NextFin News - American consumers are still spending, but the engine has quietly changed. In August, inflation-adjusted personal spending jumped 0.6% — the fastest monthly pace in more than a year — even as prices kept climbing. The resilience is not coming from rising paychecks or fat savings accounts. It is being funded by two parallel engines that did not exist in this shape a decade ago: homeowners are tapping a record pool of housing wealth through second mortgages and HELOCs, while households are stacking paid AI subscriptions into their monthly budgets even as those same subscriptions help push prices higher.
The combination matters because it splits the consumer into two camps and turns a cyclical spending story into something more structural. Homeowners with equity are using their houses as ATMs. Everyone else is leaning on 22% credit cards and a subscription stack that now behaves like a fixed cost. Both groups are keeping the economy moving — for now.
The Situation: Spending Holds While Savings Thin
The headline number is deceptively strong. Inflation-adjusted personal spending rose 0.6% in August from a month earlier, the biggest monthly jump since March 2025, according to Commerce Department data released in late September. Core PCE — the Federal Reserve's preferred inflation gauge — edged up 0.2%. On the surface, the American consumer looks unbreakable.
Beneath it, the funding mix has deteriorated. The personal saving rate fell to 4.1% in August from 4.6% in July, against a historical average of 8.4% stretching back to 1959. Households are spending a larger share of every dollar they take home. At the same time, consumer credit is expanding again: the Federal Reserve reported that revolving credit grew at a 2.5% annual rate in July, and total credit card balances reached $1.263 trillion in the second quarter, up from $1.242 trillion in the first. The average APR on cards accruing interest sat at 22.15% in the second quarter.
Into that gap step two funding sources. First, housing wealth. US homeowners held a record $11 trillion in tappable home equity in the first quarter of 2026, according to Intercontinental Exchange. Second, digital subscriptions — led by generative AI tools — which have become a new, sticky line item in the household budget even as they feed the very price pressures squeezing that budget.
The tension is stark: consumers are paying more for the things AI makes expensive — computers, software, electricity — while simultaneously subscribing to more AI. And homeowners are borrowing against houses they cannot sell without giving up rock-bottom mortgage rates. The spending number stays green. The funding mix gets riskier.
The Lock-In Effect: Why Home Equity, Not Refinancing, Is the Funding Channel
The first thing to understand is why home equity extraction is happening through second liens rather than cash-out refinances. Millions of homeowners locked in first mortgages well below current market rates during the 2020–2022 window. Refinancing would mean surrendering those rates. A second lien — a HELOC or home equity loan — lets them access cash while keeping the cheap first mortgage intact.
The data confirms the channel shift. Equity withdrawals rose 2% year over year in the first quarter of 2026, reaching the highest first-quarter level since 2021. More than half — 54% — of all equity extraction came through second liens. Second-lien withdrawals posted their strongest first-quarter performance in nearly two decades, an 18-year high for the quarter. Cash-out refinances also picked up, hitting their highest first-quarter level since 2022, but they remain the junior partner.
The vintage math is decisive. Nearly two-thirds of second-lien originations in the first quarter came from borrowers who took out their primary mortgage between 2020 and 2022. In total, 3.9 million homeowners from that vintage now carry a second lien.
"The housing market continues to be defined by the lock-in effect," said Andy Walden, head of mortgage and housing market research at ICE. "Millions of homeowners are sitting on first mortgages with rates well below current market levels, making second liens and HELOCs an attractive way to access equity without giving up those loans."
Rates on the second liens themselves have become more inviting. The average HELOC rate fell to 6.6% in March 2026, the most attractive level since late 2022. At that rate, a $50,000 draw carries a monthly payment of roughly $275 — far below what the same borrower would face on a refinanced first mortgage when the average 30-year fixed rate sits above 7%. Introductory HELOC rates have even dipped slightly below the prime rate as lenders compete aggressively for home-equity business.
