Three forces are acting on the secondhand market at once, and they don’t all push the same direction.

Ask what AI is doing to electronics resale, and the usual answer is “better pricing algorithms.” That’s true, but it’s the least interesting part of the story.
The bigger shift is that AI is acting on used tech from three directions at once: as a demand shock in the components market, as an inspection technology on the warehouse floor, and as a new layer between buyers and listings. Those forces don’t agree with each other. One is quietly adjusting what your three-year-old laptop is worth. One is streamlining the listing process. And one is rewriting who decides what a shopper sees.
Here is how each one works, and what it means if you are the person with an old gadget sitting in a drawer.

AI is Making Your Old Laptop More Valuable for This One Component

The most direct effect AI has had on resale value now has less to do with the brand and release year. It is memory.

AI accelerators need high-bandwidth memory, and it is manufactured on the same wafers as the DDR5 and NAND that go into laptops, phones, and consoles. The three companies that make nearly all the world’s DRAM have been steering capacity toward the higher-margin product. J.P. Morgan Global Research estimates DRAM prices will have climbed more than 400% from the start of 2024 through the end of 2026, and IDC projected 2026 DRAM supply growth at roughly 16% year over year, below what the market needs.

The mechanism matters more than any single figure: AI demand is raising the cost of new hardware, and the resale floor is rising underneath it. That is a change in the shape of depreciation, not a marketing line.

As a result, new hardware costs more. And when prices go up, many turn to pre-owned alternatives. Assurant reports that consumer interest in buying a certified pre-owned device rose from 17% in 2024 to 25% in 2026. Drawing on Counterpoint data, the same analysis notes refurbished smartphone sales grew 6% year over year in the second half of 2025 and are projected to rise around 11% in 2026, while new shipments are expected to decline.

A Camera Now Can Grade the Condition

Used electronics have always had an honest disclosure problem. The seller is presumed to know relevant details about the device they are selling. And the buyer expects those details to be disclosed. That’s where the disconnect often takes place. Economists call this the “lemons problem”, and it’s why secondhand markets have historically priced everything as though it might be the worst version of itself. The discount buyers demand is not really about wear. It is about uncertainty.

AI-assisted grading attacks that directly. Computer vision systems photograph a gadget and rate its cosmetic condition against a trained standard, not a technician’s subjective judgment alone. Automated diagnostic rigs run functional tests for screen dead zones, camera lens haze, touch response, audio distortion, and battery health, then attach the evidence to the grade.

The scale is real. Assurant, a U.S. company that grades trade-ins and returns on behalf of carriers and retailers, says it processed 13.5 million devices globally in 2025 through a recovery network reaching 1,500 downstream partners across 33 countries, using AI-assisted diagnostics.

Two consequences follow:

1 – Grades get more consistent. The classic trade-in dispute — you said Good, they are calling it Fair — comes from two humans applying a fuzzy standard on different days. A model applying a fixed standard with photographic evidence attached narrows that gap considerably.

2 – The demand for reconditioned tech grows. Of the 162.8 million refurbished smartphones sold worldwide in the second half of 2025, Counterpoint data cited by Assurant shows 52.9 million changed hands with no mechanical work performed: wiped, cleaned, reboxed, sold. The other 109.8 million were repaired first. Dependable grading is precisely what makes that no-repair channel safe to operate at volume.

If you want the underlying variables grading systems actually score, we broke them down in our “How much your laptop is worth and the factors that determine resale value” blog post.

The third force gets the least attention from sellers, but it may matter most in the coming years.

eBay has rolled out a conversational AI shopping agent to a share of U.S. customers, surfacing inline on the page to help with product discovery rather than waiting for a keyword search, Retail Dive reported. In its own 2026 outlook, eBay leadership described a shift from transactional to agentic commerce, with AI handling verification and other nuanced steps of a transaction from beginning to end.

But here is the complication: Platforms want to control the agents. eBay moved to restrict third-party “buy-for-me” bots from transacting on its marketplace, following a similar tightening at Amazon, as covered by EMARKETER. In other words, marketplaces are betting on AI-driven search discovery, as long as it’s their agents doing the searching.

The NPU is Splitting the Used Market in Two

A fourth effect emerges from the first three: the divide between pre- and post-AI-release pre-owned technology.

This divide hurts the resale value of computers and smartphones released before 2024. Models without AI capabilities depreciate faster than they normally would over the same period. At the same time, memory crunch slows this rapid resale value decline. 

Here is the whole picture overview:

Four AI forces and which way each one pushes used device valueForceHow it worksDirection on used valueWhat to do about itMemory demand from AI data centersHBM production crowds out the consumer DRAM and NAND that new devices need, raising new prices UpwardLifts the floorStop assuming last year’s depreciation curve still applies to your deviceAI grading and automated diagnosticsVision models and test rigs score condition against a fixed standard and attach evidence UpwardShrinks the uncertainty discountPhotograph and document real condition before shipping; accuracy now paysAgentic discovery and AI searchConversational agents mediate what buyers see; platforms are restricting outside agents IndirectShifts where demand landsChoose the channel by where buyer traffic actually is, not by habitThe NPU capability lineA hard boundary between NPU-equipped machines and everything older DownwardPremium pre-NPU models give up ground fasterIf your machine sits just below the line, selling sooner beats waiting

What Does All of This Mean for the Resale Value of Your Tech?

