A crypto exchange token on sale this Uptober

The market just had its worst week in four months, and the central argument driving the selling was simple: AI infrastructure costs too much to borrow for. The Nasdaq booked its worst week since late June, falling 4.5%, as investors grew more skeptical that heavily debt-funded AI expansion can stay profitable while borrowing costs remain elevated. Into that environment, Meta is expected to report Wednesday after the close, and the question it has to answer is more pointed than usual.

Ten days ago, Mark Zuckerberg said on a podcast that he told his chief AI officer, the company committed to spending up to $200 billion before seeing whether its models would work, saying “We’re going to write this check for like, I don’t know, $100 billion, $200 billion.” The exchange took place at the end of 2025 at a leadership offsite in Hawaii, when his chief AI officer had little more than a “vibe report” on the research and a few charts. “It was certainly not clear at all that it was going to work,” he said. Wednesday is when that commitment gets its price tag.

The Bull Case

Meta’s structural advantage over every AI startup borrowing in private credit markets is that it funds its buildout from advertising, not from lenders. That gap widened materially this week. Reports tied to an Anthropic financing described Blackstone being offered a 10% plus coupon to anchor a $9 billion junior slice of a roughly $60 billion Broadcom-arranged debt package linked to Anthropic’s $125.2 billion Google TPU commitment. In June, reports on an earlier Anthropic structure put the junior layer around an 8.5% yield. Meta pays none of that. It writes the check from its own income statement.

The advertising engine that makes this possible is still accelerating. Total revenue for Q2 2026 was $60.80 billion, an increase of 28% compared to Q2 2025, due to an increase in advertising revenue. Ad impressions delivered across Meta’s Family of Apps increased 14% year-over-year, and average price per ad increased 12% year-over-year. For Q3, Meta guided total revenue to a range of $61 billion to $64 billion. The bull thesis is linear: ads fund chips, chips improve ads, ads fund more chips.

Management continues to expect 2026 operating income to exceed 2025’s level, a promise that, if kept, means the AI spending is additive rather than destructive to earnings. When Anthropic needs to promise a 10% coupon to buy compute time, that promise carries a different weight entirely.

The Bear Case

The advertising revenue line and the cash flow statement tell two completely different stories about Meta’s financial health right now, and the cash flow statement is the one that should concern investors. Free cash flow fell to $784 million because $31.1 billion of quarterly capex consumed nearly all operating cash flow. A year earlier the same metric stood at about $8.5 billion. The company converted a 28% revenue quarter into a roughly 1.3% free-cash-flow margin. A year ago that figure was about 18%.

The spending is not slowing. Meta guided 2026 capital expenditures, including principal payments on finance leases, to $130 billion to $145 billion, and it guided full-year total expenses to $165 billion to $169 billion. Investors will be closely tracking expenses and capital spending, since Meta’s Q2 EPS missed expectations as costs surged. Those figures leave almost no room for the AI products to disappoint. If Muse or the enterprise AI initiatives lag, there is no debt covenant forcing Meta’s hand, but there is also no lender to renegotiate with. The losses land directly on shareholders.

The contrast extends beyond Meta. Alphabet has been making a parallel case to investors — that self-funded AI infrastructure converts to profit faster than debt-funded rivals can match.

The challenge is that the company’s expenses continued soaring. In Q2, total expenses rose to about $42 billion, up 55% year-over-year, driven in part by higher investment in AI infrastructure and also including legal charges and severance related to a May 2026 headcount reduction. Self-funded does not mean consequence-free.

Where the Evidence Leads

The comparison with Anthropic is instructive precisely because it clarifies what Meta’s model actually is. Anthropic received a BB+ credit rating from KBRA, placing it one notch below investment grade as it seeks access to billions in financing. Meta’s equity holders are, in effect, providing the equivalent of that debt, but at no fixed coupon, with no maturity date, and with the added risk that if returns disappoint, there is no floor on the downside. That is better than 10% debt only if the AI products actually pay off.

The Q3 report will not resolve that question. What it will do is tell the market whether the ad engine held up while the $130 billion to $145 billion capex program continues to consume cash. The question is no longer simply how much these companies are investing in AI, but whether those investments are translating into measurable financial returns. The core tension is whether revenue growth can outpace the depreciation and operating costs associated with expanding infrastructure.

Final Verdict

The bull case is modestly stronger heading into Wednesday, but the confidence interval is narrow. Meta’s ad-funded model genuinely is safer than double-digit private credit. The advertising numbers support that view. What gives pause is the Q2 free cash flow: a company spending $31 billion a quarter on infrastructure while generating $784 million in free cash flow is not, in any conventional sense, paying for AI out of its profits yet. It is deferring the cost onto the balance sheet and the future income statement, just without a formal lender to hold it accountable. Watch for whether management tightens or widens the 2026 capex range on Wednesday. Any move above $145 billion, in this week’s market, will be read as an admission that the check Zuckerberg described is still being written.