Broadcom — FQ2 FY2026 Earnings Preview
FQ2 FY2026 = Quarter ended ~May 3, 2026 ·
Reports Wednesday, June 3, 2026 AMC (5:00pm ET call) ·
Prepared May 24, 2026 · FYE late October · 10 days out · Stock $414.14
Earnings Date
June 3, 2026
AMC · 5pm ET call · 10 days out
Consensus Revenue
~$22.04B
Mgmt guide: ~$22B (+47% YoY)
Consensus EPS
~$2.40
+52% YoY (Q2 FY25 = $1.58)
Stock / 52-wk
$414.14 · +3% YTD
Closed May 22 · ~$2T market cap
Q2 AI Revenue Guide
$10.7B
+140% YoY · 6th customer (OpenAI) added
Q2 Semi Guide
$14.8B
+76% YoY · Non-AI $4.1B (+4%)
Q2 Infra SW Guide
$7.2B
+9% YoY · VMware ARR +19%
RPO (Q1 FY26)
$45.0B
+102% sequential — $11.7B QoQ jump
Executive Summary
Broadcom reports FQ2 FY2026 on Wednesday, June 3 AMC. This print
is the cleanest "is the AI capex super-cycle still accelerating?" data check of 2026.
Q1 FY26 was an inflection: total revenue accelerated to +29% YoY ($19.3B), AI semiconductor revenue
hit $8.4B (+106% YoY), and management guided Q2 to $22B (+47% YoY) with AI at $10.7B (+140% YoY) —
both step-functions higher than the prior trajectory. RPO jumped from $33.3B to $45.0B in a single
quarter (+$11.7B / +35% sequentially) — the largest backlog build in company history. Hock added a
sixth XPU customer (OpenAI deploying 1+ gigawatt in 2027) and reiterated the ">$100B AI revenue in
FY27" line of sight. The setup is for a sustained beat-and-raise into FY27.
Bull case: The $22B Q2 guide is mathematically a low-bar.
Q1 came in $55M above the mid-point ($19.31B vs ~$19.25B guide) but with AI at $8.4B vs initial
$7.4B guide — a +13% AI beat against guidance. If Q2 AI tracks similarly, AI lands at ~$12B vs
guide $10.7B, putting total revenue at ~$23.3B. The $45B RPO underwrites multi-year visibility:
even at the disclosed 33% recognized in next 12 months, that's ~$15B of contracted revenue locked
in. The 6 XPU customers (Google Ironwood TPU, Anthropic 3GW+, Meta MTIA, customers 4-5, OpenAI 1GW
in 2027) all ramp through 2027. AI networking (Tomahawk 6 / Jericho 4 / Tomahawk Ultra) continues
to win share at the scale-out fabric layer. VMware ARR +19% with $9.2B in Q1 bookings — Hock made
the explicit point that "AI creates need for MORE VMware, not less." Adjusted EBITDA margin of 68%
demonstrates operating leverage at scale. Capital return: $10B in incremental buyback authorization
on top of $10.9B already returned in Q1.
Bear case: Stock at $414 = ~$2T market cap at ~40x NTM EPS —
priced for sustained acceleration with little margin for any AI customer concentration or hyperscaler
capex air-pocket. Non-AI semi is stuck at ~$4.1B with only +4% growth guided (broadband/storage
recovering, wireless seasonal). AI is now 65%+ of semi (and ~50% of total) — a customer pause at
Google or Meta is a single-stock event. OpenAI deal is 2027, not immediate. Custom silicon competition
risk: NVDA Vera/Rubin ramp, AMD MI400, Marvell custom wins (e.g., Microsoft Cobalt successors).
The +$11.7B RPO jump came largely from new long-dated OpenAI/customer-6 commitments — revenue
recognition is back-end loaded. China export controls, taxes, and the +16.5% non-GAAP tax rate hike
all bleed margin. If Q2 misses or guide for Q3 falls short of consensus "next leg" ($25B+), the
multi-month +50% YTD rally re-rates.
