came in at $3.32, ahead of the $3.24 estimate;

  • Free cash flow reached $13.665 billion, equivalent to 46% of revenue;

  • AI semiconductor revenue reached $16.7 billion, up 221% year over year and 54% sequentially.

The company also expects fourth-quarter AI semiconductor revenue to rise to $21.7 billion, representing growth of 236% year over year.

Almost every major indicator—from revenue and earnings to the AI business—continued to grow. Yet Broadcom shares fell more than 6% at one point after hours before narrowing the decline to around 3.5%. (investors.broadcom.com)

The reaction shows how the standard for evaluating AI infrastructure earnings has changed. Beating past performance is no longer enough; companies must also exceed the expectations already embedded in their share prices.

1. How Strong Was the Report?

Broadcom reported record quarterly revenue, operating profit and free cash flow.

AI semiconductor revenue reached $16.7 billion, accounting for 56% of total revenue. This segment includes both custom AI accelerators and the networking chips used to connect large-scale computing clusters.

Management expects:

  • Fiscal 2026 AI semiconductor revenue of approximately $58 billion, up 186% year over year;

  • Fiscal 2027 AI semiconductor revenue to double again to approximately $115 billion, up from its previous target of more than $100 billion;

  • Fiscal 2028 AI semiconductor revenue to double once more to approximately $230 billion.

Broadcom said it has largely secured the supply needed to support these targets, while customer demand continues to exceed its current outlook.

If these targets are achieved, Broadcom’s AI business alone would grow within two years to a scale larger than most major semiconductor companies.

2. Why Did a 0.7% Gap Overshadow the Entire Report?

The issue was Broadcom’s fourth-quarter revenue guidance.

The company expects revenue of approximately $34.8 billion, up 93% year over year. The LSEG consensus estimate was $35.03 billion, leaving a gap of around $230 million, or approximately 0.7%.

Revenue of $34.8 billion is not weak. Its implied 93% growth rate would even represent an acceleration from the third quarter’s 86% growth.

However, earnings trades are driven by the difference between reported results and market expectations—not by absolute growth alone. Broadcom had already presented investors with an enormous long-term AI revenue target, so the market also expected its near-term guidance to come in clearly above consensus.

As a result, the positive impact of the higher long-term AI target was offset by fourth-quarter revenue guidance that came in slightly below expectations.

This is an increasingly visible risk during the current AI earnings cycle: good results may only be enough to maintain a valuation, while another leg higher requires a substantial beat.

3. What Does the $115 Billion Target Mean?

Broadcom’s core advantage lies in its custom AI accelerators, or XPU business.

Once large technology companies have stable, large-scale AI workloads, they can design specialized chips around their own models to improve energy efficiency, cost and inference performance.

Broadcom currently has six XPU customers. Their potential demand roadmaps represent approximately 30 gigawatts in total, but management cautioned that not all 30 gigawatts will necessarily be deployed during fiscal 2027 and 2028.

Land, power, data-center construction, HBM availability and substrate supply could all affect the final delivery schedule. (www.investing.com)

Disclosed deployment plans include:

  • Anthropic is expected to add 5 gigawatts in 2027 and approximately 10 incremental gigawatts in 2028;

  • OpenAI plans to deploy approximately 1.3 gigawatts in 2027 and more than 5 gigawatts in 2028;

  • Meta is expected to deploy approximately 3 gigawatts in total by 2028;

  • Google has signed a long-term agreement with Broadcom covering multiple generations of TPUs and AI networking products.

The $115 billion target is therefore closer to a revenue outlook built around customer deployment roadmaps. Its eventual realization still depends on whether data centers are completed on schedule, chips enter production smoothly and customer capital expenditure remains strong.

Management also said the fiscal 2027 and 2028 targets are intended to illustrate the company’s long-term growth trajectory and will not be updated every quarter.

The hard indicators investors can track will therefore remain quarterly AI revenue and actual deployment progress. (www.investing.com)

4. What Else Is the Market Worried About?

1. Product Mix Is Affecting Gross Margin

Broadcom’s consolidated gross margin was 75% in the third quarter, down 2.1 percentage points sequentially. The company expects it to fall further to approximately 73% in the fourth quarter.

The main reason is the increasing contribution from custom AI chips. XPUs contain more high-cost memory and complex components, giving them lower gross margins than Broadcom’s infrastructure software business.

However, Broadcom’s non-GAAP operating margin still reached 67.9%, an increase of 2.4 percentage points year over year. The scale benefits created by rapid revenue growth partly offset the effect of the lower-margin AI product mix. (www.investing.com)

Gross-margin pressure deserves attention, but it does not yet mean Broadcom’s overall profitability is deteriorating.

2. Capital Expenditure Is Increasing

Broadcom spent $532 million on capital expenditure in the third quarter and expects that figure to rise to $1.4 billion in the fourth quarter.

The additional investment will support semiconductor capacity, including substrate production in Singapore and indium phosphide capacity used for optical components. Management expects its Singapore substrate facility to begin operations in fiscal 2027. (www.investing.com)

Whether Broadcom can convert its orders into revenue will increasingly depend on supply-chain execution, data-center construction and capacity expansion.

