tapebrief
AVGO · Q4 2025

AVGO · Q1 2026 Earnings

Bullish

Broadcom

Reported March 4, 2026

30-second summary

30-second take. Broadcom delivered Q1 FY2026 revenue of $19.31B (+29.4% YoY, +1.1% above the $19.11B consensus), with AI semiconductor revenue of $8.4B (+106% YoY) beating the $8.2B prior guide and Q2 FY2026 guided to $22.0B with AI semis stepping to $10.7B (+140% YoY) — a sharp re-acceleration from Q1's 106% pace. Consolidated non-GAAP gross margin landed at 77% versus a guide of ~76.9% (down ~100bps QoQ from the Q4 77.9% base) — modestly above guide. The signal that matters: management quantified 2027 AI chip-only revenue at "in excess of $100 billion," named OpenAI as the sixth qualified XPU customer with volume deployment in 2027, and confirmed capacity is fully secured through 2028. This is the most quantitatively committed forward statement Broadcom has ever made, and it explicitly takes off the table both the "is the backlog real" and "will supply gate the ramp" debates.

Headline numbers

EPS

Q1 FY2026

$2.05

+1.5% vs est.

Revenue

Q1 FY2026

$19.31B

+29.4% YoY

+1.1% vs est.

Gross margin

Q1 FY2026

68.1%

Free cash flow

Q1 FY2026

$8.01B

Operating margin

Q1 FY2026

44.3%

Key financials

Q1 FY2026
MetricQ1 FY2026YoYQ4 FY2025QoQ
Revenue$19.31B+29.4%$18.02B+7.2%
EPS$2.05—$1.95+5.1%
Gross margin68.1%—68.0%+12bps
Operating margin44.3%—41.7%+263bps
Free cash flow$8.01B—$7.47B+7.3%

Guidance

Guidance is issued one quarter forward. The Prior-guide column references the guide issued last quarter for the period just reported; the New-guide column is for next quarter.

Actuals vs prior guidance

MetricPeriodPrior guideActualΔResult
RevenueQ1 FY2026approximately $19.1 billion$19.311 billionin-lineMet
Revenue YoY growthQ1 FY202628%29.4%+1.4pts above guideBeat
Adjusted EBITDA marginQ1 FY202667% of projected revenue68%in-lineMet
AI Semiconductor RevenueQ1 FY2026$8.2 billion$8.4 billion+$0.2B above guideBeat
Non-AI Semiconductor RevenueQ1 FY2026approximately $4.1 billion$4.115 billion+$0.015B above guideBeat
Infrastructure Software RevenueQ1 FY2026approximately $6.8 billion$6.796 billionin-lineMet
Gross MarginQ1 FY2026down approximately 100 basis points sequentially68.1%above guideBeat

New guidance

MetricPeriodGuideYoY
RevenueQ2 FY2026approximately $22.0 billion+47% YoY
Semiconductor RevenueQ2 FY2026approximately $14.8 billion+76% YoY
AI Semiconductor RevenueQ2 FY2026$10.7 billion+140% YoY
Infrastructure Software RevenueQ2 FY2026approximately $7.2 billion+9% YoY
Consolidated Gross MarginQ2 FY202677% (flat sequentially)—

Segment performance

Q1 FY2026
SegmentQ1 FY2026YoY
Semiconductor Solutions$12.515B+52.4%
Infrastructure Software$6.796B+1.4%

Capacity & utilization

Q1 FY2026
SegmentQ1 FY2026
AI Revenue$8.4B
AI Revenue Growth106% YoY
Q2 FY2026 AI Revenue Guidance$10.7B

Profitability

Q1 FY2026
SegmentQ1 FY2026
Adjusted EBITDA$13.1B
Adjusted EBITDA Margin68%
Free Cash Flow$8.0B
FCF Margin41%
Operating Cash Flow$8.3B

Management tone

Narrative arc: prior-quarter customer additions and backlog quantification → Q1 FY2026 sixth customer (OpenAI) + $100B+ 2027 chip-revenue commitment.

Management committed a dollar floor on FY2027 AI chip revenue. Management leapfrogged the FY2026 quantification question entirely by anchoring to 2027: "we have line of sight to achieve AI revenue for from chips just chips in excess of $100 billion in 2027." The willingness to bypass FY2026 and commit a year further out tells you management thinks the FY2026 number is so obviously large that the analytical battle has moved one year forward.

The customer-addition cadence continued, and the sixth customer is OpenAI. "We also now have a sixth customer. We expect OpenAI to deploy in volume their first generation XPU in 2027 and over one gigawatt of compute capacity." Naming OpenAI explicitly signals the agreement has hardened past the point where ambiguity served either party.

