Broadcom CEO Hock Tan and Intel CEO Lip-Bu Tan warned at a Singapore conference that access to capital could emerge as a major constraint on the AI boom4. The same day, Tencent told fixed-income investors that its planned bond offering of about $5 billion would fund AI expenditures, with the bulk going toward chip purchases2. ANALYSIS Taken together, the moves mark a structural pivot: the AI boom's funding base is migrating from equity risk capital toward fixed-income markets, even as the investors supplying that capital openly doubt the boom's durability.
Why it matters
The AI infrastructure race has been framed for two years as a supply-side problem: not enough GPUs, not enough power, not enough fab capacity. The warnings from Broadcom and Intel reframe it as a demand-side financing problem. "[What] I think might be the biggest constraint is capital," Hock Tan said at the DBS Asia Future Forward Conference. Broadcom's CEO argued that financing could become a bigger obstacle to AI expansion than shortages of semiconductors, memory chips, or electricity. Intel's Lip-Bu Tan reinforced the point: "If there's any challenge in the front, it's capital and the cost of capital, and then the other part is in the longer run, people want to see the ROI". ◆ When the two companies that sell the building blocks of AI infrastructure identify the buyer's ability to pay as the binding constraint, the competitive landscape tilts toward firms with the cheapest access to debt markets.
The big picture
Tencent's bond offering illustrates how that tilt plays out in practice. The company told fixed-income investors that its planned offering of about $5 billion would fund AI expenditures, with the bulk going toward chip purchases. The deal includes $1.5 billion to $2 billion in dollar tranches. ◆ A bond sale of that scale, directed explicitly at chip procurement, treats AI compute not as a speculative venture bet but as a capital-intensive infrastructure project financed the way utilities or telecom networks have long been financed: with debt.
Meanwhile, the capital that is flowing into AI carries a paradox. A survey of 155 North American family offices conducted by RBC Wealth Management and Campden Wealth found that 85% ranked artificial intelligence as their top investment theme3. Yet 75% of respondents expected an AI bubble to burst within the next five years. RBC's Bill Ringham, director of private wealth strategies, offered a blunt diagnosis: "Even patient capital can be caught up in the fear of missing out".
ANALYSIS The family-office data and the CEO warnings describe two sides of the same tension. Capital is available, but it is arriving with a hedge attached. Investors are deploying into AI while simultaneously pricing in a correction, a posture that favors instruments with downside protection over pure equity exposure. That dynamic helps explain why a company like Tencent would tap bond markets rather than rely solely on operating cash flow or equity issuance.
Hock Tan offered a timeline that frames the window of opportunity. He said demand for compute infrastructure could persist through 2029 and potentially into 20301. He also suggested wafer shortages could ease by 2028 or 2029, and that memory shortages might not persist a year from now. Intel's Lip-Bu Tan said the industry was unlikely to experience a significant slowdown over the next two to three years. ◆ If the supply bottlenecks in chips and memory are easing on a one-to-three-year horizon while capital costs are rising, the competitive advantage shifts to organizations that can lock in financing now at today's rates.
Broadcom's Tan added a caveat: demand for computing infrastructure could slow if adoption of advanced AI models disappoints or their capabilities plateau. Intel's Lip-Bu Tan said companies would increasingly need to demonstrate tangible returns from AI investments through improved efficiency and new business models. ◆ The ROI question is the thread connecting all three strands. Family offices expect a bubble but invest anyway. Tencent borrows billions for chips. And the CEOs who sell those chips warn that the entire cycle depends on whether buyers can eventually prove the spending was worth it.
What's next
The family-office survey found that 76% of respondents ranked excessive government borrowing as the top market risk likely to crystallize over the next five years, ahead of even an AI bubble burst. ◆ Rising sovereign debt loads feed directly into the cost-of-capital concern Hock Tan and Lip-Bu Tan raised. Tencent's bond pricing, once finalized, will offer a concrete data point on what the market charges a major AI spender to borrow. If that spread is tight, the debt-funded infrastructure model has room to run; if it widens, the capital bottleneck the two CEOs warned about arrives faster than the chip shortages recede.