Broadcom’s $100B AI Bombshell: Is AVGO About to Smash Through $350?

Broadcom (AVGO) delivered the kind of quarter that reminds investors why it sits near the center of the AI hardware trade: modest upside versus consensus in fiscal Q1, a meaningfully better-than-expected guide for Q2, and unusually confident multi-year demand commentary. Shares are up about 7% in early trade as the report lands at a moment when semis have been trying to reassert leadership, even as broader equities chop around on geopolitics and rate-cut timing. In the tape, this is the “good AI” setup: results weren’t just fine, the guidance and the narrative tightened the line-of-sight story in a way that tends to get rewarded.

On the numbers , Broadcom posted adjusted EPS of $2.05 versus the Street at $2.03, with revenue of $19.31B also slightly above expectations. The quality of the beat mattered more than the size: EBITDA margin came in at 68%, roughly 125bp better than anticipated, which helped offset the reality that revenue was only modestly ahead. Segmentally, Semiconductor Solutions revenue was $12.52B versus roughly $12.25B expected, while Infrastructure Software was $6.79B–$6.80B, a bit light versus consensus (roughly mid-$6.8B to ~$7.0B depending on the source). In other words, the quarter did what bulls needed: AI and semis carried, software didn’t derail, and profitability held up.

Guidance is what pushed this from “solid” to “market-moving.” Broadcom guided fiscal Q2 revenue to about $22.0B versus the Street around $20.5B–$20.6B, and kept adjusted EBITDA margin at 68% versus consensus closer to ~66%–67%. The semiconductor guide was particularly loud: ~$14.8B in Q2 semiconductor revenue, and AI semiconductor revenue of $10.7B (well above consensus), implying AI growth accelerating sharply year over year. Layer on top a new $10B buyback authorization through December 2026 and a steady dividend, and you have an earnings package that speaks both to growth and shareholder returns.

The AI discussion did most of the heavy lifting, and it wasn’t generic “we love AI” filler. Management said Q1 AI revenue rose 106% year over year to $8.4B, driven by custom AI accelerators and AI networking, and then raised the bar with “line of sight” to AI chip revenue (chips only) exceeding $100B in 2027. That “line of sight” phrase matters because it implies committed programs and a clearer demand funnel, not just a hopeful extrapolation. Multiple sell-side teams highlighted improved visibility, the durability of engagements, and management’s effort to refute the two big fears in custom silicon: customers insourcing and margin dilution.

On ASICs (Broadcom’s custom accelerators / XPUs), the key message was customer breadth and multi-year capacity. Management described the ramp of custom AI accelerators across its customers as progressing well and said it has fully secured supply chain capacity for key components through 2026–2028. That’s effectively backlog-by-another-name: you may not get a single neat backlog line item, but when a vendor says they’ve secured multi-year component capacity against defined hyperscaler programs, that is a strong signal of forward demand and execution confidence. The Street also latched onto the “sixth customer” disclosure (OpenAI) and reaffirmation of Meta’s roadmap, both of which help reduce concentration anxiety and the constant “is this program real?” noise.

TPU commentary was equally specific and, importantly, tied to scale. For Google, Broadcom called out strong demand for the seventh-generation Ironwood TPU and implied increasing demand into 2027 with next generations. It also spoke to Anthropic demand in TPU compute terms (one gigawatt in 2026, potentially surging beyond three gigawatts in 2027), and said OpenAI is expected to deploy its first-generation XPU in 2027 at over one gigawatt of compute capacity. Those “gigawatt” references aren’t just colorful language; they frame the magnitude of deployments and reinforce that this is infrastructure-scale, not pilot-scale.

AI networking was another underappreciated lever that can keep Broadcom’s AI mix healthier. Management suggested AI networking could be a large share of AI revenue over time, and the call’s tone (plus the guide) points to networking content rising as clusters scale and inference expands. If inference is indeed the explosive growth driver (as management emphasized), that tends to pull through accelerators, networking, and system-level content more broadly.

On valuation, Broadcom is not “cheap,” but the multiple is easier to defend when guidance is stepping up. In the material you provided, Broadcom was cited around ~27x forward earnings, roughly in the neighborhood of AMD and richer than Nvidia on that specific snapshot. The debate for investors is whether Broadcom deserves a premium as a more diversified AI infrastructure compounder (custom silicon + networking + VMware cash engine), or whether the market should keep it closer to a historical band given software questions and macro uncertainty. The company’s margin stability commentary (including reassurance around rack economics and AI mix) is aimed directly at that valuation question.

So, does this report have the strength to push through $350 resistance? Fundamentally, yes: you have upside guide, accelerating AI revenue, clearer multi-year visibility, and supportive capital return. Technically, a 7% gap toward a well-watched level often invites profit-taking and “sell the news” games, but this is the type of print that can sustain a breakout if the broader tape cooperates and semis don’t roll over. The cleanest bullish path is follow-through above $350 on real volume, with the stock holding the level on any intraday retest; the main risk is that macro headlines (rates/oil/geopolitics) swamp a single-name catalyst and turn a strong report into a one-day wonder. This time, Broadcom gave bulls enough substance to argue it should be more than that.