A $150 Android phone used to spend most of its bill of materials on the screen, the battery and the camera module. That math broke this year. Intel chief executive officer Lip-Bu Tan, speaking at the AI Infrastructure Summit 2026 in Santa Clara on September 15, 2026, said memory chips now swallow “70-80% of the cost” of budget smartphones and laptops. The remark landed at an event built around AI infrastructure, but the sharpest damage is showing up in devices that have nothing to do with AI at all.
Tan’s comments add a hard number to a shortage that chipmakers, phone brands and PC vendors have been describing in vaguer terms since early 2026. He told the Santa Clara audience: “I expected memory to be a big bottleneck early last year, but it actually happened and it will become more serious next year.” He followed with a blunter diagnosis of the supply side: “The production capacity is very limited, so many businesses are being delayed because they do not have enough memory.” For a company that makes CPUs, not memory, the fact that Intel’s own CEO is sounding the alarm shows how far the shortage has spread beyond the chipmakers who actually produce DRAM and NAND.
What Tan Told the AI Infrastructure Summit
The AI Infrastructure Summit 2026, held at the Santa Clara Convention Center, drew the kind of crowd that usually talks about GPU clusters and power budgets. Tan used his slot to pivot the conversation toward a quieter crisis: memory. He said prices for memory chips have climbed “five, six, and seven times” depending on the product category, a range wide enough to cover both mainstream DDR5 modules and the high-bandwidth memory that AI accelerators depend on.
That 70-80% cost-share figure is the number worth sitting with. It does not describe a flagship phone with a $1,200 price tag and a titanium frame. It describes the sub-$200 devices that make up most of global phone shipments and the entry-level laptops schools, small businesses and budget-conscious buyers rely on. When memory eats most of a device’s cost, there is almost no room left for the vendor to absorb the hit. The price gets passed straight to the buyer, or the device gets redesigned with less RAM and less storage than the previous generation.
Why Budget Devices Are Taking the Hardest Hit
Flagship phones and premium laptops can absorb a memory price spike because memory was never the dominant cost driver in the first place. A $1,000 phone spends heavily on its camera system, its display panel and its brand premium. A $120 phone spends heavily on memory, full stop. That is the structural reason budget hardware is where this shortage shows up first and hardest.
Reporting from Reuters on the shortage described smaller phone and laptop makers redesigning products and passing costs on to buyers because the scarcity is expected to run for years, not months. ABC News Australia covered the same squeeze from the demand side, tying the AI buildout directly to component costs that ripple down into ordinary consumer electronics. A separate report from Briefs found that sub-$100 smartphones, long a staple in price-sensitive markets across South Asia and Africa, are being squeezed out of production entirely because there is no margin left to cut once memory takes its share.
The redesign trade-off
Vendors facing this math have three levers: raise the price, cut the memory spec, or eat the margin loss. Most are choosing some blend of the first two. A device that shipped with 6GB of RAM last year ships with 4GB this year at the same price point, or the same 6GB spec now costs $20 more. Neither option is popular with buyers, but the alternative for many smaller manufacturers is exiting the segment altogether.
The Numbers Behind the Shortage
Memory research firm TrendForce has tracked the price climb in granular detail through 2026. Its DRAM spot price tracker showed mainstream DDR5 16Gb chips holding at an elevated average in mid-September, well above where the category sat a year earlier. A TrendForce market update from September 2, 2026 noted that even as some end-demand softened, suppliers held firm on quotes, keeping prices elevated rather than letting them slide back down. That is an unusual dynamic. Normally, softer demand brings prices down quickly. This time, tight supply is doing more work than demand in setting the price floor.
| Memory Category | Approx. Price Move in 2026 | Primary Driver |
|---|---|---|
| NAND Flash contract price | Up roughly 55-60% in Q1 2026 alone | Reduced fab output amid AI storage demand |
| Conventional DRAM contract price | Up 13-18% quarter-over-quarter after a prior 93-98% jump | Server and AI accelerator memory demand |
| Mainstream DDR5 16Gb spot price | Holding at elevated levels through mid-September 2026 | Suppliers maintaining firm quotes despite softening demand |
| Budget device memory cost share | 70-80% of device bill of materials, per Intel’s Tan | Fixed memory floor price against a low device price ceiling |
The table above pulls together the pattern across memory categories that TrendForce and Intel’s own leadership have described this year. NAND moved first and fastest. DRAM followed with a slower but still severe climb. The budget-device cost share is the downstream effect: the same dollar increase in a memory chip hits a $2,000 laptop as a rounding error and hits a $150 laptop as an existential pricing problem.
