Broadcom’s $100B AI target fails to satisfy Wall Street’s high bar

The selloff signals that 'priced for perfection' valuations in the AI infrastructure stack are highly vulnerable to even minor guidance misses. For the streaming industry, this volatility directly impacts the cost and availability of custom ASICs and networking silicon required for high-efficiency video encoding and personalized recommendation engines. As Broadcom and Marvell dominate the custom chip market for hyperscalers like Google and Meta, any pivot or caution in their roadmaps could slow the transition from general-purpose GPUs to more cost-effective, custom streaming hardware. Watch for whether hyperscalers accelerate internal chip development to bypass the premium pricing and supply constraints of merchant silicon providers.

The volatility in semiconductor markets comes as hyperscalers face accelerating infrastructure costs. Per the Financial Times in May 2026, the combined capital expenditure of Google, Amazon, Microsoft, and Meta is projected to reach $725 billion this year, a 77% increase over 2025. This spend is increasingly consumed by a 'super-cycle' in memory pricing. BNN Bloomberg reported in June 2026 that memory chip prices have spiked six-fold over the past year, as manufacturers prioritize high-margin AI data center chips over components for consumer electronics and traditional streaming devices.

These rising costs are starting to reach the consumer layer. Research from IDC in early 2026 suggests that consumer electronics prices could rise by up to 20% due to persistent component shortages and the prioritization of AI-specific silicon. For streaming services, this means the hardware used for edge caching and content delivery is becoming more expensive just as the market shifts from training AI models to the more compute-intensive inference phase. According to semi-analysis data from June 2026, inference now accounts for roughly two-thirds of all AI compute demand.

Competitive dynamics are also shifting as major players secure limited supply. Tom's Hardware reported in May 2026 that Nvidia has effectively booked over half of TSMC's advanced packaging capacity for the year, leaving Broadcom, AMD, and Google's TPU program to compete for the remaining volume. This bottleneck ensures that even as Broadcom maintains a high-revenue target, its ability to scale production remains tied to physical manufacturing constraints that are unlikely to ease before 2027.