What's at stake: Confirmation that (a) AI semi prints >$10.7B
guide (consensus $11B+), (b) Q3 guide is >$24B (implies sustained acceleration), (c) RPO holds
$40B+ (proves the Q1 jump wasn't a pull-forward), (d) more color on the OpenAI 2027 ramp,
(e) Anthropic 3GW timeline confirmed, (f) infra software ARR sustains +19%. A clean print plus
constructive Q3 guide likely takes the stock toward $450+; a soft AI guide or guide-down on Q3
could re-test the $370 50-day moving average.
Guidance & Consensus
Guidance issued on Q1 FY26 earnings call, March 4, 2026. AVGO guides Q+1 revenue + EBITDA margin only — no EPS guide. EPS consensus is bottoms-up from Street models.
| Metric | Low | Midpoint / Guide | High | Consensus | Prior Year | Note |
|---|---|---|---|---|---|---|
| Q2 FY26 — Total Revenue | — | ~$22.0B | — | $22.04B | $15.00B (Q2 FY25) | Mgmt: +47% YoY |
| Q2 FY26 — AI Semi Revenue | — | $10.7B | — | ~$11.0B | $4.4B (Q2 FY25) | Mgmt: +140% YoY |
| Q2 FY26 — Non-AI Semi Revenue | — | $4.1B | — | — | $4.0B (Q2 FY25) | Mgmt: +4% YoY |
| Q2 FY26 — Semi Total | — | $14.8B | — | ~$14.9B | $8.41B (Q2 FY25) | Mgmt: +76% YoY |
| Q2 FY26 — Infra Software Revenue | — | $7.2B | — | ~$7.2B | $6.60B (Q2 FY25) | Mgmt: +9% YoY; VMware ARR +19% |
| Q2 FY26 — Adj EBITDA Margin | — | ~68% | — | ~68% | 67% (Q2 FY25) | Operating leverage on AI scale |
| Q2 FY26 — Non-GAAP Tax Rate | — | ~16.5% | — | — | 14.0% | Up from global min tax + mix |
| Q2 FY26 — Non-GAAP Diluted EPS (Street) | — | — | — | $2.40 | $1.58 (Q2 FY25) | +52% YoY implied |
| FY27 — AI Revenue (Mgmt line-of-sight) | — | >$100B | — | TBD | ~$40B (FY26E) | Hock Q1 call: '>$100B AI revenue in '27' |
| FY27 — Total Revenue (Street model) | — | ~$118B | — | $118B | ~$94B (FY26E) | Implied +25%+ YoY; AI dominant |
Historical 8-Quarter Metrics — Daloopa Sourced
| Metric | Q2 FY24 | Q3 FY24 | Q4 FY24 | Q1 FY25 | Q2 FY25 | Q3 FY25 | Q4 FY25 | Q1 FY26 |
|---|---|---|---|---|---|---|---|---|
| Total Net Revenue ($M) | $12,487 | $13,072 | $14,054 | $14,916 | $15,004 | $15,952 | $18,015 | $19,311 |
| Total Revenue YoY % | +43% | +47% | +51% | +25% | +20% | +22% | +28% | +29% |
| Semiconductor Solutions ($M) | $7,202 | $7,274 | $8,230 | $8,212 | $8,408 | $9,166 | $11,072 | $12,515 |
| AI Semi Revenue ($B) | n/d | n/d | $3.7 | $4.1 | $4.4 | $5.2 | $6.5 | $8.4 |
| AI Revenue YoY % | n/d | n/d | +150% | +77% | +46% | +63% | +74% | +106% |
| Non-AI Semi Revenue ($B) | n/d | n/d | $4.5 | $4.1 | n/d | $4.0 | $4.6 | $4.1 |
| Semi Gross Margin % | 67% | 68% | 67% | 68% | 69% | 67% | 68% | 68% |
| Semi Operating Margin % | 55% | 56% | 56% | 57% | 57% | 57% | 59% | 60% |
| Infrastructure Software ($M) | $5,285 | $5,798 | $5,824 | $6,704 | $6,596 | $6,786 | $6,943 | $6,796 |
| Infra SW Gross Margin % | 88% | 90% | 91% | 92.5% | 93% | 93% | 93% | 93% |
| Infra SW Operating Margin % | 60% | 67% | 72% | 76% | 76% | 77% | 78% | 78% |
| Adj EBITDA % of Revenue | 61% | 64% | 65% | 68% | 67% | 67% | 68% | 68% |
| FCF % of Revenue | 36% | 37% | 39% | 40% | 43% | 44% | 41% | 41% |
