The Information Machine

NVIDIA Launches Vera CPU and Vera Rubin NVL72 at COMPUTEX / GTC Taipei · history

Version 11

2026-06-02 08:31 UTC · 237 items

What

NVIDIA's COMPUTEX 2026 keynote (June 1) produced a second NVL72 rack validation milestone within 48 hours of the first: Jensen Huang announced that Microsoft has completed bring-up of its first Rubin VR200 NVL72 rack via Foxconn as ODM [5], one day after Dell delivered the world's first fully validated NVL72 to CoreWeave [1][3]. Jensen simultaneously disclosed that wafer-level mass production of Rubin has started but rack-level mass production has not yet begun [5]. SemiAnalysis rated the keynote 'F tier,' noting no new AI datacenter products were unveiled [6], while Jensen used the event to announce a $2B NVLink Fusion stake in Marvell and publicly predict it would become the next trillion-dollar company [8].

Why it matters

Two independent ODM chains — Dell→CoreWeave and Foxconn→Microsoft — completing NVL72 bring-up within 48 hours confirms the Rubin validation ecosystem is operational across multiple operators simultaneously. The gap between wafer-level and rack-level production start [5] is now the concrete signal to watch for mass availability against the large commitment volumes already announced. SemiAnalysis's F-tier critique [6] marks the first prominent independent voice pushing back against NVIDIA's deployment momentum narrative.

Open questions

  • Rack-level mass production of Rubin VR200 NVL72 had not started as of the COMPUTEX keynote [5] — when does it begin, and does the wafer-to-rack lag create a delivery gap for the large commitment volumes announced by Microsoft, CoreWeave, and others?

  • Both CoreWeave and Microsoft have achieved single-rack bring-up (L11), but full-cluster L12 validation with scale-out networking has not been demonstrated by either [3][4][5] — when does the first L12 milestone occur, and at what cluster scale?

  • NVIDIA's $2B equity stake in Marvell via NVLink Fusion [8] pulls custom third-party silicon into NVIDIA's interconnect ecosystem — does this represent a genuine platform opening or a mechanism for NVIDIA to retain ecosystem control over alternative GPU architectures?

  • Micron's IR press release asserts high-volume HBM4 production for Vera Rubin [22] while multiple independent sources conclude NVIDIA designated only Samsung and SK Hynix as suppliers [23][24][25] — has Micron secured an actual production allocation or is the claim aspirational?

Narrative

NVIDIA's Vera Rubin NVL72 crossed its first two concrete deployment milestones in rapid succession. On May 31, Dell delivered the world's first fully validated Vera Rubin NVL72 rack to CoreWeave, with L11 diagnostics confirming the rack's internal NVLink/IMEX scale-up domain operational [1][2][3]. In NVIDIA's three-stage validation hierarchy, L10 certifies single-server firmware, L11 certifies a single rack's scale-up domain, and L12 certifies a full compute cluster with scale-out networking [4]. The following day, at COMPUTEX 2026, Jensen Huang announced that Microsoft has completed bring-up of its first Rubin VR200 NVL72 rack via Foxconn as ODM [5] — a second validation chain becoming operational within 24 hours of the first. Jensen also disclosed that wafer-level mass production of Rubin has started, while rack-level mass production has not yet begun [5], a distinction that matters for assessing when the large announced deployment volumes can actually ship.

The COMPUTEX keynote drew a sharply divided reception. SemiAnalysis graded it 'F tier,' stating Jensen announced nothing new on the AI datacenter side and that the headline announcement — Windows running on NVIDIA's ARM CPU — faces structural conditions unlike Apple's x86-to-M1 transition [6]. Jensen's own framing was expansive: he called this 'the greatest era in history to build software,' argued AI agents will create massive new software demand rather than displace existing software [7], announced a $2B NVLink Fusion investment in Marvell, and publicly predicted Marvell would become the next trillion-dollar company [8]. The Marvell partnership pulls custom AI silicon into NVIDIA's NVLink interconnect ecosystem, extending vertical integration strategy to third-party chipmakers in a way that has no precedent in NVIDIA's prior platform history.

