Reported by TechPowerUp on . Summarized and analyzed by XAIPART Editorial.
What happened
Intel senior vice president Cho Min-sung used the Semiconductors Powering Humanoids summit in South Korea to confirm that the next-generation "Razor Lake-AX" platform will integrate 700 TOPS of AI compute on die, measured at INT8/FP8 precision.S1 That figure represents a 3.88x jump over the current "Panther Lake" mobile parts, which Intel rates at 180 TOPS in total.S1
The compute uplift is attributed to a large integrated GPU paired with a substantial CPU complex, and the iGPU is set to ship in two Xe3-based variants: a 16-core version and a flagship with 32 Xe3 cores.S1 Earlier reports cited by the source peg the GPU die at roughly 162.84 mm², signaling a physically large package.S1 Against the Arc B390 that currently serves the Panther Lake lineup, the source describes Razor Lake-AX as nearly three times as powerful.S1
Intel also used the stage to reverse an earlier memory decision: the source reports that Intel will reintroduce on-package memory on the "AX" silicon, after previously calling Core Ultra 200V "Lunar Lake" the last design with that approach. Razor Lake-AX is said to use on-package LPDDR5X or LPDDR6.S1
| Metric | Reported value | Source |
|---|---|---|
| Razor Lake-AX on-die AI compute (INT8/FP8) | 700 TOPS | S1 |
| Panther Lake total compute | 180 TOPS | S1 |
| Reported uplift vs. Panther Lake | 3.88x | S1 |
| iGPU configuration | 16-core; flagship 32-core Xe3 | S1 |
| Reported GPU die area | 162.84 mm² | S1 |
| Uplift vs. Arc B390 | ~3x (per source wording) | S1 |
| Memory | On-package LPDDR5X or LPDDR6 | S1 |
| Process node, TDP, launch date, pricing | Not disclosed | S1 |
Background timeline
- - At the Semiconductors Powering Humanoids summit in South Korea, Intel SVP Cho Min-sung confirmed 700 TOPS for "Razor Lake-AX." [S1]
- - Source restates that Intel previously called Core Ultra 200V "Lunar Lake" the last on-package-memory design, then reverses course for the AX chip. [S1]
- - Source cites prior reporting that the iGPU comes in 16-core and 32-core Xe3 variants. [S1]
XAIPART take
Technical impact
XAIPART analysis: Routing 700 TOPS through an integrated Xe3 GPU (up to 32 cores) and a "substantial CPU complex" shifts the on-die NPU/accelerator balance; designers targeting edge AI inference should expect GPU-side INT8/FP8 paths to dominate, with the NPU role de-emphasized relative to Panther Lake.
XAIPART analysis: The return of on-package LPDDR5X or LPDDR6 - after Lunar Lake was framed as the final such design - tightens memory bandwidth and latency but removes socketed SO-DIMM flexibility, which matters for configurable industrial and embedded SKUs.
Market impact
XAIPART analysis: A 3.88x headline jump over Panther Lake and a roughly 3x lead versus Arc B390 positions Razor Lake-AX against discrete low-power AI accelerators in humanoid and edge-AI reference designs, pressuring competing mobile-SoC TOPS claims.
XAIPART analysis: The reported 162.84 mm² GPU die and on-package memory indicate a large, expensive BGA package, so procurement teams should plan for tighter allocation and longer lead times than compact Panther Lake mobile parts, especially on the 32-core flagship SKU.
Action
XAIPART analysis: For new mobile or edge-AI designs, freeze Panther Lake or Lunar Lake BOMs only if the 700 TOPS figure and on-package LPDDR5X/LPDDR6 are not on the critical path; for humanoids and coprocessor designs, hold a parallel Razor Lake-AX feasibility study until process node, TDP and launch date are disclosed.
XAIPART analysis: Buyers should pre-qualify a 16-core Xe3 variant as a hedge against 32-core flagship supply constraints, and confirm on-package memory type (LPDDR5X vs. LPDDR6) before locking firmware and validation plans.
XAIPART Forecast
XAIPART analysis: By , if Intel publishes a Razor Lake-AX 32-core Xe3 datasheet confirming at least 700 TOPS at INT8/FP8 on a sub-45W mobile envelope, XAIPART expects the platform to capture meaningful design wins in humanoid reference boards; if the published on-package memory configuration stays at LPDDR5X rather than LPDDR6, treat that signal as a delay indicator for the LPDDR6 ramp.
Facts are as reported by the cited source. The design-in, sourcing and forward-looking statements above are XAIPART's analysis, not the source's claims. Uncertainty: figures rest on a single summit-stage announcement; process node, TDP, launch window and pricing are not disclosed; the ~162.84 mm² die area and Arc B390 comparison language are paraphrased from earlier reporting cited inside the source.
Corrections: none at publication. This brief will be updated with a timestamp if the source corrects its report.