| RPO Backlog ($B) | $22.9 | $21.2 | $20.5 | $22.4 | $24.7 | $27.5 | $33.3 | $45.0 |
| Non-GAAP Diluted EPS | $1.10 | $1.24 | $1.42 | $1.60 | $1.58 | $1.69 | $1.95 | $2.05 |
Revenue accelerating: +20% Q2 FY25 → +22% → +28% → +29% Q1 FY26. AI semi is the engine — went from $4.1B (Q1 FY25) to $8.4B (Q1 FY26), a 2x in 4 quarters. Q2 FY26 guide of $10.7B AI is +27% sequential. Operating leverage evident: Semi op margin 55% → 60% over 8Q, Infra SW op margin 60% → 78% post VMware integration. RPO jumped $11.7B QoQ in Q1 FY26 — the single-largest sequential build in company history, signaling multi-year XPU contracts locked in. EPS up nearly 2x (from $1.10 to $2.05) in 8 quarters. Cells link to Daloopa source pages.
Six XPU Customers — The Multi-Year AI Thesis
On the Q1 FY26 call (March 4, 2026) Hock Tan formally added OpenAI as the 6th custom XPU customer, joining Google, Meta, Anthropic, and two unnamed customers (consensus: ByteDance + Apple). Hock reiterated the multi-year line of sight of >$100B AI revenue in FY27. Watch June 3 for any updates on ramp timing, customer-by-customer gigawatts, and the disclosed customer count.
- 1. Google (Ironwood TPU v7) — Continued growth in CY26 with strong demand for the 7th-generation Ironwood TPU. Even stronger demand expected in 2027 and beyond from next-gen TPUs.
- 2. Meta (MTIA) — Hock pushed back on reports MTIA was discontinued. "Meta's custom accelerator MTIA roadmap is alive and well. We're shipping now." Next-gen MTIA scales to multiple gigawatts in 2027+.
- 3. Anthropic — Off to a "very good start" in 2026 for 1 gigawatt of TPU compute. In 2027, demand expected to "surge in excess of 3 gigawatts" of compute.
- 4 + 5. Two undisclosed customers (consensus: ByteDance + Apple) — Strong shipments this year; expected to "more than double in 2027."
- 6. OpenAI (NEW — added Q1 FY26 call) — Expected to deploy in volume their first-generation XPU in 2027 at over 1 gigawatt of compute. The "5 customers → 6 customers" disclosure is the single most important development of the quarter for the AI XPU narrative.
Total 2027 deployment math: Anthropic 3GW+ + Meta multiple GW + OpenAI 1GW+ + Google sustained TPU + customers 4-5 doubling = a multi-decagigawatt 2027. AVGO's stated position is that they have "fully secured capacity" for leading-edge wafers, HBM, and substrates through 2027.
RPO Backlog — The Lock-In Signal
The number: RPO grew from $33.3B (Q4 FY25) to $45.0B (Q1 FY26), a sequential jump of +$11.7B (+35%) — the largest single-quarter backlog build in Broadcom history. For context, Q4 FY25 itself was a $5.8B sequential build (already large). The acceleration in net new RPO suggests multi-year XPU and software contracts being formalized.
Recognition cadence: Per the 10-Q footnote, only 33% of RPO is expected to be recognized over the next 12 months (was 35% in Q4, 49% at the start of FY24). The mix is shifting toward longer-dated revenue — consistent with multi-year XPU supply agreements that Hock described as "multi-year supply agreements as our customers scale-up deployment."