Essentially, the practical read is this: Used gadgets most exposed to the downside are consumer-grade models that sit one generation below a capability line (pre-AI tech) and are being held rather than sold. 

The cause for wasted value, though, is often not depreciation but inertia. Counterpoint counted 494 million pre-owned smartphones globally in the second half of 2025 and found that only 178 million of them were actually resold. The remaining 316 million went into drawers or became spares, so no materials returned to circulation.

That gap is worth closing, and it is why the reselling side of the market matters as much as the recycling side. At Gadget Salvation, we calculate buyback offers against current resale market data rather than a fixed depreciation schedule. The secondary market moves constantly—that is the whole point of this article—and a valuation built on last year’s trends gets old before you finish reading it. The variables are the same ones AI grading systems evaluate: model, configuration, functional condition, cosmetic condition, and live demand for that specific gadget. The offer is locked for 30 days, so any market shift between your checkout and our receiving the shipment is our headache, not yours. Extending the life of a working device is the whole point. Every used laptop, smartphone, or tablet that gets a second owner is one that doesn’t need to be manufactured and doesn’t become e-waste.

An Algorithmic Offer is Still an Algorithm

One thing is worth noting: automated pricing is under active scrutiny in the United States. The Federal Trade Commission has sought public comment on a proposed enforcement policy statement covering personalized pricing — the practice of using a consumer’s own data to estimate what they will pay — and several states have moved on disclosure requirements or outright restrictions, as Wiley summarizes.

Most of that activity targets consumer-facing retail rather than buyback, and the two are genuinely different: a trade-in quote is an offer made to you, not a price extracted from you. But the underlying principle travels. An automated valuation is a set of assumptions someone chose, expressed as a number. A model generating it doesn’t make an offer fair, and it doesn’t make it unfair either. It makes it worth checking.

Frankly, no platform in the recommerce category volunteers its valuation model, and ours is no exception. Pricing logic is competitive information, and logic that tracks a market moving this fast would be out of date the moment it was published as a fixed statement anyway. So do not evaluate a platform on whether it shows you the arithmetic. Evaluate it on whether it stands behind the number it gave you.

Three questions to ask any buyback company:

  • What specifically does each condition grade mean, in plain language?
  • What can prompt a price revision after inspection?
  • If the offer is revised downward and I reject it, do I get the device back at no cost?

Conclusion

AI is changing the electronics resale space in several ways. Rebuilding discovery around agents and reshaping demand for older tech are just a few of the more obvious ones. It has also drawn a spec line that is widening the gap between the top and bottom of the used market further.

One thing doesn’t change: a depreciating gadget that stays in a drawer is worth nothing to you and nothing to the planet. That part is still entirely manual.

Sources:

  • J.P. Morgan Global Research — The AI-Driven Memory Shortage: DRAM Prices, Inflation and Market Risks — Estimates that DRAM prices rise more than 400% from the start of 2024 through the end of 2026.
  • IDC — Global Memory Shortage Crisis: Market Analysis and the Potential Impact on the Smartphone and PC Markets in 2026 — 2026 DRAM supply growth of roughly 16% year over year.
  • Consumer Reports — With AI Data Centers Scooping up RAM, Laptop Prices Could Spike in 2026 — The chain from open-ended hyperscaler memory orders through to higher retail laptop prices.
  • Tom’s Hardware — Memory price surge begins to cool as consumers hit affordability limit — Contract prices still climbing through the third quarter of 2026 at a slowing rate, citing TrendForce.
  • Assurant — DRAM demand is shifting the smartphone supply conversation — Certified pre-owned consumer interest rising from 17% in 2024 to 25% in 2026; refurbished smartphone sales up 6% year over year in H2 2025 and projected to rise roughly 11% in 2026; the 15% to 20% seasonal decline in pre-owned pricing; 13.5 million devices processed in 2025 across 1,500 downstream partners in 33 countries; AI-assisted diagnostics in grading.
  • Retail Dive — eBay’s conversational AI shopping agent rollout — Staged release of eBay’s shopping agent to a portion of U.S. customers, surfacing inline for product discovery.
  • eBay Inc. — AI Trends to Watch In 2026 — The shift from transactional to agentic commerce, and AI helping shoppers discover pre-owned alternatives.
  • EMARKETER — eBay joins Amazon in tightening control over AI shopping agents — Both marketplaces restricting third-party “buy-for-me” agents from transacting.
  • IDC, via AMD-commissioned white paper — The Rise of AI PCs in the Commercial Market — The IDC definition of an AI PC as CPU, GPU, and NPU on one physical processor, and Microsoft’s Copilot+ threshold of 40 TOPS.
  • Gartner — Gartner Forecasts Worldwide Shipments of AI PCs to Account for 43% of All PCs in 2025 — AI PCs at 43% of PC shipments and AI laptops at 51% of the laptop market in 2025, and the projection that AI laptops become the only option for large businesses in 2026.
  • Wiley — Algorithmic Pricing Attracts Widespread Scrutiny as FTC Seeks Comment on Personalized Pricing Policy Statement — The FTC comment period on personalized pricing and related state-level disclosure requirements and restrictions.