Broadcom CEO Warns AI Is a Casino, Targets $230B Chip Sales
AiX — Free Expert Advisor
Trades XAUUSD on autopilot. Verified Myfxbook performance. Free forever.
Risk warning: CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage. The majority of retail investor accounts lose money when trading CFDs. AiX is informational software — not investment advice. Past performance does not guarantee future results.
Broadcom Chief Executive Hock Tan told an audience in Singapore on Monday that many investors buying into artificial intelligence have "no clue" what they own, comparing the boom to casino gambling, Bloomberg reported. The same executive forecasts Broadcom's annual AI chip revenue rising roughly fourfold from about $58 billion this fiscal year to $230 billion in fiscal 2028. AI semiconductor revenue more than tripled year on year to around $17 billion in the fiscal third quarter, with about $22 billion expected in the fourth.
Context — why the casino warning lands harder from Broadcom
Tan's position makes the warning unusual. Broadcom designs custom AI chips and data centre networking equipment for customers including OpenAI and Anthropic, two of the largest buyers of computing power in the market. Few firms have captured more of the AI investment wave, which means few executives are better placed to describe what they see inside the order book.
His scepticism reached the buyers themselves. Tan said OpenAI's annualised revenue of around $50 billion in September showed strong demand, but argued that its enormous training costs give little sign of long-term profitability. Broadcom's growth plan leans on OpenAI as a customer, so the CEO is publicly questioning the economics of a company his own forecast depends on.
He also questioned whether businesses are getting real value from the technology, likening companies experimenting with AI to children who burn through arcade tokens quickly without learning anything. That framing shifts the debate from whether compute demand exists to whether the spending produces returns.
The catalyst for the remarks was a public appearance rather than a corporate disclosure. Broadcom did not publish new guidance alongside them, and the company did not disclose the terms of its financing arrangements. The numbers Tan cited for AI chip revenue stand as his own forward view, not audited results.
What changed is the funding question. Tan said access to capital, rather than shortages of memory chips, could become the industry's biggest constraint. That moves the binding limit from silicon supply to credit availability, a different problem with a different set of watchers.
Data — the numbers behind the shovel-seller's warning
Broadcom's AI semiconductor revenue more than tripled from a year earlier to around $17 billion in the fiscal third quarter. The company expects about $22 billion in the fourth quarter, a sequential step up of roughly $5 billion.
The longer arc is steeper. Tan forecasts annual AI chip revenue rising roughly fourfold from about $58 billion this fiscal year to $230 billion in fiscal 2028.
| Metric | Figure |
|---|---|
| Q3 AI semiconductor revenue | ~$17 billion |
| Q4 AI semiconductor revenue (guided) | ~$22 billion |
| FY2026 AI chip revenue | ~$58 billion |
| FY2028 AI chip revenue (target) | ~$230 billion |
OpenAI's annualised revenue stood at around $50 billion in September, per Tan's remarks. That figure matters because it sits below Broadcom's single-year AI chip target for fiscal 2028, meaning the customer base would need to scale its own top line substantially to absorb the chips Broadcom plans to ship.
On the financing side, banks last week began syndicating a $60 billion AI financing package tied to Anthropic's lease of Google-designed chips. The structure includes $42 billion of senior loans supported by Broadcom's credit rating. The Financial Times reported the arrangement.
That $42 billion sits against roughly $58 billion in expected AI chip revenue this fiscal year, a reminder that Broadcom's balance sheet now backs a meaningful share of the demand it forecasts.
Analysis — where the credit risk actually sits
The Anthropic-linked package converts a chip order into a financing exposure. Broadcom's credit rating supports $42 billion of senior loans, so if the lessee struggles to meet lease commitments, the strain reaches Broadcom's own credit profile rather than stopping at the customer.
Lessor structures of this kind spread through several layers. The chip designer carries contingent exposure, the banks syndicating the debt carry the paper, and the model developer carries the lease obligation. Each layer prices the risk of the layer below it.
Semiconductor peers with heavy data centre exposure face a read-across, though the report gives no figures for them. The relevant question for the sector is not whether orders exist but whether the buyers can fund them without vendor support.
A counter-argument deserves weight. Tan's $230 billion target implies he expects the demand to materialise despite his own scepticism, and tripling AI revenue in a single quarter is evidence that customers are paying today. Warnings about long-term profitability do not invalidate current cash flows.
The limitation is disclosure. Broadcom has not detailed how much of its forward revenue depends on vendor-backed financing, so investors cannot size the contingent exposure from public statements alone.
Positioning reflects the tension. Investors holding semiconductor exposure for AI demand have reason to stay, while credit-focused desks now have a new instrument to analyse in the $42 billion Broadcom-backed tranche.
Outlook — what to watch next
Broadcom's next results are the first checkpoint. Investors will look for any sign of how financing exposure is evolving and whether customer commitments remain on track, particularly the OpenAI-linked portion of the forward book.
The syndication of the $60 billion Anthropic package is the second. How the $42 billion senior tranche prices will indicate how credit markets judge Broadcom-backed AI paper, and whether the structure attracts demand at tighter spreads than a standalone developer could command.
Any update on memory chip supply is a third, because Tan has argued capital access may replace component shortages as the binding constraint. If he is right, headlines about chip supply matter less than headlines about debt pricing.
No specific dates for these catalysts appear in the report, and none should be assumed. The conditionals are simple: stronger syndication demand supports the vendor-financing model, while weak uptake would validate Tan's own warning about funding as the limit.
Frequently Asked Questions
What does Hock Tan's casino comment mean for Broadcom shareholders?
It means the CEO is flagging that AI demand may be less durable than order books suggest, while his own $230 billion fiscal 2028 target assumes the opposite. Shareholders face a company whose growth depends on customers whose profitability its own chief executive has questioned, plus $42 billion of senior loans supported by Broadcom's credit rating.
Why is vendor financing a risk for chip companies?
When a chipmaker's credit rating backs loans that let customers buy its products, the sale and the credit exposure arrive together. If the customer cannot meet lease or repayment commitments, the loss reaches the vendor's balance sheet. Broadcom has not disclosed how much of its forward revenue depends on such arrangements.
How does OpenAI's revenue compare with Broadcom's chip targets?
Tan said OpenAI's annualised revenue was around $50 billion in September, while Broadcom targets about $230 billion in AI chip revenue by fiscal 2028. The gap implies the customer base must grow its own top line several times over to absorb the chips Broadcom plans to ship, which is why Tan's profitability doubts matter for his own forecast.
Bottom Line
Broadcom is selling the picks, lending the money, and warning that the prospectors may not find gold.
Disclaimer: This article is for informational purposes only and does not constitute investment advice. CFD trading carries high risk of capital loss.
Trade XAUUSD on autopilot — free Expert Advisor
AiX is our free MetaTrader 5 Expert Advisor. Verified Myfxbook performance. No subscription. No fees. XAUUSD breakout engine.
PartnerPosition yourself for the macro moves discussed above
Start TradingSponsored
Ready to trade the markets?
Open a demo account in 30 seconds. No deposit required.
CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage. You should consider whether you understand how CFDs work and whether you can afford to take the high risk of losing your money.