The performance picture for Vera Rubin has two distinct layers. Phoronix benchmark results, published via NVIDIA's promotional blog, show a 1.5x overall advantage over current 128-core x86 and a 1.6x geometric mean improvement over NVIDIA's own Grace CPU [9][10]. SemiAnalysis technical analysis adds the critical nuance: Rubin GPU FP4/FP8 FLOPs scale approximately 3.5x over GB200 while FP16 gains are only ~1.6x and HBM capacity remains flat [11]. The FP16 ceiling means Rubin's efficiency headline is concentrated in low-precision inference, not uniformly distributed across training or high-precision workloads. NVIDIA's own DGX Rubin NVL8 reference system supports Intel Xeon 6 as a host CPU alongside Vera [12], implicitly acknowledging continued x86 relevance within the platform.

The infrastructure buildout surrounding Vera Rubin is structurally complex. NVIDIA holds approximately $674M in equity in Nscale [13], a cloud provider that is also its primary European deployment partner — meaning Microsoft's 130,000-GPU commitment at Start Campus Portugal and the 1.35GW Letter of Intent for Nscale's West Virginia Monarch Compute Campus [14][15] are transactions where NVIDIA functions simultaneously as chip designer, equity holder in the deploying operator, and co-party to the public announcement. Two hardware constraints establish binding ceilings on the deployment pace: HBM4 supply is dominated by SK Hynix (~70% of NVIDIA's orders) and Samsung (2026 allocation sold out) with shortage projected until 2028 and rack prices at $8.8M [16][17][18]; and the NVL72's 600kW per-rack power requirement is incompatible with most existing data center infrastructure, requiring greenfield construction [19][20]. Taiwan's manufacturing ecosystem provides the assembly backbone: more than 500 NVIDIA partners assemble over one million MGX rack components for Vera Rubin across 25 factory sites [21].

Timeline

  • 2026-01-05: NVIDIA debuts Rubin chip at CES: 336 billion transistors, 50 petaflops AI performance. [41]
  • 2026-01: Jensen Huang announces at CES 2026 keynote that Vera Rubin NVL72 is in full production. [42][43]
  • 2026-02: SK Hynix begins HBM4 mass production shipments to NVIDIA, holding approximately 70% of NVIDIA's HBM4 orders. [44][16]
  • 2026-03: NVIDIA invests $2B in Marvell as part of NVLink Fusion partnership; Jensen later predicts Marvell will become the next trillion-dollar company. [8]
  • 2026-03-17: Nscale acquires 8GW Monarch Compute Campus in West Virginia; Microsoft signs 1.35GW LOI co-announced with NVIDIA and Caterpillar. [45][46][14][47]
  • 2026-05-18: First Vera CPUs hand-delivered to OpenAI, Anthropic, and other leading AI labs. [48][49][50]
  • 2026-05-18: Jensen Huang keynotes Dell Technologies World: projects $3–4T AI infrastructure buildout by 2030 and flags memory supply as primary bottleneck. [26][51][52]
  • 2026-05-21: NVIDIA reports Q1 2026 earnings: $81.6B revenue, up 85% year-over-year. [27][53]
  • 2026-05-21: NVIDIA GTC Taipei: Vera Rubin NVL72 wins Computex Best Choice Golden Award; Meta, Google Cloud, and Microsoft formalize partnerships. [54][55][56][57][28]
  • 2026-05: Samsung sells out entire 2026 HBM4 supply; rack prices reach $8.8M; HBM4 shortage projected to persist until 2028. [36][18][37][17]
  • 2026-05: Multiple analyses document Vera Rubin NVL72's 600kW per-rack power requirement as incompatible with existing data centers, requiring greenfield construction. [19][20][38]
  • 2026-05: NVIDIA equity stake in Nscale confirmed at approximately $674M; Microsoft's Rubin GPU deployment via Nscale revised upward to 130,000 units. [40][13][15]
  • 2026-05-26: Phoronix benchmarks of Vera CPU published via NVIDIA blog: 1.5x overall x86 advantage, 1.6x geometric mean over Grace CPU, 90% peak bandwidth utilization. [9][10]
  • 2026-05-31: Dell delivers world's first fully validated Vera Rubin NVL72 rack to CoreWeave; L11 diagnostics confirm scale-up domain operational, full-cluster L12 remains ahead. [1][2][3][4]
  • 2026-06-01: Jensen Huang at COMPUTEX announces Microsoft has completed bring-up of its first Rubin VR200 NVL72 rack via Foxconn as ODM; wafer-level mass production started, rack-level mass production not yet begun. [5]
  • 2026-06-01: SemiAnalysis rates Jensen's COMPUTEX keynote 'F tier': no new AI datacenter products announced; Windows-on-NVIDIA-ARM transition unlikely to succeed. [6]
  • 2026-06-01: NVIDIA AI Cloud blog confirms CoreWeave among first Vera Rubin adopters; Taiwan ecosystem confirmed assembling 1M+ MGX rack components across 25 factory sites. [30][21]