3. Customer Concentration Remains High

Six XPU customers account for most of the growth in Broadcom’s AI business, with Anthropic, OpenAI, Google and Meta planning particularly large deployments.

These customers have significant financial resources, but the concentration also means Broadcom’s long-term revenue is becoming increasingly dependent on the budgets, financing capacity and construction schedules of a small number of technology companies.

5. DELL and PANW Sent the Same Message on the Same Day

On September 2, Dell Technologies $DELL surged by double digits after strong AI-server demand led the company to raise its full-year revenue and profit guidance.

Palo Alto Networks $PANW also reported results above expectations, yet its shares fell by approximately 9%. (www.reuters.com)

The common thread across all three companies is clear: beating expectations has become the entry requirement for earnings season. What determines the stock-price reaction is the size of the beat.

Dell started from a lower expectations baseline, so its substantial full-year guidance increase received a strongly positive response.

Broadcom was already carrying an AI revenue expectation of more than $100 billion. As a result, even after raising that target to $115 billion, a 0.7% gap in its near-term revenue guidance was enough to attract attention.

For the AI infrastructure chain, earnings-day risk is becoming more asymmetric: a miss can trigger a rapid decline, while a modest beat may not generate a comparable gain.

Related Stocks

  • $AVGO: A core name in custom AI accelerators and networking chips;

  • $MRVL: A direct comparable in custom ASICs and networking;

  • $MU: Exposure to HBM and AI-related memory demand;

  • $ANET: Exposure to Ethernet networking for AI data centers;

  • $SMH and $SOXX: Semiconductor-sector ETFs.

Tiger’s View

This report does not suggest that AI demand is cooling.

Quarterly AI semiconductor revenue of $16.7 billion, fourth-quarter guidance of $21.7 billion and the fiscal 2027 target of $115 billion all indicate that custom AI chips remain in a period of rapid expansion.

What has changed is how the market values that growth.

Investors were previously willing to assign higher valuations directly to larger long-term forecasts. Now, they are checking delivery schedules, gross margins, capital expenditure and customer budgets quarter by quarter.

The long-term story still matters, but the timing of revenue recognition is receiving more weight.

Key indicators to watch next:

  1. Whether fourth-quarter AI semiconductor revenue reaches $21.7 billion;

  2. Whether consolidated gross margin stabilizes at approximately 73%;

  3. Whether the planned $1.4 billion in capital expenditure can ease substrate and optical-component bottlenecks;

  4. Whether data centers for customers such as Anthropic and OpenAI are completed on schedule;

  5. Whether future earnings beats can once again produce positive stock-price reactions.

Short-term after-hours moves can still diverge from the following regular trading session, so using the after-hours decline alone to determine direction carries considerable risk.

How do you view Broadcom’s decline after raising its long-term outlook?

A. The market is reasonably discounting execution risk
B. The reaction is excessive—the $115 billion target is not fully priced in
C. Gross-margin pressure is the bigger concern
D. Customer concentration reduces the credibility of the long-term outlook

Sources: Broadcom Fiscal Q3 Results | Reuters | Broadcom Earnings Call Transcript

This content is provided for market discussion only and does not constitute investment advice.

# Nvidia Reclaims AI Narrative — But at What Cost to Gross Margins?

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Report

Comment5

  • Top
  • Latest
  • 苏36
    ·10:11
    TOP
    I’d choose A: The market is reasonably discounting execution risk.

    Broadcom’s results were exceptional: AI semiconductor revenue jumped 221% YoY to $16.7 billion, while management raised its FY2027 AI revenue target to $115 billion. The long-term AI story clearly remains intact.

    But expectations have changed. With so much optimism already priced in, investors are no longer asking whether AI demand is strong—they are asking whether Broadcom can deliver the chips on schedule, maintain margins and convert customer roadmaps into actual revenue.

    The 0.7% Q4 guidance gap is tiny fundamentally, but meaningful when valuation and expectations are this high.

    To me, this isn’t an AI warning. It’s a “show me the execution” moment. Broadcom can still win—but the bar is now extremely high.

    @Tiger_comments [暗中观察]

    Reply
    Report
  • 靖润
    ·14:34
    选A:合理折价交付风险。


    博通上调长期AI收入目标到1150亿,但四季度指引比预期低0.7%,股价盘后跌。问题不是这0.7%有多大,是市场现在的定价逻辑变了——远期目标调得再高,当期指引有一点点缺口,市场先折价再说。


    戴尔那470亿到账了不用折,博通这1150亿还在路上,得按4.78%重新算。客户集中度、执行风险都在定价里,不是市场没看到上调目标,是把上调目标的时间成本也折进去了。


    超额交付才给溢价,符合预期就得先交折扣。这是当前市场环境下的定价原则,也是博通盘后跌的原因。
    Reply
    Report
  • AfraSimon
    ·10:12
    B for now. The miss was tiny, but customer concentration is doing more damage than the headline target. If Anthropic or OpenAI insource faster, that 115B starts looking way less solid
    Reply
    Report
  • Joy34
    ·10:12
    A for me. Does a 0.7% guide miss really explain a 6% drop? I care more about that 73% gross margin and whether AI revenue is getting more expensive to scale.
    Reply
    Report
  • Kipbana
    ·11:20
    Going to see a strong rally for Nvidia
    Reply
    Report