XBOX Project Helix (Zen 6 + RDNA 5), Nvidia RTX 5050 9GB 🤣, PS6 2027 Explained | February Loose Ends

NVIDIA 5050 “9 GB” rumor

• Evidence: the 9 GB RTX 5050 spec (referred to as “5059 GB”) appears real according to the host’s sources (a retail contact plus two other contacts). NVIDIA reportedly gave no clear explanation to partners.
• Likely technical rationale:
  • A GDDR7 9 GB part on a 96‑bit bus can be tuned to yield similar bandwidth to a 12 GB GDDR6 part on a 128‑bit bus.
  • Using 3 Gbit GDDR7 modules could produce bandwidth/performance that aligns the 9 GB SKU with existing 8–12 GB parts without upsetting the product stack.
• Broader possibilities:
  1. A wider GDDR7 refresh across the 50‑series (different capacities/Ti/Super parts) is planned.
  2. NVIDIA may be buying up GDDR7 in bulk and needs SKUs to absorb inventory or provide a release valve during transitions from GDDR6 to GDDR7.
• Conclusion:

  The 9 GB SKU looks like a last‑minute product/stacking move, driven by bandwidth/module choices and supply/contract strategy.

---

Xbox Project Helix / “Magnus” APU (analysis)

Overview

• Project Helix appears to be the hybrid Xbox APU (previously leaked as Xbox Magnus). Asha Sharma’s comments confirm a hybrid design and performance ambitions.

Leaked/spec’d hardware (summary of available docs)

• Largest console APU to date: ~408 mm², all 3 nm silicon (possible N3 variants).
• GPU: RDNA5, ~68 compute units (after cut-down), a significant step up versus Series X.
• CPU: Zen 6 cores (high‑performance cores), likely high clocks (host estimates ~5–6 GHz practical, not 7 GHz).
• RAM: options discussed around 36–48 GB total (final choice based on cost/supply).
• NPU: up to ~110 TOPS (6 W mode) or ~46 TOPS (1.2 W mode) — implies on‑device AI features.

Performance expectations

• Raster: projected roughly in the RTX 5080 ballpark or better, due to per‑CU improvements and higher CU count.
• Ray tracing: substantially improved over RDNA4; depending on clocks and optimization, could approach 5090 levels.

Positioning and price

• Hybrid PC-console angle — can run Xbox and PC games.
• Target price estimates: $900–$1,500 (host thinks $999–$1,200 most likely; $900 possible if aggressive).
• Uses the same RDNA5 chiplet AMD targets for a ~$550 GPU SKU; shared chiplet economics inform cost estimates.

Highlighted use cases

• High‑frame‑rate/PC gaming (240 Hz scenarios)
• VR
• AI features (assistant/NPU‑powered)
• Backward compatibility

---

PlayStation 6 (PS6) overview and timing

Hardware summary (restating prior leaks/specs)

• APU die: ~280 mm² (smaller than Xbox Magnus), ~54 CUs, RDNA5 GPU.
• CPU: 8 Zen 6C main cores + 2 low‑power cores for OS/background tasks.
• Target experience: TV‑centric 4K/120 Hz (not meant to be a PC alternative).
• Estimated clocks: CPU likely mid/high GHz (host estimates ~4–6 GHz range depending on configuration); GPU clock targets could yield much higher raster TFLOPS than PS5.
• Memory:
  • Home console: around 30–40 GB.
  • Handheld: 24–36 GB depending on configuration choices.
• Handheld variant specifics: lower clocks (estimated ~1.2–1.6 GHz docked vs undocked), OLED 1080p 120 Hz screen, designed to scale down games using PSSR3/upscalers.

Key feature emphasis

• Ray tracing and AI performance are expected to be the main next‑gen differentiators (not raw raster teraflops).
• PSSR3 (Sony’s upscaler) is central to Sony’s strategy.

Launch timing and RAM concerns

• Host’s view: PS6 is very likely to ship in 2027 (TSMC contracts indicate a Q2 2027 ramp) and is unlikely to be pushed to 2029.
• Rationale:
  • Huge sunk development costs and manufacturing contracts.
  • Historical precedent: PS5 launched despite RAM shortages and shipping cost spikes in 2020.
  • Expectation that RAM supply/prices will improve by late this year / next year.
• Final RAM choices will be balanced against cost and developer feedback; choices likely finalized close to manufacturing.