Implied 12-month locked-in revenue: $45.0B × 33% = ~$14.9B already booked for the next 12 months. Layered on the $7.2B Q2 software guide and the $14.8B Q2 semi guide, this implies significant additional visibility into FY27.
What to watch on June 3: Does RPO hold above $40B (proves it wasn't a pull-forward)? Does the next-12-month recognition % go even lower (longer-dated contracts)? Any disclosure of which customer drove the Q1 build (OpenAI deal? Anthropic 3GW extension? VMware multi-year renewals?).
Management Tone Assessment — Post Q1 FY26 (Mar 4, 2026)
| Topic | Tone | Evidence (Q1 FY26 Call, Mar 4, 2026) |
|---|---|---|
| Overall framing | Confidently Bullish | Hock: 'top-line strength translated into exceptional profitability… our scale continues to drive significant operating leverage.' Most aggressive tone since VMware close. |
| AI Semiconductors | Strongly Bullish | Q1 AI rev $8.4B vs $7.4B guide ('way above outlook'). Q2 guide $10.7B (+140% YoY). 'Custom accelerator business grew 140% YoY in Q1. This momentum continues in Q2.' |
| Customer 6 (OpenAI) | Step-Function Bullish | First disclosure: 'We expect OpenAI deploying in volume their first-generation XPU in 2027 at over 1 gigawatt of compute capacity.' Largest customer addition since Anthropic. |
| FY27 AI Outlook | Highly Confident | Hock: 'line of sight that our revenue in '27 will be significantly in excess of $100 billion.' Reiterated multi-quarter framing — investor expectations now anchored. |
| Infrastructure Software / VMware | Stable Bullish | VMware revenue +13% YoY. ARR +19%. Q1 bookings $9.2B in TCV. Op margin 78% (+190bps YoY). Hock: 'AI creates need for MORE VMware, not less.' |
| Non-AI Semiconductors | Recovering / Stabilizing | Q1 $4.1B flat YoY. Enterprise networking, broadband, server storage all up YoY; wireless down seasonally. Q2 guide $4.1B (+4% YoY) — first positive YoY in 6 quarters. |
| Margin Trajectory | Bullish | Adj EBITDA 68% in Q1 (above 67% guide). Q2 guide 68% maintained. Semi op margin hit 60% (record). Hock: 'operating leverage' is the recurring framing. |
| Capital Return | Aggressively Bullish | Q1 buyback $7.8B + dividends $3.1B = $10.9B returned. Added $10B incremental authorization through CY26. Share count down ~23M in Q1 alone. |
| Tariffs / China / Policy | Quantified, Not Discussed | Light commentary on tariffs vs peers — implies limited direct impact on AI semis. Tax rate moves to 16.5% in Q2 from 14% (global min tax + mix). |
| Capacity / Supply Chain | Fully Secured | Hock: 'We have fully secured capacity of these components for '26 through '27 — wafers, HBM, substrates.' Reduces supply-side concern for the AI ramp narrative. |
Beat / Miss Track Record — 15 / 16 vs Street (last 8 quarters)
| Quarter | Rev Cons. | Rev Actual | Rev Beat | EPS Cons. | EPS Actual | EPS Beat |
|---|---|---|---|---|---|---|
| Q2 FY24 | $12.06B | $12.49B | +3.6% | $1.09 | $1.10 | +0.9% |
| Q3 FY24 | $12.98B | $13.07B | +0.7% | $1.22 | $1.24 | +1.6% |
| Q4 FY24 | $14.07B | $14.05B | -0.1% | $1.38 | $1.42 | +2.9% |
| Q1 FY25 | $14.62B | $14.92B | +2.1% | $1.51 | $1.60 | +6.0% |
| Q2 FY25 | $14.96B | $15.00B | +0.3% | $1.57 | $1.58 | +0.6% |
| Q3 FY25 | $15.83B | $15.95B | +0.8% | $1.66 | $1.69 | +1.8% |
| Q4 FY25 | $17.47B | $18.02B | +3.1% | $1.87 | $1.95 | +4.3% |
| Q1 FY26 | $19.26B | $19.31B | +0.3% | $2.03 | $2.05 | +1.0% |
Beat Rate (L8Q)
15 / 16 — 94%
EPS 8/8; Revenue 7/8 (Q4 FY24 razor-thin)
Avg Rev Beat L4Q
+1.1%
Q4 FY25 was largest at +3.1%
Avg EPS Beat L4Q
+1.9%
Range +0.6% to +4.3%
Key takeaway: AVGO is a consistent beater. The Q4 FY25 EPS beat of +4.3% (the largest in the
window) corresponded with the largest AI semi beat — Q1 FY26 mirrors that pattern but with even greater AI
beat magnitude (Q1 AI rev $8.4B vs $7.4B guide = +13%). Repeat of that pattern on the Q2 print would put EPS
near $2.45-$2.50 vs $2.40 consensus. Revenue print near $22.3-$22.5B (vs $22.0B guide / $22.04B consensus)
likely on AI overdelivery alone.