Management directly rebutted analyst skepticism on Meta's MTIA — a tonal posture that is genuinely new. Broadcom historically avoids commenting on specific customers under any circumstance. This quarter: "Contrary to recent analyst reports, Meta's custom accelerator MTIA roadmap is alive and well. We're shipping now. And in fact, for the next generation XPUs, we will scale to multiple gigawatts in 27 and beyond." The shift to naming customers and defending their roadmaps publicly reflects either a different competitive calculus or a level of confidence in the customer pipeline that no longer needs protective ambiguity.

Supply chain framing shifted from a constraint to a competitive moat. This quarter: "we have fully secured capacity of these components for 26 through 28." The capacity narrative has flipped — what was a question mark has become a stated barrier-to-entry against customer-owned-tooling competitors. Management's framing in the Harlan Sur Q&A — that COT alternatives are "2x less performant and complex" — depends on Broadcom being the only party that can deliver leading-edge wafers, HBM, and packaging at volume.

Infrastructure software language stayed defensive while the print decelerated. Q1 software at +1.4% YoY is the slowest growth in the post-VMware era. Management's response is theological rather than operational: "VCF cannot be disintermediated or replaced... AI will create the need for more VMware, not less." The Q2 guide of +9% YoY is an improvement but still half the FY26 "low double-digit" framing carried over from last quarter. This is the one part of the story where the language and the numbers are diverging, and neither this brief nor any analyst on the call pressed on it adequately.

Recurring themes management leaned on this quarter:

Custom AI XPU franchise acceleration across six customers2027 AI revenue visibility exceeding $100 billionSupply chain security and capacity assurance through 2028AI networking gaining share with Tomahawk 6/7 leadershipInfrastructure software indispensability in AI-native data centersOperating leverage driving 68% adjusted EBITDA margins

Risks management surfaced:

Constrained capacity in leading edge wafers and high bandwidth memoryCompetitive optical networking alternatives requiring more powerCustomer concentration in custom accelerator businessSupply chain component availability for substrate and HBMExecution risk scaling to multiple gigawatts across six customers

Q&A highlights

Blaine Curtis · Jefferies

Clarification on >$100B AI chip revenue forecast (ASICs vs networking vs RAC), and perspective on investor pessimism regarding hyperscaler ROI requirements on AI infrastructure investments

Management clarified the $100B+ revenue is silicon-based (XPUs, switch chips, DSPs). On ROI concerns, emphasized that customers are focused on creating LLMs and platforms for enterprise/consumer use, driving sustained demand for both training and inference compute capacity. Noted customers are designing custom accelerators and networking architectures, supporting long-term growth.

Revenue forecast for 2027 significantly exceeds $100 billionRevenue composition: XPUs, switch chips, DSPs (silicon content)Customers include 5-6 players creating their own custom acceleratorsStrong demand for both training and inference capacity

Harlan Sir · JP Morgan

Assessment of competitive threat from customer-owned tooling (COT) initiatives by hyperscalers developing internal XPU/TPU designs, and how Broadcom maintains technical leadership given current 12-18 month advantage

Management stated COT initiatives face tremendous challenges (technology, silicon design, packaging, clustering expertise) and are 2x less performant and complex than current Broadcom solutions. Emphasized COT competitors won't be meaningful for many years due to need for best-in-class performance versus Nvidia, execution risk, and production volume capability. Broadcom plans to continue widening gap through next-gen products.

COT competitors are 2x less performant and complex than Broadcom solutions12-18 month technical lead in XPU designProduction volume and yield expertise critical differentiatorNext-generation products (Tomahawk 7) launching with 2x performance in 2027

Ross Seymour · Deutsche Bank

What is driving networking to 40% of AI revenues in near-term, and what is expected mix composition in the $100B+ 2027 revenue forecast; leadership position expectations in scale-out and scale-up

Growth driven by 200G bandwidth GPUs/XPUs, Tomahawk 6 (only 100 terabit/sec switch) providing highest bandwidth, and 1.6 terabit DSP optical transceivers. Management expects networking to normalize to 33-40% range of total AI revenue long-term. Next-gen Tomahawk 7 launching in 2027 with 2x performance. Optimization across compute and networking sides provides competitive advantage.

Networking currently 40% of AI revenuesTomahawk 6: only 100 terabit/sec switch available1.6 terabit DSP optical transceivers (only player)Tomahawk 7 launching 2027 with 2x performance

Timothy Arcuri · UBS

Impact of AI rack shipments on gross margin; whether margin compression of ~500bps is expected from lower-margin rack business, and if there's a floor below which Broadcom won't pursue racks

Management firmly rejected margin compression thesis, stating gross margin will not be affected by increasing AI product mix. Clarified that yields and costs have been optimized such that AI model margins are consistent with rest of semiconductor business. CFO confirmed minimal impact from overall mix shift.