Alternativas e ciclo de vida
| Peça | Relação | Ciclo de vida | Base |
|---|---|---|---|
| — | Nenhuma alternativa registrada FALTAM DADOS | — | Requer match_type antes que qualquer rótulo de antecessor possa aparecer |
Os rótulos “predecessor” e “alternativa” vêm apenas de um tipo de correspondência registrado, de um status de ciclo de vida ou do texto da fonte. Quando nenhuma correspondência é registrada, nenhum rótulo é exibido. Nossos próprios anúncios não são referência de mercado.
Cenário competitivo
- 21764 anúncios ativos nesta categoria na XAIPART
- 88 fabricantes representados
- 57 fabricantes mencionados em nosso arquivo de notícias nos últimos 30 dias
As contagens refletem apenas o catálogo da XAIPART — não representam participação de mercado.
Peças afetadas
Nenhuma correspondência no catálogo. Esta história não corresponde a nenhuma peça atualmente listada na XAIPART. Não inferimos correspondências e não exibimos uma lista de espaço reservado.
Onde isso se encaixa na história
- intel-s-core-2000hx-plan-brings-ddr4-support-back-to-high-end-laptop-cpus
- asus-zenbook-14-lands-at-u-s-best-buy-with-snapdragon-x-plus-an-intel-core-ultra-9-386h-sku-due-q4-2026
- akasa-euler-mx2-fanless-mini-itx-case-targets-35-w-industrial-builds-without-moving-parts
- inno3d-ags-6220v2-packs-eight-rtx-pro-6000-blackwell-server-gpus-in-6u-what-it-means-for-ai-rack-buyers
- intel-nova-lake-hx-and-h-die-shots-surface-six-tile-mobile-package-mapped
- intel-razor-lake-ax-targets-700-tops-what-3-88x-panther-lake-means-for-mobile-ai-boms
- A Intel está preparando uma família móvel Core 2000HX em dies 'Raptor Lake Next', revivendo a compatibilidade com DDR4 para laptops enquanto o Nova Lake-HX assume o nome Core Ultra 4000HX. trade
- A ASUS iniciou a venda no varejo dos EUA do Zenbook 14 (UX3480QA) nas variantes de 8 GB e 16 GB, com um SKU com Intel Core Ultra 9 386H previsto para o início do 4º trimestre de 2026. trade
- A Akasa lança um chassis Mini-ITX de alumínio totalmente passivo com placa DC-DC integrada de 220 W e duas baias de 2,5", voltado a POS, quiosques e caixas de edge com CPUs Intel LGA1851 até LGA115X de até 35 W TDP. product launch trade
- A INNO3D lança o servidor rack 6U AGS-6220V2 com oito GPUs PCIe Gen 5 RTX PRO 6000 Blackwell Server Edition, voltado a frotas de treinamento e inferência de IA. product launch trade
- Fotos de die exposto do Nova Lake-HX e -H revelam um pacote móvel de seis tiles usando nós TSMC N2P, Intel 3-E e Intel 18A-P, com novas contagens de núcleos e tamanhos de pacote. trade
- A Intel afirmou que o Razor Lake-AX entregará 700 TOPS em INT8/FP8, 3,88x os 180 TOPS do Panther Lake, usando uma iGPU Xe3 de até 32 núcleos e trazendo de volta memória LPDDR5X/LPDDR6 no pacote. trade
- A Minisforum anunciou o NAS S5 sem ventoinha de 5 baias com Intel Core 3 304, com 5x M.2 NVMe, 10GbE + 2.5GbE, USB4 duplo e um runtime de Agente local sempre ativo. product launch trade
A cadeia é construída apenas com briefs publicados — um nó sem brief não é desenhado.
Pressão de custos por dimensão
| Dimension | FRED series ID | Unit | Lag | As of |
|---|---|---|---|---|
| Input cost | PCOPPUSDM, PALUMUSDM, PNICKUSDM | USD/mt | 96 d | |
| Supply/demand | PCU334413334413, IPG3344S | index(1982=100) / index(2017=100) | 46 d | |
| FX & energy | DEXCHUS, DCOILBRENTEU, DHHNGSP | rate / USD/bbl / USD/MMBtu | 6 d |
No free authoritative interface for: gold · silver · rare earth · polysilicon · lithium carbonate · market share — listed rather than approximated.
Como isso foi produzido
Evidências do processo — coleta, sanidade da fonte, modelo e filtro
| Source fetched | TechPowerUp · www.techpowerup.com · 2,057 chars · fetch status success |
|---|---|
| Source sanity (S1) | ok — passed before any model call (no s1_ block recorded) |
| Model / prompt | MiniMax-M3 · prompt Blog Agent - News Summary (v2.8-final) · 6,733 tokens · 30.6s · run completed |
| Gate | verdict pass · 6/6 applicable checks passed · spec v2.9 · C10 ratio 0.050996 · C10 rounds 0 |
| Time contract | 5 <time datetime> values in this brief — 0 violations (+08:00 or date-only) |
Cada linha acima é produzida pelo pipeline, não escrita à mão.
Comentários
Seja o primeiro a comentar.
🔐 Por favor entrar para publicar um comentário.
Não tem uma conta? Cadastre-se