---

Sony upscaling (PSSR) vs AMD FSR

• Sony is building its own PSSR family rather than relying solely on AMD FSR.
• Reasons given:
  • Greater control over image and feature roadmap.
  • Faster in‑house development cadence and independence from AMD’s software schedule.
  • Marketing differentiation.
• Host notes that PSSR has quickly reached parity with AMD’s FSR4 performance in some cases (e.g., resident‑level game examples), motivating the in‑house approach.

---

RAM capacity recommendations and decision process

• Microsoft likely choices for Helix: 36 GB or 48 GB (48 GB if they want to pitch a “power PC” angle).
• Sony likely choices:
  • Home console: 30–40 GB.
  • Handheld: 24–36 GB.
• Final capacities will depend on developer feedback, devkit testing, and RAM pricing at manufacturing time.
• Host emphasis: development costs and contract timing make delays to wait for lower RAM prices unlikely; Sony will probably accept higher memory costs rather than delay a multi‑year development schedule.

---

GPU process node, wafer capacity and economics

• Many GPU makers are using 3 nm rather than immediately moving to 2 nm because:
  • TSMC capacity is heavily booked by AI customers (3 nm slots are in demand).
  • 2 nm has higher costs plus reticle/die‑stitching limitations for very large GPU dies, making it less economical for gaming GPUs.
  • Console/APU die sizes and cost tradeoffs differ from massive datacenter AI dies, making 3 nm attractive for consoles.

---

Steam Machine / small console PCs

• A Valve or third‑party small console PC (Steam Machine concept) could be competitively priced ($500–$600) and attractive if:
  • It’s upgradeable (RAM/SSD).
  • It uses modern upscaling and ray‑tracing optimizations.
  • It hits a PS5‑like performance target.
• Host views this as opportunistic hardware using cost‑effective chipsets, with a likely lifecycle of ~4–5 years if reasonably successful.

---

Other practical notes & opinions

• PSU guidance: quality 750 W PSUs have supported high‑end cards (e.g., 4090) stably for years; only extreme future GPUs might necessitate 1,000 W+.
• Laptops: mid/high‑tier laptop GPUs (50‑series) deliver strong real‑world performance and good 1440p experiences with DLSS.
• Leaks & community: FSR leaks occur and sometimes AMD employees or community interactions are plausible causes.
• Games/PC ports: rumors (e.g., Jason Schreier) suggest Sony may limit PC ports for first‑party titles; reasons include economics and a smaller PC install base capable of running heavily ray‑traced next‑gen games.

---

Reviews, guides, and related content referenced

• Host references prior deep‑dive leaks and videos:
  • Xbox Magnus leak
  • PlayStation 6 render/video leaks
  • Resident Evil 9 gameplay impressions (PS5 Pro)
  • Upcoming Die Shrink podcast (Die Shrink 172) covering Xbox leadership/strategy and technical leaks
• These are presented as validation/expanded commentary rather than step‑by‑step guides; the host encourages viewing earlier videos for full leak detail and Patreon for extended Die Shrink episodes.

---

Main speakers and sources

• Primary speaker: Tom (host of the channel “Moore’s Law Is Dead” / “Mors Law Is Dead”).
• Sources cited by the host:
  • Retail contact and two anonymous sources regarding the NVIDIA 5050 9 GB SKU.
  • An unnamed contact at NVIDIA (no explanation provided about the SKU).
  • AMD documents and internal leaks regarding RDNA5, chiplet targets, and pricing/roadmaps (e.g., chiplet targeted at ~$550).
  • Asha Sharma (head of Xbox) — referenced for confirming hybrid design and related statements.
  • Jason Schreier (rumored reporting on Sony PC port strategy).
  • Developer contacts and unspecified leaked documents (devkit details, TSMC contracts, manufacturing timelines).
• Sponsor mentioned in the stream: Bullstrap.

---

A compact spec table comparing the rumored Helix/Magnus, PS6 home, and PS6 handheld (memory, CPU/GPU cores, estimated performance/price) can be produced based on the host’s figures.