HBM Is Pulling Capacity Away From Everything Else
Behind the DRAM and NAND numbers sits a more specific culprit: high-bandwidth memory, the stacked chips that feed AI accelerators like Nvidia’s GPU lines. Fab capacity used to produce HBM is capacity that does not go toward the DDR5 and LPDDR modules that phones and laptops need. CNBC’s coverage of the memory crunch earlier in 2026 reported that Micron, Samsung and SK Hynix executives all pointed to the same underlying constraint: AI customers are willing to pay premium prices for guaranteed HBM allocation, and that pulls wafer capacity away from commodity memory production.
Micron chief executive officer Sanjay Mehrotra described supply conditions as very tight in CNBC’s coverage of Micron’s fiscal Q2 2026 earnings, with customers receiving only a fraction of the memory volume they were requesting. Micron’s business leadership separately described demand as having outpaced the industry’s ability to supply it at all, a dynamic that plays out identically whether the customer is a hyperscaler buying HBM for a GPU cluster or a phone brand trying to secure LPDDR for its next budget model. The difference is leverage. Hyperscalers can pay up. Budget phone brands mostly cannot, which is exactly why the squeeze shows up first at the low end of the market.
Samsung, SK Hynix and Micron: The View From the Suppliers
The companies that actually make memory are not shy about how tight the market has become. Samsung’s memory business flagged continued shortages stretching into 2027 in its own 2026 earnings commentary. SK Group’s leadership has floated an even longer timeline for the crunch to ease, framing it as a multi-year supply and demand mismatch rather than a temporary bottleneck tied to one product cycle.
What makes this shortage different from past memory cycles is that it is not purely cyclical. Historically, DRAM and NAND prices moved in boom-bust waves tied to fabs overbuilding capacity and then flooding the market. This time, a structural new source of demand, AI accelerator memory, is competing directly with consumer electronics for the same fabrication lines. Samsung and SK Hynix both build HBM and commodity DRAM on related process technology, which means every wafer allocated to HBM for a GPU customer is a wafer not available for a phone or laptop chip.
Historical Context: How Memory Shortages Usually Play Out
Memory has always been the most cyclical part of the semiconductor industry. The 2017-2018 DRAM shortage pushed prices up sharply before oversupply crashed them back down within two years. The 2021 shortage tied to pandemic-era demand for laptops and gaming consoles followed a similar arc: prices spiked, manufacturers built more capacity, and by 2023 memory was cheap again.
What breaks that pattern this time is the durability of AI demand. Past shortages were driven by a temporary spike, like a new game console launch or a pandemic work-from-home surge, that manufacturers could plan around and eventually outbuild. AI accelerator demand for HBM does not look temporary. Every major cloud provider and chipmaker is locked into multi-year capacity commitments, which means the fabs that would normally pivot back to commodity DRAM production once a shortage eases have less incentive to do so this time. That is the mechanism behind warnings, including Tan’s own past comments, that relief may not arrive until 2028.
Competitive Comparison: How the Major Players Are Positioned
Not every company in the memory supply chain is affected the same way. Some are direct beneficiaries of the price spike. Others, like Intel, are on the buying side and absorbing higher input costs across their own product lines.
| Company | Position in the Shortage | 2026 Stance |
|---|---|---|
| Samsung | Major DRAM, NAND and HBM producer | Flagged shortages continuing into 2027 in earnings commentary |
| SK Hynix | Leading HBM supplier for AI accelerators | Reported its 2026 HBM output effectively committed to customers |
| Micron | DRAM, NAND and HBM producer | CEO Sanjay Mehrotra described supply as very tight, per CNBC |
| Intel | Buyer of memory for its own CPU platforms | CEO Lip-Bu Tan is publicly warning the shortage is worsening, not easing |
| Budget phone and laptop OEMs | End buyers with the least pricing power | Redesigning products and raising prices, per Reuters and Briefs reporting |
The pattern in that table is straightforward. Companies that make memory are, on balance, benefiting from higher prices even as they acknowledge the strain it puts on the rest of the industry. Companies that buy memory to build finished devices, whether that is Intel building CPU platforms or a budget phone brand building a $130 handset, are on the losing side of the same trade.
Market Impact: Who Feels the Squeeze First
The immediate market impact splits into three groups. Consumers in price-sensitive markets face fewer choices at the bottom of the phone and laptop market, or the same choices at higher prices. Smaller device manufacturers without the scale to negotiate favorable memory supply contracts face margin compression that larger rivals can better absorb. And memory producers themselves are enjoying a rare period of pricing power after years of commodity-style competition that kept margins thin.
There is a secondary effect worth watching too. When budget hardware gets more expensive or gets spec-cut, the replacement cycle in emerging markets slows down. Phones and laptops that would normally get replaced every two to three years stay in service longer, which in turn affects everything from software update support windows to the used-device resale market. None of that shows up in a quarterly earnings call, but it compounds over the two-to-three year horizon that most industry watchers now expect this shortage to last.
