Record Highs on AI Capital, But the Consumer Is Cracking

The S&P 500 closed the week of August 10–14 at 776.34 (SPY), slipping 0.20% on Friday but capping its third consecutive weekly gain.^[1] The index briefly traded above 7,800 on Thursday for the first time, powered by cooler-than-expected producer inflation data and the heavyweight technology cohort.^[2] The Nasdaq (QQQ) finished at 731.07, up modestly from 723.03 at the prior week’s close^[3] — a far quieter week than the 5.2% surge it posted the week before.^[4]

The calm at the top of the tape masks a divergence widening underneath. On one side, the AI infrastructure build-out just received the largest single capital commitment in its history. On the other, the American consumer — still 68% of GDP — is flashing warning signals that grow harder to dismiss.

The AI Capital Surge

On August 10, Nvidia announced memorandums of understanding with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to establish AI compute infrastructure financing platforms aimed at mobilizing more than $500 billion in third-party capital.^[5] The partnership would allow hyperscalers, frontier AI labs, and enterprises to borrow against future Nvidia compute revenue streams — effectively securitizing AI infrastructure the way mortgages were securitized a generation ago.^[5]

The announcement landed at a moment when capital expenditure from the largest technology companies has already reached unprecedented levels. It also reframed the central question around AI spending: not whether the build-out is sustainable, but whether Wall Street’s balance sheet can absorb enough of the risk to keep corporate balance sheets clean.

Intel’s $20 Billion Bet

The same week, Intel priced a $20 billion common stock offering at $95 per share — upsized from an initial $15 billion plan.^[6] CEO Lip-Bu Tan disclosed he and a family member would personally purchase $12 million of shares at the offering price, buying 105,263 shares through a family trust.^[6] The proceeds are earmarked for “general corporate purposes,” but the context is clear: Intel is racing to expand foundry and manufacturing capacity to capture AI-driven semiconductor demand.^[6]

Intel’s stock rose 3.54% intraday on the announcement^[6] — a sign that the market is rewarding conviction, not just scale. The CEO’s personal capital at risk is a sharper signal than any press release.

Earnings Still Ahead of Expectations

With roughly 88% of the S&P 500 having reported Q2 results, blended earnings growth stands at 50.4%.^[4] Even excluding the outsized contributions from Alphabet and Amazon, growth still runs at 29.2%, comfortably above longer-term averages.^[4] The share of companies raising forward guidance is at its highest level since 2021.^[4]

Not every beat was rewarded. AMD, SanDisk, and Western Digital all fell on high expectations despite solid results^[4] — a reminder that when the bar is this elevated, beating it is no longer sufficient.

The Consumer Cracks

Beneath the earnings strength, the data flowing in on Friday told a different story. The University of Michigan’s preliminary August consumer sentiment index fell to 51.0 from 55.2 in July — a 7.6% month-over-month decline and 12.4% below a year ago.^[7] Consumers cited rising living costs tied to Middle East conflict and persistent inflation as primary concerns.^[7]

Simultaneously, the Census Bureau reported that July retail sales fell 0.6% — the largest monthly decline in more than a year.^[8][7] This followed a July jobs report that showed nonfarm payrolls fell by 23,000 against expectations for an 80,000 gain, with May and June payrolls revised down by a combined 103,000.^[4]

The unemployment rate ticked down to 4.1%^[9], but only because labor-force participation fell to its lowest level in over five years.^[4] The decline in participation is not strength — it is a shrink in the denominator.

The Macro Backdrop

The macro picture is one of controlled disinflation with growth intact — at least on paper.^[9] The Fed funds rate has fallen 70 basis points year-over-year, inflation is cooling, and GDP growth remains positive. Credit spreads are tight, and the VIX sits near 14.63^[9], a level historically associated with complacency rather than caution.

The closest historical analogs the data identifies are mid-2006 — a period of soft landing that preceded the 2007–2009 credit crisis by roughly a year^[9] — and late 2007, when the yield curve had just un-inverted and the economy was on the threshold of recession.^[9] These analogs are not predictions. They are reminders that the transition from “soft landing” to “something harder” often arrives quietly, through consumer channels, long after the equity market has declared victory.

What Would Have to Be True

For the bullish case to hold, three things need to line up: the $500 billion in AI infrastructure capital needs to translate into actual data-center deployments and recurring revenue, not just memorandums; Q2 earnings strength needs to carry into forward guidance — and next week’s retail earnings will test whether companies see the consumer deterioration in their own numbers; and the Fed needs to hold, not hike, despite CPI at 3.3% and a 10Y Treasury at 4.63%. Friday’s weak jobs report and soft retail sales pushed September rate-hike odds down to around 44%, from above 54% earlier.^[4]

For the bearish case, the consumer is already doing the work. Sentiment at 51.0 is recession-adjacent territory.^[7] Retail sales falling at the fastest pace in over a year^[8] — combined with negative payroll revisions^[4] — suggests the “bad news is good news” dynamic may be reaching its limit. There is a threshold beyond which bad news stops being good for equities and simply becomes bad.

What to Watch Next

The week of August 17 is a retail earnings checkpoint.^[10] Target reports August 19, Walmart on August 20, Home Depot on the 19th, and Lowe’s on the 20th.^[10] These reports will either confirm or contradict the consumer sentiment and retail sales data that rattled Friday’s session. If Walmart and Target guide conservatively, the divergence between AI-driven corporate earnings and consumer-facing reality becomes harder for the market to bracket as transitory.

Then, on August 26, Nvidia reports Q2 FY2027 earnings after the close.^[10] This is the single print that will either validate or challenge the $500 billion infrastructure thesis. The stock closed Friday at $225.16^[11], and the options market will be pricing in a move that sets the tone for the entire technology sector heading into September.

The stakes are asymmetric. If Nvidia’s numbers and guidance are strong, the AI infrastructure narrative gets another lease on life and the market’s third consecutive weekly gain looks like a floor. If the consumer data continues to deteriorate and retail earnings confirm the weakness, the market faces a question it has been able to avoid all summer: can AI infrastructure spending sustain a market when the broader economy is slowing?

The answer is not yet knowable. What is knowable is that the data points arriving over the next two weeks — Target, Walmart, and Nvidia — will narrow the range of plausible answers considerably.