Perspectives

NVIDIA / Jensen Huang

Maximally bullish: Q1 2026 earnings ($81.6B, +85% YoY) validate parabolic AI demand; COMPUTEX framing positions this as the greatest era in history to build software; $2B Marvell NVLink Fusion investment extends vertical integration to third-party silicon.

Evolution: Expanded: Microsoft's Rubin VR200 NVL72 bring-up announcement and the Marvell trillion-dollar prediction broaden the deployment and ecosystem narrative beyond NVIDIA's own silicon stack.

[26][27][7][5][8]

SemiAnalysis (technical analysis and keynote critique)

Technical arm: Rubin FP4/FP8 FLOPs scale ~3.5x over GB200 while FP16 gains are only ~1.6x and HBM capacity is flat — headline efficiency is workload-specific. Keynote critique: COMPUTEX was F tier with no new AI datacenter products and a Windows-on-ARM announcement unlikely to replicate Apple's transition success.

Evolution: Sharpened from pure technical analysis to an active critical voice on the deployment narrative; the F-tier assessment is the first prominent independent pushback against NVIDIA's keynote momentum framing.

[11][6][5]

Microsoft / Foxconn / Nscale

Anchor customer for Vera Rubin NVL72 via Nscale (130,000 GPUs at Start Campus Portugal, 1.35GW LOI for West Virginia) and now first hyperscaler to complete Rubin VR200 NVL72 bring-up, with Foxconn as ODM partner.

Evolution: Evolved from LOI-level commitment to physical rack validation milestone — Microsoft and Foxconn are the second actor pair to complete NVL72 bring-up, within 24 hours of Dell and CoreWeave.

[28][14][29][15][5]

Dell / CoreWeave

First-mover operators: Dell delivered the world's first fully validated Vera Rubin NVL72 rack to CoreWeave, clearing L11 diagnostics; CoreWeave is also confirmed among the first cloud providers adopting Vera Rubin and Spectrum-X Ethernet Photonics.

Evolution: Consistent since the May 31 delivery milestone; CoreWeave's expanded adoption of Spectrum-X further reinforces its first-mover positioning in the Rubin ecosystem.

[1][2][3][30]

OEM / ODM ecosystem (Dell, Supermicro, HPE, Foxconn, Compal, PEGATRON, Aivres)

Full multi-vendor tier covers both the NVL72 and HGX Rubin NVL8 form factors; Foxconn now confirmed as Microsoft's ODM for the NVL72 bring-up, making it the second ODM chain to complete rack validation alongside Dell.

Evolution: Foxconn's confirmation as an NVL72 ODM — not previously identified in that role — adds a new actor to the production execution tier.

[31][32][33][34][12][5]

Micron

Officially asserts high-volume HBM4 production specifically designed for NVIDIA Vera Rubin via investor relations press release, directly contradicting industry reports concluding NVIDIA designated only Samsung and SK Hynix as HBM4 suppliers.

Evolution: Consistent; the contradiction with multiple independent sources remains unresolved.

[22][35][23][24][25]

Memory and supply chain analysts

HBM4 shortage is the binding structural constraint: SK Hynix holds ~70% of NVIDIA's orders, Samsung has sold out its 2026 supply, rack prices have surged to $8.8M, and shortage is projected to persist until 2028.

