Broadcom Q2 FY2026 Earnings: AI Infrastructure Does Not End With GPUs
Broadcom’s Q2 FY2026 results are more than a strong earnings print. They show that AI infrastructure is expanding beyond GPUs into custom ASICs, AI networking, and data-center optimization.
Bottom line: Broadcom is a structural AI infrastructure beneficiary. But a great company and a great entry price are not the same thing. Q2 strengthens the growth thesis, while new entries still require price and liquidity discipline.
The Q2 numbers were strong
Broadcom reported Q2 FY2026 revenue of $22.187 billion, up 48% year over year. GAAP net income was $9.310 billion, while non-GAAP net income was $12.074 billion. GAAP diluted EPS was $1.91 and non-GAAP diluted EPS was $2.44.
The quality of profitability matters even more. Adjusted EBITDA reached $15.244 billion, roughly 69% of revenue, while free cash flow was $10.262 billion, about 46% of revenue. Few AI-linked growth companies combine this level of growth with this level of cash generation.
An AI growth story with serious free cash flow
Broadcom’s appeal is not just revenue growth. It is growth plus margin, asset efficiency, and shareholder-return capacity.
AI semiconductors are becoming Broadcom’s center of gravity
| Metric | Result | Meaning |
|---|---|---|
| Q2 revenue | $22.187B | 48% year-over-year growth |
| Q2 AI semiconductor revenue | $10.8B | 143% year-over-year growth |
| Q2 AI revenue share | About 49% of total revenue | AI is no longer a side business |
| Q3 AI semiconductor outlook | $16.0B | More than 200% year-over-year growth expected |
| Q3 total revenue outlook | $29.4B | 84% year-over-year growth expected |
Custom ASICs and AI networking are the core
Semiconductor solutions revenue was $15.009 billion, up 79% year over year. Broadcom attributed AI semiconductor growth to demand for custom AI accelerators and AI networking.
This matters because AI infrastructure does not end with GPUs. At scale, data movement, latency, connectivity, and power efficiency become bottlenecks. Broadcom is positioned in that connectivity layer.
VMware is less the growth engine and more the cash-flow stabilizer
Infrastructure software revenue was $7.178 billion, up 9% year over year. That growth rate is less dramatic than AI semiconductors, but the segment helps stabilize Broadcom’s margin and free cash flow.
Broadcom is combining high-growth AI silicon on one side with recurring software cash flow on the other. That mix supports a different valuation discussion than a pure semiconductor-cycle stock.
AI infrastructure is not a one-GPU story
GPUs
The center of general-purpose training and inference, and the layer where NVIDIA remains strongest.
Custom ASICs
The layer where large customers optimize repeated workloads for lower cost and lower power.
AI Networking
The layer that reduces data-movement and connectivity bottlenecks inside large AI clusters.
Custom ASIC demand is a result of AI cost pressure
As AI investment scales, hyperscalers need more compute while also reducing the unit cost of that compute. GPUs provide flexibility and ecosystem strength, but custom chips can improve efficiency for specific large-scale workloads.
That is where Broadcom benefits. The next phase of AI infrastructure competition may be less about who buys the most chips and more about who produces the most useful AI service at the lowest cost per unit.
Growth is strong, but liquidity prices the multiple
From a growth perspective, Broadcom’s Q2 was extremely strong. AI semiconductor revenue accelerated, Q3 guidance became stronger, and margins and free cash flow remained impressive.
But asset prices do not move on growth alone. Rates, inflation, oil, the dollar, and risk appetite can compress multiples even for high-quality AI names. Broadcom’s business quality and entry timing should therefore be separated.
What to monitor after this print
- Whether AI semiconductor revenue actually reaches the Q3 target of $16.0 billion.
- How concentrated and repeatable the custom ASIC revenue base is.
- Whether AI networking demand is structural rather than a one-time buildout effect.
- Whether VMware and infrastructure software continue to stabilize high-margin cash flow.
- Whether rates and liquidity conditions support the valuation of AI infrastructure stocks.
The main risk is not weak demand; it is expectations and concentration
The demand signal is clearly strong. The risks sit elsewhere. Custom ASIC revenue can be tied to a small number of large customer projects, making timing and order concentration important.
Broadcom has also already been re-rated as a core AI infrastructure name. When expectations are this high, even excellent results do not always translate into immediate share-price upside.
Do not confuse a great company with a great entry
- Do not turn strong earnings into an automatic buy at any price.
- Do not use the word AI to justify every valuation premium.
- Watch customer concentration and project lumpiness alongside the AI growth rate.
- Remember that liquidity can hit multiples before the long-term thesis changes.
Final view: Broadcom belongs on the core list, but not at any price
Broadcom’s Q2 FY2026 results show the direction of AI infrastructure. AI investment began with GPUs, but it does not end there. As data centers scale, custom chips, networking, power efficiency, and inference-cost optimization become increasingly important.
Broadcom sits near the center of that shift. The print supports the idea that the company is not merely a semiconductor supplier, but an AI infrastructure platform that helps solve cost and connectivity bottlenecks.
The key discipline is simple: a great company is not always a great entry. Broadcom’s growth story is stronger after this quarter, but if the market has already priced in too much of that story, new entries should still be staged rather than chased.
Korean version: Read the Korean version
This is public market interpretation based on public sources, not a buy or sell recommendation.
Public sources checked
This article is anchored on Broadcom’s official Q2 FY2026 earnings release, supplemented by the investor center and major public financial sources. Market prices and reactions can vary by provider and timestamp, so the article focuses on the structural meaning of the results and guidance.