This is not a cyclical blip. It is a structural feature of a rate environment where the average 30-year fixed mortgage averaged 7.03% as of late September 2026, while millions of existing borrowers sit on rates near 3%. As long as that spread persists, second liens will be the default equity-extraction channel.
"As refinance opportunities become more limited, home equity products are playing a larger role in helping homeowners access liquidity and meet financial goals," said Bob Hart, president of ICE Mortgage Technology.
The Subscription Stack: AI as the New Household Fixed Cost
The second engine is younger, renter-heavy, and shows up in card data rather than mortgage filings. Subscription spending rose 7.7% year over year in July 2026, outpacing overall card spending growth by more than a percentage point and a half — and it has done so for two straight years, according to Bank of America payments data. Entertainment and retail subscriptions account for roughly 43% of the total, but the fastest growth is in reading and information services, a category that includes AI offerings: up 61% year over year among Gen Z and 51% among younger Millennials.
Generative AI has moved from novelty to budget line item with unusual speed. ChatGPT reached 900 million weekly active users by early 2026 — a gain of 500 million users in a single year, meaning more than 10% of the global population now uses it every week. It is 2.7 times larger than Google's Gemini on the web. Paid adoption is compounding even faster than usage: Claude's paid subscribers grew more than 200% year over year and Gemini's 258% as of January 2026, according to Yipit Data, with the category pricing toward a $20-per-month floor.
The burden is stacking. A November 2025 survey of 2,000 US AI users found Americans pay for an average of four premium AI tools at roughly $66 a month. Nearly a quarter spend more than $100 a month, and 14% pay for eight or more AI services. Churn has become the default management strategy — 53% say they cancel and restart AI tools as needed — but the aggregate stack keeps growing. US households now spend roughly $273 a month on subscription services overall, and 89% underestimate that total.
Here is the irony: these subscriptions are simultaneously the coping mechanism and part of the problem. AI spending is feeding the inflation that is forcing households to seek extra cash in the first place.
The Feedback Loop: AI Subscriptions Help Drive the Inflation They Strain Against
The transmission channel is measurable. Investment in AI and data centers has left a clear imprint on prices. The Richmond Fed documented that year-over-year growth in the personal consumption expenditure category for computer software and accessories rose to 14.5% in May 2026. Strong demand is pushing up prices for graphics processors, computer storage, and software subscriptions, and firms are passing component costs through to shoppers.
The macro footprint is larger than the consumer sticker shock suggests. A St. Louis Fed study found that AI-related investment added 0.97 percentage points to real GDP growth in the first three quarters of 2025. That demand surge is inflationary in the near term even as it promises productivity gains later. Mark Zandi, chief economist at Moody's Analytics, estimated that AI is contributing roughly 0.2 percentage points to overall inflation — small in the aggregate, but concentrated in the exact categories households feel: computers, software, and the electricity that powers data centers. He calculated that higher inflation means households must spend just over $375 more to buy the same basket of goods and services they did a year earlier, with AI's impact embedded in that increase.
Oxford Economics expects the dynamic to persist. Lead US economist Bernard Yaros wrote earlier this year that he anticipates AI-driven price surges will "continue to provide an atypical boost to core inflation over the next two years." The Fed's chairman, Kevin Warsh — sworn in on May 22, 2026, after the narrowest confirmation vote in the position's history — inherits the dynamic: a spending boom in one sector lifting prices economy-wide while households finance consumption through debt and equity extraction.
So the loop closes on itself. AI investment lifts incomes and GDP in the near term. It also lifts the price of the tools, the hardware, and the power. Households subscribe to the tools anyway — because they are now embedded in work, school, and daily life — and then tap home equity or credit cards to cover the rest of the basket. The spending number stays green. The funding mix gets riskier.
