Samsung has reportedly asked suppliers to cut Q4 phone production by 20% to 30%, taking 2026 output from a planned 270 million toward 200 million. Memory prices have climbed far enough that some Galaxy models barely make money, and Samsung would sooner build fewer of them.
Perspective Coverage
7 publishers
- Builder
- Builder 10%
- Operator
- Operator 46%
- Investor
- Investor 44%
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+30
- Incentives50
- Confidence45
buildOne report1 publisher Dynatron cooler listing puts AMD's Verano AI host CPU on a smaller SB1 socket
AMD's EPYC Verano AI host CPU will use a new SB1 socket, smaller than SP7 and SP8, according to a listing at cooler maker Dynatron. Verano gets its own board, while the cooler listed for it copies an SP7 part's specs and leaves the power budget unknown.
Reality
- Evidence38
- Adoption
- Not scored
- Hype gap+8
- Incentives
- Not scored
- Confidence45
The $375m Series C priced Positron at $3.5bn before the money, and the up-to-$500m follow-on has to be sitting at $4.5bn pre-money to reach the $5bn headline, which only holds if every dollar of it is drawn.
Perspective Coverage
4 publishers
- Builder
- Builder 34%
- Operator
- Operator 21%
- Investor
- Investor 45%
Reality
- Evidence55
- Adoption30
- Hype gap+35
- Incentives70
- Confidence60
CXMT pitches new DRAM platform to electronics makers, but only CXMT has measured it
China's largest DRAM maker has two 24-gigabit LPDDR5X parts in mass production and says the platform behind them matches the industry's most advanced nodes. The only measurements behind that claim are CXMT's own.
Reality
- Evidence45
- Adoption32
- Hype gap+38
- Incentives78
- Confidence55
Apple now charges about $44 a gigabyte for the Mac Studio's local model memory
A maxed 256GB Mac Studio has gone from $7,499 to $11,299 in 18 months, and the reported rise in LPDDR5X prices explains under a third of that. Anyone refreshing a local-inference box needs both numbers.
Reality
- Evidence45
- Adoption20
- Hype gap+12
- Incentives55
- Confidence45
Changxin Memory says its fifth-generation platform is in mass production at 11.95nm with at least 50 per cent more dies per wafer. Both of the figures that would let a buyer price it come from the company.
Perspective Coverage
3 publishers
- Builder
- Builder 38%
- Operator
- Operator 19%
- Investor
- Investor 43%
Reality
- Evidence40
- Adoption45
- Hype gap+35
- Incentives70
- Confidence58
CXMT moved LPDDR5X from mass production to a flagship handset in about 13 months
ChangXin Memory has announced 24-Gbit LPDDR5X products. Its own timeline runs from LPDDR5X mass production in May 2025 to LPDDR6 mass production by September 2026, on a target of about 300,000 wafers a month.
Reality
- Evidence34
- Adoption60
- Hype gap+32
- Incentives72
- Confidence44
buildConfirmed2 publishers Positron says the money covers a TSMC N3P tapeout by the end of 2026, a new 2MW engineering data center and a production ramp in the second half of 2027. The larger of the round's two tranches is described as up to $500 million.
Reality
- Evidence55
- Adoption35
- Hype gap+32
- Incentives72
- Confidence62
Investors quintupled Positron's valuation months before Asimov's tapeout
The $875 million round prices an inference chip whose performance figures come from cycle-accurate simulations and whose production date has moved from early 2027 into the second half. The system Positron ships today runs on HBM.
Reality
- Evidence42
- Adoption38
- Hype gap+45
- Incentives80
- Confidence52
Positron raises $875M for an inference chip built around commodity LPDDR5X instead of HBM
The Reno startup says its Asimov ASIC carries 288 GB of on-package LPDDR5X and realizes more than 90% of its memory bandwidth. Both figures come from the company, and no independent benchmark accompanies them.
Reality
- Evidence26
- Adoption5
- Hype gap+58
- Incentives74
- Confidence54
Positron bets $875m that smartphone memory can carry inference
Positron is now worth $5 billion on a design that swaps HBM for the memory that ships in phones, but Asimov has not taped out, mass production is set for the second half of 2027, and the 26x tokens-per-dollar figure comes from simulation.
Reality
- Evidence28
- Adoption4
- Hype gap+58
- Incentives78
- Confidence55
Apple debuted its first 2nm chip in an $899 Mac mini instead of an iPhone
The M6 reached TSMC's N2 process before the A20 Pro did, pinning down what the iPhone 18 Pro's chip is built on. Its memory bandwidth number tells on-device AI teams how much room next year's phone actually adds.
Reality
- Evidence58
- Adoption30
- Hype gap+12
- Incentives62
- Confidence55
buildOne report1 publisher Samsung puts MAC trees in every LPDDR5X bank because HBM costs too much
At Hot Chips 2026 Samsung detailed a validated LPDDR5X-PIM part claiming 614 GB/s of internal bandwidth and 2.28x to 3.01x inference gains over plain LPDDR5X.
Reality
- Evidence52
- Adoption18
- Hype gap+34
- Incentives74
- Confidence55
buildOne report1 publisher Intel's Crescent Island bets 480GB of LPDDR5X and 350W beats paying for a liquid retrofit
Intel's Hot Chips disclosure details 32 Xe Cores, 32MB of L2 and 16-deep matrix engines. The figure it does not include is the one that decides whether the capacity is servable.
Reality
- Evidence58
- Adoption
- Not scored
- Hype gap+30
- Incentives68
- Confidence52
buildOne report1 publisher Nvidia's 88-core Vera bets agentic serving runs out of bandwidth before it runs out of cores
One SKU, one compute die, 1.2 TB/s of LPDDR5X, and benchmark wins over a 96-core EPYC. The comparison with AMD's Venice is being set on Nvidia's terms rather than on core counts.
Reality
- Evidence38
- Adoption12
- Hype gap+33
- Incentives78
- Confidence44