What OEMs Are Actually Doing About It
Device makers facing this squeeze are not sitting still. Reuters reporting on smaller phone and laptop manufacturers described active redesign work aimed at reducing memory footprint without gutting the user experience entirely. That can mean shifting more workload to cloud processing to reduce on-device RAM needs, trimming storage tiers, or delaying product refreshes until supply loosens.
Larger OEMs with more purchasing leverage are locking in longer-term supply contracts at fixed prices, a hedge that smaller competitors generally cannot afford. That dynamic tends to consolidate market share toward brands large enough to secure guaranteed allocation, which is one more way this shortage reshapes competitive position beyond the immediate price tag on a store shelf.
What This Means for Anyone Buying Hardware Right Now
For buyers shopping in the fourth quarter of 2026, the practical takeaway is simple: budget-tier phones and laptops are the segment most likely to see price hikes or spec cuts in the coming months. If a purchase can wait, waiting carries some upside, since manufacturers have not yet finished passing the full cost increase through to retail pricing. If a purchase cannot wait, checking the exact RAM and storage spec against last year’s equivalent model is worth the extra five minutes, since a device with the same name and price this year may ship with meaningfully less memory than its predecessor.
Predictions: Where This Goes From Here
- Budget phone and laptop prices climb further into early 2027 as fixed-price supply contracts signed earlier in 2026 expire and get renegotiated at higher rates.
- More sub-$100 smartphone models get discontinued rather than redesigned, concentrating the entry-level market among fewer, larger brands with better supply access.
- Memory makers post stronger margins through 2027 even if unit shipment volumes stay flat, since price gains are outrunning volume growth.
- Pressure builds on regulators and industry groups in memory-dependent markets to examine allocation practices, given how directly AI accelerator demand is now colliding with consumer device affordability.
- Relief does not arrive on a fast timeline. Industry commentary, including from Intel’s own leadership, has pointed to 2028 as the earliest realistic point for meaningful easing, meaning budget hardware pricing pressure is likely a multi-year story rather than a short-term spike.
The Bigger Picture
What makes Tan’s comments notable is not that a chip executive is warning about memory prices. That has been happening all year. It is that the warning came with a specific, human-scale number attached: 70 to 80 percent of a device’s cost. That figure translates an abstract supply chain story into something a shopper can actually picture the next time they look at a phone price tag that went up $30 for no obvious reason. The AI buildout that is producing record earnings for chipmakers and cloud providers is, at the same time, quietly repricing the cheapest tier of consumer electronics on the planet. Those two stories are usually told separately. Tan’s remarks in Santa Clara tied them together in one sentence.
Frequently Asked Questions
Why did Intel’s CEO comment on memory prices instead of a memory company executive?
Intel buys memory for its own CPU platforms and partner devices, so rising memory costs affect Intel’s customers and, by extension, Intel’s business even though Intel itself does not manufacture DRAM or NAND at scale. Lip-Bu Tan’s comments at the AI Infrastructure Summit 2026 reflected that buyer-side view of the shortage.
What percentage of a budget phone’s cost is now memory?
According to Tan, memory can account for 70 to 80 percent of the total cost of a budget smartphone or laptop, up sharply from prior years when memory was a smaller share of the bill of materials.
Is this shortage the same as past DRAM price cycles?
Not entirely. Past shortages, like the ones in 2017-2018 and 2021, were driven by temporary demand spikes that manufacturers eventually outbuilt. This shortage is being sustained by long-term AI accelerator demand for high-bandwidth memory, which competes for the same fabrication capacity as consumer DRAM and NAND.
Which memory makers are most exposed to AI demand right now?
Samsung, SK Hynix and Micron are the three largest producers affected, since all three make both commodity DRAM and NAND for consumer devices as well as high-bandwidth memory for AI accelerators. Executives at all three companies have described 2026 supply as tight.
When are memory prices expected to come back down?
Industry commentary has pointed to 2028 at the earliest for meaningful relief, with some executives suggesting the tightness could persist even longer given the scale of committed AI infrastructure spending.
Should I delay buying a budget laptop or phone right now?
If the purchase is not urgent, it is worth comparing the RAM and storage specs of the current model against last year’s version before buying, since some manufacturers are quietly reducing specs rather than raising sticker prices. If the purchase is urgent, buying sooner avoids the risk of further price increases as supply contracts renew at higher rates.
Are premium phones and laptops affected too?
Yes, but less severely. Memory makes up a smaller share of a premium device’s overall cost, so price increases are easier for manufacturers to absorb without passing the full amount on to buyers or cutting specs.