Key Catalysts — Bull vs Bear
Bull Catalysts
- Q2 AI revenue prints >$11.5B (vs $10.7B guide) — sustains the +13% beat pattern
- Q3 revenue guide >$24B — implies continued sequential acceleration into back half
- RPO holds above $42B — proves Q1 jump wasn't a pull-forward
- OpenAI 2027 timing pulled earlier (late 2026 production ramp)
- Anthropic 3GW timeline confirmed for 2027 with quarterly cadence
- Meta MTIA next-gen volume targets disclosed
- VMware ARR sustains +19%; total contract value >$10B in Q2
- Non-AI semi guidance up >+6% YoY (broader cyclical recovery confirmed)
- Adj EBITDA margin guides to >68% for Q3 (operating leverage continues)
- Additional buyback authorization or accelerated repurchase
- Tomahawk 6/7 or Jericho 4 milestone (AI networking share takes)
- Disclosure of 7th XPU customer or expansion of existing customer commitment
Bear Risks
- Q2 AI revenue light vs $10.7B guide — first AI miss in 8 quarters
- Q3 guide misses Street ($22B vs $24B expected) — deceleration signal
- Non-AI semi guides flat-to-down — broader cyclical concerns
- RPO drops back below $40B — Q1 jump was a one-time pull-forward
- Google Ironwood TPU ramp delay (process-node or design issue)
- OpenAI deal terms light on $$ — 1GW spread over multi-year
- Tax rate steps up further (above 16.5%) on additional global min tax application
- Capacity comments turn more cautious (HBM, substrates, advanced packaging)
- Custom silicon competitive pressure: MRVL wins displace AVGO at customer 4 or 5
- VMware churn or pricing fatigue surfaces — bookings decelerate from $9.2B Q1
- Hock makes any comments about 'lumpy' or 'pause' in AI semi (would re-rate the stock)
- China export controls tighten further; impacts non-AI semi mix
Peer Earnings — Read-Throughs into the Print
| Peer | Earnings Date | Read-Across |
|---|---|---|
| NVDA (NVIDIA) | Reported May 27 (Q1 FY27) | Single most important read for AI semi. Data center growth + capex commentary from Jensen sets the bar for AI demand visibility. |
| TSM (Taiwan Semi) | Reported Apr (Q1 CY26) + Monthly sales | AVGO's foundry. AI accelerator wafer demand color, advanced packaging capacity (CoWoS-S/L), HBM cadence. |
| MRVL (Marvell) | Reported May 28 (Q1 FY27) | Direct AI custom silicon competitor (Microsoft, Amazon). Custom silicon momentum sets backdrop for AVGO XPU narrative. |
| ARM (ARM Holdings) | Reported May 7 (Q4 FY26) | IP royalties from AVGO XPU customers (Anthropic Trainium/TPU, Meta MTIA all ARM-based). |
| AMD (Advanced Micro) | Reported May 6 (Q1 CY26) | MI400 series, EPYC server CPUs. Read on hyperscaler GPU vs custom silicon mix. |
| META (Meta Platforms) | Reported Apr 30 (Q1 CY26) | Capex commentary on AI infrastructure spend. MTIA roadmap and any custom silicon disclosure. |
| GOOGL (Alphabet) | Reported Apr 22 (Q1 CY26) | Capex on TPU/AI infra. AVGO's largest XPU customer — TPU v7 'Ironwood' ramp commentary. |
