Broadcom's $100B AI Forecast, NVDA Margins, SpaceX IPO: Mar 8

Broadcom’s CEO put a bold number on the table this week, Nvidia’s latest quarter showed margins that most hardware companies cannot dream of, and SpaceX’s IPO is moving from rumor to process. Here is what matters for AI investors right now.

Key points

• Broadcom CEO Hock Tan said AI chip revenue will come in “significantly” above $100 billion in 2027, after AI revenue more than doubled in the most recent quarter.
• Nvidia posted Q4 revenue growth of 73% year-over-year with gross margins of 75%, reinforcing its premium positioning in AI data center hardware.
• SpaceX added Citigroup to its IPO bank lineup and is targeting a confidential filing as early as March, with a reported valuation target of $1.75 trillion and a $50 billion raise.
• TSMC is targeting 2028 for completion of a new Tainan fabrication plant to meet AI chip demand.
• Alphabet CEO Sundar Pichai was awarded a pay package worth up to $692 million over three years, underscoring how much Big Tech boards are paying to retain AI-era leadership.

Is Broadcom becoming the second pillar of AI chip revenue?

The $100 billion figure from Broadcom deserves context. Hock Tan’s guidance is specifically tied to custom silicon, the application-specific chips that hyperscalers like Google, Meta, and others are commissioning to reduce dependence on Nvidia’s off-the-shelf GPUs. Broadcom sits at the center of that trend as a leading custom chip designer and supplier.

Demand from Big Tech companies investing heavily in AI infrastructure is the primary driver cited for that projection. If the forecast holds, Broadcom’s AI segment alone would rival the total annual revenue of most semiconductor companies. The stock gained on the news, according to reporting from Investing.com.

For investors, this is a different kind of AI chip bet than Nvidia. Broadcom’s custom chip business is inherently tied to long-term design wins with specific customers, which creates stickiness but also concentration risk. The two companies are not direct competitors so much as they serve different parts of the same AI infrastructure buildout.

What do Nvidia’s margins say about AI infrastructure pricing power?

Nvidia’s Q4 gross margin of 75% on revenue that grew 73% year-over-year is the kind of combination that rarely lasts in hardware. The analysis from Seeking Alpha frames this as a “generational lead premium,” meaning the market is pricing in continued dominance rather than just current performance.

The critical question for investors is durability. Broadcom’s custom chip momentum, AMD’s ongoing GPU push, and internal chip programs at Google and Amazon all represent pressure on Nvidia’s pricing power over time. For now, those 75% margins suggest customers are still paying whatever Nvidia asks. That dynamic will eventually shift, but the timeline is uncertain.

TSMC’s planned Tainan expansion, targeting completion in 2028, is partly a capacity response to Nvidia and Broadcom’s volume requirements. More fab capacity in the 2028 timeframe could ease supply constraints and, in theory, reduce some of the scarcity premium baked into current AI chip pricing.

SpaceX IPO: what investors should watch

SpaceX is not an AI pure-play, but its IPO has relevance for AI investors in a few ways. The company’s Starlink satellite network is increasingly being positioned as AI-adjacent infrastructure, and a $1.75 trillion valuation would make this one of the largest public offerings in history, pulling significant capital from the broader tech and growth investment pool.

Citigroup joining the bank lineup alongside earlier reported banks signals the deal is becoming more concrete. A confidential filing as early as March would mean a public listing could come within months. A $50 billion raise at that scale would be a significant liquidity event for the private markets.

Separately, Sundar Pichai’s three-year pay package of up to $692 million reflects a broader reality: the competition for executives seen as capable of navigating the AI transition is intense, and boards at Alphabet and its peers are paying accordingly. For shareholders, that is a cost worth monitoring as AI investment cycles stretch longer and capital allocation decisions grow more consequential.

Nothing on this site constitutes investment advice. This update is for informational purposes only.