Evolution: Consistent; rack-level mass production not yet started adds a further constraint alongside memory supply at the moment large-scale commitments are outstanding.

[16][17][36][18][37][5]

Data center infrastructure analysts

Vera Rubin NVL72's 600kW per-rack power requirement is a fundamental incompatibility with existing data center infrastructure — not a retrofit problem but a greenfield requirement establishing a second binding structural bottleneck alongside HBM4 supply.

Evolution: Consistent; NVIDIA's DSX MaxLPS tool claims 40% more GPUs within a fixed power budget, but this addresses power efficiency, not the absolute 600kW ceiling that mandates greenfield construction.

[19][20][38][30]

Tensions

  • Micron's official IR press release states high-volume HBM4 production for NVIDIA Vera Rubin [22], while TechPowerUp and multiple independent analyses conclude NVIDIA designated only Samsung and SK Hynix as HBM4 suppliers [23][24][25] — two claims that are mutually incompatible unless they refer to different allocation tiers. [22][23][24][25]
  • NVIDIA markets Vera Rubin on a 10x cost-per-token reduction, but SemiAnalysis documents FP4/FP8 gains of ~3.5x over GB200 while FP16 gains are only ~1.6x and HBM capacity is flat [11] — the efficiency claim is workload-specific, not uniform across training or high-precision inference. [11][18][19]
  • NVIDIA holds approximately $674M in Nscale equity [13] while publicly describing Nscale only as a commercial partner — making the large deployment announcements co-publicized by both companies non-arm's-length commercial transactions [14][39]. [39][40][13][14]
  • NVIDIA positions Vera CPU as the purpose-built successor to x86 in agentic AI [9][10], while NVIDIA's own DGX Rubin NVL8 supports Intel Xeon 6 as a host CPU option alongside Vera [12] — implicitly acknowledging continued x86 relevance within the platform. [9][10][12]
  • Wafer-level mass production of Rubin has started, but rack-level mass production has not yet begun [5] — creating a gap between NVIDIA's production narrative and actual rack availability at a moment when large-scale deployment commitments from Microsoft and others are outstanding. [5]
  • SemiAnalysis rated Jensen's COMPUTEX keynote 'F tier' for announcing no new AI datacenter products [6], while NVIDIA's promotional coverage treats the same event as a major milestone featuring Microsoft's NVL72 bring-up, ecosystem expansion, and AI factory blueprints [30][5] — a direct divergence on whether the keynote advanced or stalled the narrative. [6][30][5]