| MSFT (Microsoft) | Reported Apr 24 (Q3 FY26) | Azure AI capacity, Cobalt CPU progress (potential MRVL/AVGO swing). $80B+ FY26 capex run-rate. |
| NET (Cloudflare) | Reported May 1 (Q1 CY26) | Edge AI inference demand pull-through. Read on network bandwidth growth from AI training. |
What to Watch on June 3 (AMC)
1. Q2 AI semi revenue vs $10.7B guide:
+13% beat pattern from Q1 implies ~$12.1B. Anything >$11B is constructive; <$10.7B would be the first AI miss in 6 quarters.
2. Q3 FY26 revenue guide (the headline):
Street consensus implies ~$24B+. Mgmt typically guides Q+1 only. A guide above $23.5B confirms continued acceleration.
3. Q3 AI semi guide:
Implied range $11.5B-$13B+. A guide above $12.5B is bullish; below $11.5B raises concern that customer-2 / customer-4 ramp paused.
4. RPO trajectory:
Q1 was $45B (+$11.7B QoQ). Hold above $42B = locked in; below $40B = pull-forward.
5. OpenAI ramp specifics:
Any earlier-than-2027 timing? Disclosure of $$ commitment or gigawatt schedule. Confirms or de-risks the $100B AI FY27 line.
6. Anthropic 3GW timeline:
Hock said '2027.' Watch for quarterly cadence: ramp Q1 → Q4 2027, or front-loaded?
7. Meta MTIA volume:
Hock pushed back on cancellation reports. Watch for explicit shipping schedule or gigawatt target.
8. Customer 7 disclosure?:
Adding OpenAI as #6 in Q1 was massive. Any reference to additional customer engagements (e.g., xAI, ByteDance follow-on)?
9. Non-AI semi outlook:
Q2 guide $4.1B (+4% YoY) — first positive in 6Q. Sustained recovery? Wireless seasonal commentary?
10. VMware metrics:
ARR was +19% Q1; bookings $9.2B TCV. Net retention, deal sizes, hyperscaler partnerships.
11. Capacity / HBM / substrate:
Q1: 'fully secured through 2027.' Any signs of constraint or relief? CoWoS-L allocation?
12. Adj EBITDA margin trajectory:
Q2 guide 68%. Watch for Q3 commentary — operating leverage continues or hits R&D investment wall?
13. Tax rate sustainability:
Q2 guides to 16.5% from 14%. Is this the new normal? Implication for FY27 EPS.
14. Capital return:
$10B incremental authorization added Q1. Watch for additional authorization or pace of repurchase.
15. FY27 commentary:
Any quantification beyond Hock's '>$100B AI' line of sight? Q3 guide will imply FY27 starting point.
Source: Daloopa (8-quarter historical data — segment revenue Semi Solutions / Infra Software, AI semi revenue, non-AI semi revenue, segment gross/operating margins, EBITDA % of revenue, FCF % of revenue, RPO backlog; cells link to Daloopa source pages, company_id=22). Additional sources: Broadcom Q1 FY26 earnings call transcript (March 4, 2026, via Financial Modeling Prep API), Q4 / Q3 FY25 transcripts, Broadcom Q1 FY26 press release & 10-Q, FMP earnings history (consensus + actuals last 8 quarters), Daloopa stock price ($414.14 close May 22, 2026). EPS values are non-GAAP diluted (mgmt-reported).