Gross margin remains solid despite AI product mix shiftAI business margins consistent with overall semiconductor business modelYields and costs optimized to support margin consistencyNo anticipated 500bps margin compression from rack shipments

Stacey Raskon · Bernstein

Validation of gigawatt deployment math for 2027 ($100B+ revenue); estimated 8-9 gigawatts based on known customer commitments and $20B per gigawatt content assumptions

Management confirmed approximately 10 gigawatts of deployments expected in 2027. Acknowledged that dollars per gigawatt varies significantly by customer, but confirmed overall magnitude consistent with analyst calculations. Supported $20B-range content per gigawatt as reasonable approximation.

Approximately 10 gigawatts of deployments expected in 2027Content per gigawatt varies by customerAggregated content per gigawatt approximately $20 billion range

Answers to last quarter's watch list

Q1 FY2026 AI semiconductor revenue vs. the $8.2B guide (+100% YoY). AI semis came in at $8.4B (+106% YoY), beating the doubling guide by $200M and 6pts. More importantly, management leapfrogged FY2026 quantification by committing to FY2027 chip revenue "in excess of $100B" — converting the multi-quarter floor into a multi-year floor.
Resolved positively
Q1 FY2026 gross margin landing vs. the down ~100bps QoQ guide. Consolidated non-GAAP gross margin landed at 77% vs. a guide that implied roughly 76.9% on the prior-quarter non-GAAP basis of 77.9% — modestly above guide. Q2 FY2026 is now guided to 77% flat QoQ, and management explicitly rejected the ~500bps AI rack compression thesis in Q&A.
Resolved positively
Sixth XPU customer qualification. Sixth customer named explicitly as OpenAI, with volume XPU deployment in 2027 and over 1 GW of compute capacity.
Resolved positively
AI backlog growth from $73B. Management did not refresh the consolidated backlog dollar figure this quarter — the new disclosure framework anchors to 2027 chip revenue ($100B+) and gigawatt deployment scale (~10 GW), not a rolling 18-month backlog. The metric has effectively been retired in favor of forward revenue commitments.
Not resolved
Non-AI semiconductor Q1 print vs. the $4.1B flat-YoY guide. Non-AI semis came in at ~$4.1B, in-line with the flat-YoY guide. Management offered no recovery commentary; non-AI continues to be the stagnant counterweight to the AI ramp. The structural read — Broadcom is now an AI-plus-software company — is confirmed for another quarter.
Resolved negatively
OpenAI XPU program disclosure. OpenAI is now explicitly named as the sixth qualified XPU customer with 2027 volume deployment and >1 GW. The parallel XPU program flagged last quarter as "advanced stage" has graduated to a named customer commitment in one quarter — the fastest conversion in the AI customer pipeline to date.
Resolved positively

What to watch into next quarter

Q2 FY2026 AI semiconductor revenue vs. the $10.7B guide (+140% YoY). A beat extends the sequential AI dollar add above $2.3B and pressure-tests the FY27 $100B+ glide path. An in-line print is still a +140% YoY quarter — but anything softer than that re-opens the deceleration debate.

Q2 gross margin holding at 77% flat QoQ vs. UBS's 500bps compression bear case. Management has now planted a flag that AI margins match the corporate average; a step-down would force a reframe back to "dollars not percentages" and undermine the credibility upgrade.

Seventh XPU customer or first concrete OpenAI 2026 revenue disclosure. The one-new-customer-per-quarter cadence is now structural. Either a seventh customer or a disclosed dollar contribution from OpenAI for 2026 would mark the next discrete catalyst beyond the $100B headline.

Infrastructure software Q2 print vs. the $7.2B (+9% YoY) guide and any FY26 reaffirmation. Q1 at +1.4% YoY and Q2 guide at +9% YoY do not bridge to the previously-framed FY26 "low double-digit" growth. Either software accelerates materially in H2 or the FY26 software framing gets quietly walked back.

Capex disclosure as the multi-customer ramp accelerates. Singapore packaging facility, HBM commitments, and substrate capacity through 2028 imply real capital intensity that has not yet shown up in Broadcom's fabless-style capex line. A material step-up would inform the FCF conversion model into FY27.

Any 2026 AI revenue dollar quantification. Management has now committed to 2027 chip revenue but explicitly skipped FY2026. A concrete FY26 number — even a range — at the Q2 call would convert the FY27 commitment into a fully bracketed multi-year glide path.

Sources

  1. Broadcom Q1 FY2026 press release (8-K exhibit, filed 2026-03-04): https://www.sec.gov/Archives/edgar/data/1730168/000173016826000011/avgo-02012026x8kxex99.htm
  2. Broadcom Q1 FY2026 earnings conference call, March 4, 2026 — prepared remarks and Q&A (company webcast via investors.broadcom.com).

Get the next brief, free.

We publish analyst-grade earnings briefs the same day or morning after every call — headline numbers, segment KPIs, Q&A highlights, and tone analysis. Free during beta.

This is not investment advice.