Sources

  1. [1] BREAKING NEWS: COREWEAVE & DELL IS THE FIRST CLOUD TO ANNOUNCE THAT THEY HAVE RUBIN VR200 NVL72 WITH FULLY PASSING L… — SemiAnalysis Twitter (2026-05-31)
  2. [2] Dell just made history this weekend and it is the culmination of an execution streak that no other company in enterprise… — Milk Road AI Twitter (2026-05-31)
  3. [3] Notably, passing L11 diags means that this rack is up and running, including the IMEX channels on the NVL72 scale-up dom… — SemiAnalysis Twitter (2026-05-31)
  4. [4] At L10 your Firmware/BIOS and OS works on a single server, at L11 a single rack or scale-up domain works, and then at L1… — SemiAnalysis Twitter (2026-05-31)
  5. [5] BREAKING NEWS: JENSEN JUST ANNOUNCED MICROSOFT HAS FINISHED BRING UP ON THEIR FIRST RUBIN VR200 NVL72 RACK with their OD… — SemiAnalysis Twitter (2026-06-01)
  6. [6] F TIER KEYNOTEMAX: Jensen ComputeX presentation was one of the worst keynotes he has done. He announced nothing new on t… — SemiAnalysis Twitter (2026-06-01)
  7. [7] Jensen Huang just said this is the greatest era in history to build software. AI agents will not kill software. They wil… — Rohan Paul Twitter (2026-06-01)
  8. [8] The CEO of NVIDIA, looked at Matt Murphy and said "The next trillion dollar company, ladies and gentlemen." (Save this). — Milk Road AI Twitter (2026-06-02)
  9. [9] NVIDIA Vera CPU Is ‘Packing a Heavy-Hitting Punch’ Against Competition — NVIDIA Blog (2026-05-26)
  10. [10] NVIDIA published a report on Vera CPU benchmarks, done by Phoronix. — Rohan Paul Twitter (2026-05-28)
  11. [11] for more details on Nvidia's VR NVL72 Oberon and future roadmap, check out our article from February: — SemiAnalysis Twitter (2026-05-31)
  12. [12] NVIDIA DGX Rubin NVL8 Supports Intel Xeon 6 as Host CPU Option for x86-Based AI Inference - StorageReview.com — reactive:nvidia-vera-computex-launch
  13. [13] UK AI Infrastructure Startup Nscale Receives $674 Million (£500 ... — reactive:nvidia-vera-computex-launch
  14. [14] Nscale acquires 8GW Monarch Compute Campus, Microsoft signs on for 1.35GW of compute - DCD — reactive:nvidia-vera-computex-launch
  15. [15] 130,000 Rubin GPUs Are Being Deployed at Nscale For Microsoft, Further Showing Massive Interest In NVIDIA's Next-Gen AI Chips — reactive:nvidia-vera-computex-launch
  16. [16] SK Hynix Secures 70% of Nvidia's HBM4 Orders - Semicon — reactive:nvidia-vera-computex-launch
  17. [17] SK Hynix Surges 15% to New High: HBM Shortage Until 2028, How Much Longer Can AI Memory King Rise? — reactive:nvidia-vera-computex-launch
  18. [18] Nvidia's memory costs soar 485%, latest AI systems now cost $7.8 ... — reactive:nvidia-vera-computex-launch
  19. [19] The Data Center Isn't Ready. NVIDIA's Vera Rubin platform ships in… — reactive:nvidia-vera-computex-launch
  20. [20] NVIDIA Vera Rubin: 600kW Racks by 2027 | Introl Blog — reactive:nvidia-vera-computex-launch
  21. [21] Taiwan’s Industry Titans Turbocharge World’s AI Infrastructure Buildout With NVIDIA — NVIDIA Blog (2026-06-01)
  22. [22] Micron in High-Volume Production of HBM4 Designed for NVIDIA ... — reactive:nvidia-vera-computex-launch
  23. [23] Micron Is Locked Out of HBM4 in NVIDIA's Vera Rubin Systems — reactive:nvidia-vera-computex-launch
  24. [24] NVIDIA to Use SK hynix and Samsung HBM4 for "Vera Rubin" Without Micron | TechPowerUp — reactive:nvidia-vera-computex-launch
  25. [25] Why Nvidia Snubbed Micron For Samsung, SK Hynix - Dailymotion — reactive:hbm-memory-supply-squeeze
  26. [26] NVIDIA CEO Jensen Huang at Dell Technologies World: ‘Demand Is Going Parabolic, Utterly Parabolic’ — NVIDIA Blog (2026-05-18)
  27. [27] NVIDIA just dropped $81.6B in Q1 revenue up 85% YoY 🤯 — reactive:nvidia-vera-computex-launch (2026-05-21)
  28. [28] Microsoft's strategic AI datacenter planning enables seamless, large ... — reactive:nvidia-vera-computex-launch
