Just next week! Meta (META.US) earnings take the lead, massive AI investment faces a "confidence test"

Title context: Just next week! Meta (META.US) earnings take the lead, massive AI investment faces a "confidence test" Text: The clamor that "AI is short of money" is growing louder, dragging down a swath of US tech stocks and even hitting the share prices of energy companies. And next week, the sustainability of massive AI investment will face a major confidence testthe earnings disclosures of the giants. After the US market close on October 21 local time, Meta (META.US) will report its third-quarter earnings, becoming the first US tech giant in this wave of skepticism to face market scrutiny and have the effectiveness of its AI capital expenditure tested. "One of the most noteworthy items this earnings season is whether the profitability improvement of hyperscale cloud service providers can outpace the growth rate of capital expenditure," said Kevin Shea, senior equity analyst at BNY Wealth. Chuck Carlson, chief executive officer of Horizon Investment Services, also pointed out that rising capital costs are raising the hurdle for investment returns. "Rising capital costs mean this capex boom will become more expensive, and investment returns must clear a higher bar. Meta will be the first to try to answer this question." Commercialization results and free cash flow Meta previously guided revenue for the quarter at $61 billion to $64 billion, up 19% to 25% from $51.2 billion a year earlier, with the midpoint implying about 22% growth, a slowdown from 28% in the second quarter. According to FactSet data, the median analyst consensus expects revenue of $63.29 billion, up 23.6%, and adjusted earnings per share of $6.85. Meta's AI Agent Muse, launched in September, directly set off the global AI community and also drove a sharp rise in its own share price; this month the company launched another new model, Muse Wonder, strengthening memory focus and complex task processing capabilities. Download data after Muse's launch has been quite impressive, but on the more critical user retention metric, the data shows Muse's 14-day retention rate is 22%, slightly below ChatGPT's level over the same period; Guggenheim analyst Michael Morris, citing Apptopia data, noted that five weeks after launch, Muse reached 1.3 million daily active users, less than 0.05% of Meta's total 3.6 billion daily active users. Meta has also begun the commercialization process. On October 12, Meta started charging for Muse, launching two subscription tiers at $20 and $100 per month, but there is no revenue data yet. Baird investment strategist Ross Mayfield believes that at least Muse has proven to investors that massive AI investment can actual product value; Ben McMillan, chief investment officer at IDX Advisors, noted that Meta still needs to prove it can convert user growth into revenue. What worries the market more is Meta's free cash flow. In the second quarter of this year, Meta generated operating cash flow of $31.9 billion, but capital expenditure (including finance leases) reached $31.1 billion, leaving free cash flow at just $784 million, far below $8.549 billion a year earlier. The consensus compiled from institutional surveys shows Meta's third-quarter free cash flow will turn negative to -$9.14 billion, with capital expenditure reaching $40.97 billion. The funding gap is being bridged through borrowing. In the second quarter, Meta issued $24.9 billion in long-term debt. As of the end of the quarter, the company's long-term debt balance reached $83.7 billion, while cash and marketable securities were $90.3 billion. Meta did not repurchase shares in the second quarter, compared with $10.2 billion in buybacks a year earlier. Chief Financial Officer Susan Li explicitly stated on the earnings call that Meta plans to use a "higher proportion of debt" in financing long-term infrastructure construction. Multiple institutions warn that "AI is short of money" Cash-generation ability and the feasibility of external financing are also important factors igniting market concerns that "AI is short of money." A latest Gartner analysis report warns that if AI service providers continue to maintain the current pace of capital expenditure, by the end of 2030, more than $3.5 trillion of AI infrastructure investment may face asset impairment risk. The institution estimates that by 2030, AI data center owners will need to generate cumulative revenue of $8.5 trillion to cover announced capital expenditure, financing costs, and achieve a 12% return on invested capital; but over the same period, enterprise and consumer AI business revenue attributable to AI data centers may be at most only $3.6 trilliona huge gap of $4.9 trillion between the two. On borrowing, a recent report by Goldman Sachs Group, Inc. estimates that global hyperscale cloud providers' capital expenditure is expected to reach $1.2 trillion in 2027, while operating cash flow over the same period totals about $1.1 trillion, forcing at least 35% of construction spending to rely on debt financing, exceeding $400 billion. Carlyle Group Inc. gives a more aggressive forecast, expecting that by 2032, more than 80% of AI capital expenditure will require external financing. The problem is that external funding is not inexhaustible. The massive supply of US Treasuries brought by the US fiscal deficit, companies' own financing needs, and the impact of interest rate levels on borrowing costs will all determine whether the market can provide sufficient funds at reasonable prices. And when financing is constrained, it will inevitably affect the expansion speed of the entire industrial chain. It is worth mentioning that Goldman Sachs Group, Inc. also warned that "the real risk right now is that high capital costs and real-world delivery bottlenecks are preventing end demand from translating into actual capital expenditure." Tightening credit and rising capital costs lead to delay risks in computing power infrastructure construction, and the semiconductor supply chain may face an order cliff in 2027-2028 where demand cannot be converted into actual purchases. Against the backdrop of market turbulence, the industrial chain itself remains relatively optimistic in attitude and actions. Regarding the recent sharp selloff in AI stocks, NVIDIA Corporation CEO Jensen Huang said at an event in Berlin: "This too shall pass, and the reason is very simple. AI will bring transformation. It will affect every industry and every country in a very positive way. There is no doubt that this impact will continue, and we are at the beginning of this process." A regulatory filing submitted by Microsoft Corporation on Wednesday showed that the company plans to expand its global data center capacity from about 12 gigawatts to more than 38 gigawatts by 2032, an increase of more than twofold, to alleviate the company's computing power shortage. This article is reprinted from "STAR Market Daily", author: Zheng Yuanfang; GMTEight editor: Yan Wencai.