The news
Broadcom predicted a surge in artificial intelligence chip sales over the next two years. The forecast renews optimism that the company can challenge Nvidia’s dominance in the market for high-value accelerators. Broadcom also stated it remains on track to exceed $30 in earnings per share for fiscal 2028, a figure that outpaces current Wall Street estimates. Bloomberg Intelligence analyst Matt Bloxham reviewed the projections in detail.
Context
Nvidia currently leads the segment for accelerators used in large-scale AI training and inference workloads. Broadcom has positioned its custom silicon offerings as alternatives for hyperscale operators building out data-center capacity. The updated outlook arrives while cloud providers continue to place orders for hardware that supports expanding model-training clusters. Prior company guidance had already pointed to growth in the custom-chip business, but the two-year acceleration marks a sharper upward revision.
Details
The revenue outlook is tied directly to demand for Broadcom’s custom AI accelerators. Matt Bloxham noted that the projected growth trajectory over the next 24 months underpins the higher earnings-per-share target for fiscal 2028. No additional product road maps, specific revenue figures, or customer names were disclosed alongside the forecast. The company framed the increase as a direct result of sustained orders rather than new market entries or unannounced designs.
Why it matters
Engineers and infrastructure teams evaluating hardware choices now receive a clearer signal that at least one additional supplier plans to scale production capacity. Sustained competition in the accelerator market can influence lead times, pricing negotiations, and the availability of reference designs for new clusters. The earnings target further indicates that Broadcom views its custom-silicon business as a durable profit center rather than a short-term experiment. Teams planning multi-year infrastructure road maps can therefore treat Broadcom as a more credible second source alongside Nvidia when they model future capacity purchases.
The forecast does not address relative performance per watt or software-ecosystem maturity between the two suppliers. It does, however, reduce the probability that the market remains a single-vendor affair for the immediate planning horizon. Procurement groups at large cloud operators often maintain dual-source strategies precisely to mitigate supply risk; the updated Broadcom numbers give those groups additional data points for capacity modeling. Financial analysts tracking semiconductor margins will also watch whether the projected sales growth translates into sustained operating leverage or whether it requires heavier research-and-development spending to keep pace with evolving customer requirements.
For developers building inference pipelines or training frameworks, the existence of a second credible hardware path matters because it affects long-term portability decisions. Code optimized for one vendor’s interconnect or memory hierarchy can carry migration costs when clusters are refreshed. A stronger Broadcom presence may encourage more investment in abstraction layers that reduce those costs. At the same time, the earnings guidance suggests Broadcom expects pricing power to remain intact even as volumes rise, which could limit how much downward pressure appears on accelerator costs in the near term.
Infrastructure planners should therefore incorporate the revised Broadcom trajectory into their own three-to-five-year forecasts rather than treating the company as a niche or experimental option. The two-year sales surge and the $30 earnings target together point to a market that is expanding fast enough to support multiple suppliers at scale, provided execution on both the technical and manufacturing sides continues without major disruption.
---
Sources:
{"word_count": 612, "sources_used": 1, "headline": "Broadcom Forecasts Surge in AI Chip Demand"}
No comments yet