  29. [29] Nscale to Deliver 66,000+ NVIDIA Rubin GPUs to Microsoft at Start Campus' Site in Portugal — reactive:nvidia-vera-computex-launch
  30. [30] NVIDIA AI Cloud Ecosystem Expands Worldwide to Meet Global AI Compute Demand — NVIDIA Blog (2026-06-01)
  31. [31] KR5288 Rubin | 5U NVIDIA HGX™ Rubin NVL8 Server - Aivres — reactive:nvidia-vera-computex-launch
  32. [32] Compal Introduces High-Density NVIDIA HGX™ Rubin NVL8 Integrated Solution at GTC 2026 — reactive:nvidia-vera-computex-launch
  33. [33] Supermicro Reveals DCBBS® with New NVIDIA Vera Rubin NVL72 ... — reactive:nvidia-vera-computex-launch
  34. [34] PEGATRON Unveils Next Generation AI Platforms Powered by NVIDIA Vera Rubin NVL72 and NVIDIA HGX Rubin NVL8 at GTC 2026 | PEGATRON SVR — reactive:nvidia-vera-computex-launch
  35. [35] Micron Singapore - Facebook — reactive:nvidia-vera-computex-launch
  36. [36] Samsung sells out of 2026 HBM4 supply as memory resurgence ... — reactive:aws-garman-a100-demand
  37. [37] Price of Nvidia's Vera Rubin NVL72 racks skyrockets to as much as $8.8 million apiece, but server makers' margins will be tight — Nvidia is moving closer to shipping entire full-scale systems — reactive:nvidia-vera-computex-launch
  38. [38] Nvidia's Vera Rubin GPU: Redesigning Data Centres for 600kW Racks — reactive:nvidia-vera-computex-launch
  39. [39] Nvidia-Backed Nscale Plans Huge Data Center Cluster in West ... — reactive:nvidia-vera-computex-launch
  40. [40] Nvidia-backed UK AI firm Nscale raises $1.1 billion funding round — reactive:nvidia-vera-computex-launch
  41. [41] Nvidia debuts Rubin chip with 336B transistors and 50 petaflops of AI performance - SiliconANGLE — reactive:nvidia-vera-computex-launch
  42. [42] Nvidia CEO confirms Vera Rubin NVL72 is now in production — reactive:nvidia-vera-computex-launch
  43. [43] NVIDIA Vera Rubin AI Platform Hits Full Production CES 2026 ... — reactive:nvidia-vera-computex-launch
  44. [44] SK Hynix set to ship HBM4 for Nvidia's Vera Rubin this month — reactive:nvidia-vera-computex-launch
  45. [45] Nscale and Microsoft Announce Collaboration with NVIDIA and Caterpillar to Deliver 1.35GW of NVIDIA Vera Rubin NVL72 GPUs at Flagship AI Factory Campus in West Virginia — reactive:nvidia-vera-computex-launch
  46. [46] Nscale and Microsoft Announce Collaboration with NVIDIA and Caterpillar to Deliver 1.35GW of NVIDIA Vera Rubin NVL72 GPUs at Flagship AI Factory Campus in West Virginia — reactive:nvidia-vera-computex-launch
  47. [47] Nscale acquisition includes plan to build AI facility in Mason County — reactive:nvidia-vera-computex-launch
  48. [48] Vera Arrives: NVIDIA’s First CPU Built for Agents Lands at Top AI Labs — NVIDIA Blog (2026-05-18)
  49. [49] NVIDIA hand-delivers first 1.2 TB/s Vera CPUs to OpenAI, Anthropic ... — reactive:nvidia-vera-computex-launch
  50. [50] Nvidia unveils details of new 88-core Vera CPUs positioned to compete with AMD and Intel – new Vera CPU rack features 256 liquid-cooled chips that deliver up to a 6X gain in CPU throughput | Tom's Hardware — reactive:nvidia-vera-computex-launch
  51. [51] NVIDIA AI - Jensen Huang Says “Buy Dell” - LinkedIn — reactive:nvidia-vera-computex-launch
  52. [52] Jensen Huang today: Memory demand >> supply chain capacity. “Supply chain needs to be ready.” AI memory supercycle... — reactive:nvidia-vera-computex-launch (2026-05-18)
  53. [53] "Demand has gone parabolic. The reason is simple: Agentic AI has arrived." — reactive:nvidia-vera-computex-launch (2026-05-21)
  54. [54] NVIDIA GTC Taipei at COMPUTEX: Live Updates on What’s Next in AI — NVIDIA Blog (2026-05-21)
  55. [55] NVIDIA Vera Rubin NVL72 wins Computex 2026 awards for AI ... — reactive:nvidia-vera-computex-launch
  56. [56] Meta Builds AI Infrastructure With NVIDIA — reactive:nvidia-vera-computex-launch
  57. [57] NVIDIA GTC 2026: Google Cloud Deepens Partnership for AI ... — reactive:nvidia-vera-computex-launch