The Counter-Thesis: A Late-Cycle Sugar High, Not a New Regime
The bear case is straightforward and backed by the same data set. A 4.1% saving rate is not far above the pre-pandemic trough; record credit card balances at 22% APR are not a sign of financial health; and home equity extraction has historically peaked late in the cycle, just before downturns. Serious delinquencies on HELOCs ticked up from the first to the second quarter of 2025, and Fitch Ratings reported that the average 30-day-plus delinquency rate on HELOC transactions reached 2.2%, up 123 basis points, in early 2025. Overall default rates remain low, but the direction of travel is the warning.
The strongest version of this argument says the current spending is a sugar high funded by three exhaustible sources: pandemic-era excess savings (drawn down), home equity (finite and tied to house prices), and credit capacity (limited by delinquencies). When those run out, the consumer breaks — and with consumption roughly two-thirds of GDP, the economy follows.
There is force to that view, but it misses what is structural here. The rate lock-in is not temporary: millions of sub-market first mortgages will not refinance away until rates fall materially, and when they do, cash-out refinances — not second liens — will simply take over as the extraction channel. The funding vehicle rotates; the extraction continues. Similarly, AI subscriptions are not a fad layer on top of discretionary spending. They are embedded in workflows — Excel, PowerPoint, Gmail, Docs — with switching costs that make them behave like utilities. Churn is high at the individual tool level, but the category spend is sticky because the use case is structural, not cyclical.
That said, the bear case has a clear falsifying signal. If the S&P CoreLogic Case-Shiller US National Home Price Index posts two consecutive quarters of declines while the saving rate stays below 4%, the equity engine stalls and the sugar-high thesis wins. Falling house prices would freeze the lock-in trade — homeowners could not extract from depreciating collateral — and a sub-4% saving rate would leave no buffer for the subscription stack. That is the scenario to watch.
What Comes Next: Winners, Losers, and the Three Signals
The near-term, medium-term, and long-term readings point in different directions, and that is the point.
In the short term — the next two to three quarters — the spending floor holds. Home prices remain firm across most of the country: nearly 70% of major markets posted annual gains in May 2026, and affordability is still roughly 3% better than a year ago despite a roughly 50-basis-point rise in mortgage rates since February, according to ICE. Equity extraction will continue, and subscription growth will compound into the holiday quarter. Growth data will likely surprise to the upside, and the consumer-discretionary complex benefits.
Over the medium term — six to twelve quarters — the funding mix is the risk. Home equity is not income. If wage growth does not accelerate to match spending, the saving rate will grind lower and debt-service burdens will rise, especially for the renter cohort financing consumption on 22% credit cards rather than 6.6% HELOCs. Lenders specializing in second liens and home equity products win volume; issuers of sub-prime credit cards and unsecured consumer lenders face rising charge-offs.
In the long term, the structural read dominates. If AI delivers the productivity gains that investment implies, real incomes rise and the debt burden becomes manageable — the 0.97 percentage points of GDP growth from AI investment today becomes a deflationary force tomorrow. If the productivity payoff disappoints, households are left with a permanent new fixed cost and a levered balance sheet. The asymmetry cuts across asset classes: homebuilders and mortgage servicers with strong home-equity origination benefit from the current regime; consumer finance companies with heavy unsecured exposure are the vulnerable side.
Three signals decide which path unfolds. First, home prices: two consecutive quarterly declines in Case-Shiller would break the equity engine. Second, the saving rate: a sustained move below 4% with spending still rising signals balance-sheet stress, not strength. Third, AI pricing power: if software and computer price inflation stays in the double digits while paid-subscriber growth slows, the subscription stack is reaching saturation — and the feedback loop is about to tighten.
The American consumer is not running on fumes. But the tank is no longer being filled by wages — it is being filled by equity withdrawals and $20-a-month subscriptions. That works until house prices stop rising or the tools stop feeling essential. Watch the saving rate, not the spending number. It is the one that tells you when the engine has changed for good.
Explore more exclusive insights at nextfin.ai.
