Americas Technology: Hardware & Networking Report Interpretation

Overview

This conference compendium summarizes Goldman Sachs' discussions with 13 hardware and communications-technology companies at the September 8–11, 2026 Communacopia + Technology Conference. Its central conclusion is that AI infrastructure demand remains robust and is extending from hyperscalers and neoclouds into enterprise, sovereign and traditional infrastructure-refresh spending.

Core views

Goldman Sachs identifies AI-driven networking as the report's dominant sector theme. Cisco said agentic AI is increasing network-bandwidth consumption by 5x, supporting a multi-year demand cycle across hyperscale, neocloud and enterprise customers. Cisco retained fiscal 2027 guidance for 15% revenue growth at the midpoint, including $7.5 billion of hyperscale AI revenue, after securing $9.3 billion of fiscal 2026 hyperscaler AI orders. Arista is seeing strong back-end and scale-across networking demand as power, cooling and space constraints push customers to distribute clusters across locations. Celestica expects its 1.6T networking programs, beginning in 4Q26, to exceed 800G revenue during 2027; it expects co-packaged-optics adoption to become meaningful only alongside 3.2T switches, around 2029–2030. The report also argues that AI infrastructure demand is broadening beyond the earliest deployments. NetApp sees demand extending from dedicated AI storage to databases, analytics, streaming and data-infrastructure upgrades as customers move from pilots to production. Dell characterizes AI adoption as early in a multi-year infrastructure curve, citing more than 6,500 AI Factory customers and demand from enterprise and sovereign projects. HPE links inference-compute demand to data-center refreshes, token-cost optimization and data-sovereignty requirements, citing a $3.5 billion enterprise inference win. Penguin Solutions describes a land-and-expand model spanning AI Factory systems, ClusterWareAI, MemoryAI, GPU servers and managed services, with 3–5 year contracts supporting recurring revenue. Portfolio breadth, services and deployment capability are presented as competitive differentiators. Dell, HPE, SMCI and Penguin emphasize faster deployment, financing, lifecycle services and integrated systems rather than standalone components. SMCI illustrated the economic importance of deployment speed by estimating that an eight-rack installation delayed for 30 days could represent roughly $3 million of lost value. Cisco is embedding DPUs and NPUs in top-of-rack switches to integrate security into network infrastructure, while F5 is addressing AI application-delivery and runtime-security use cases; F5 reported roughly $50 million of AI-related bookings in the first half across data delivery, runtime security and AI-factory load balancing. Margins remain resilient in parts of the sector but are not uniform. Arista retained 2026 gross-margin guidance of 62%–64%, supported by pricing, supply-chain efficiencies and $9.7 billion of multiyear purchase commitments; tariff refunds contribute about 30 basis points in 2026E. Cisco expects fiscal 1Q27 gross margin of 65%–66% as hardware-heavy AI deployments weigh on mix, but management cited operating leverage, with gross margin down about two points year-on-year in fiscal 4Q26 while OpEx margin fell by almost four points. Celestica expects operating margin to rise from the low-8% range in 2Q26 to the mid-to-high-8% range over one to two years, despite memory and Tomahawk-silicon constraints. F5 retained fiscal 2027 gross-margin guidance of 80%–82%, while HP expects Personal Systems margin to trough in fiscal 4Q before recovering toward its 5%–7% long-term range in fiscal 2027. Company-specific conclusions remain differentiated. Goldman Sachs highlights Arista's EOS single-stack architecture, AI-networking demand and enterprise-campus expansion; it is Buy-rated with a $225 target based on 36x NTM+1Y EPS. Axon retained 2026 revenue-growth guidance of 32%–34%, supported by bookings, TASER upgrades and AI software adoption; it is Buy-rated with a $715 target based on 17x NTM+1Y EV/Gross Profit. NetApp retained fiscal 2027 midpoint guidance of 17% revenue growth and 22% EPS growth and is Buy-rated with a $210 target. Goldman Sachs is Neutral on Cisco, Ingram Micro and F5, while Sell-rated on HP, Super Micro and Cricut. Dell, HPE, Celestica and Penguin Solutions remain Buy-rated, reflecting their respective AI-infrastructure, integrated-solution and services opportunities.

Analysis framework

The report synthesizes management commentary from the conference, first identifying sector-wide demand, supply, margin and security themes and then applying those themes to each hosted company's operating outlook, competitive position, valuation and stated risks. Valuations are generally expressed using forward earnings or other company-specific multiples.

Methodology notes

• Industry AnalysisSupply-demand framework

  AI infrastructure demand and component-supply analysis

  The report links AI workloads, enterprise refreshes and sovereign deployments to demand for networking, servers, storage and services, while assessing how memory, silicon and other supply constraints affect pricing and margins.

• Valuation methodsP/E and PEG Valuation

  Forward P/E target-price valuation

  Goldman Sachs values several covered companies using NTM+1Y EPS multiples, including ANET at 36x, NTAP at 19x, DELL at 18x and HPE at 14x.

• Valuation methodsPS valuation

  EV/Sales component in blended fundamental/M&A valuation

  For PENG and CRCT, the report combines a fundamental P/E value with a theoretical M&A value based on NTM+1Y EV/Sales.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Arista Networks (ANET)

  AI networking and enterprise-campus beneficiary

  Strengths

  EOS single-stack architecture, AI data-center networking exposure and campus expansion.

  Weaknesses

  Deferred AI-deployment revenue may convert unevenly.

  Comparison

  Competes with Cisco, Chinese providers and white-box switching alternatives.

  Risks

  Cloud-capex slowdown, customer concentration, pricing pressure and enterprise-strategy execution.

• Cisco Systems (CSCO)

  Broad AI-networking and security beneficiary

  Strengths

  Full-stack portfolio across networking, security, observability and campus systems.

  Weaknesses

  Hardware-heavy AI mix can pressure near-term gross margin.

  Comparison

  Faces Arista, white-box and Chinese-provider competition.

  Risks

  Competition, commoditization, supply-chain costs and dilutive acquisitions.

• Dell Technologies (DELL)

  Integrated AI-infrastructure and refresh-cycle provider

  Strengths

  Scale, supplier relationships, end-to-end portfolio, services and financing.

  Weaknesses

  Exposure to component costs and cyclical PC and IT demand.

  Comparison

  Competes with white-box manufacturers in servers and storage.

  Risks

  Weaker IT and PC demand, margin pressure and structurally lower neocloud AI-server demand.

• Penguin Solutions (PENG)

  Enterprise and neocloud AI Factory provider

  Strengths

  ClusterWareAI, MemoryAI, integrated deployment services and 3–5 year contracts.

  Weaknesses

  Memory exposure introduces cyclicality.

  Comparison

  Fundamental valuation is benchmarked against DELL, HPE and NTAP.

  Risks

  Memory cyclicality, OEM competition, China exposure, supply disruptions and enterprise-AI demand.

Key data

• Cisco fiscal 2027 revenue outlook15% growth at midpointIncludes $7.5 billion of hyperscale AI revenue.
• Cisco fiscal 2026 hyperscaler AI orders$9.3 billionManagement said this increased from nearly zero two years earlier.
• Arista 2026E revenue guidance$12.6 billionRaised in 2Q26; gross-margin guidance remains 62%–64%.
• Celestica 2026 revenue-growth guidance65% year-on-yearManagement expects further acceleration in 2027.
• Dell AI Factory customersMore than 6,500Cited as evidence that enterprise AI adoption is expanding.
• HPE enterprise inference win$3.5 billionManagement described it as carrying traditional server-like margins.
• SMCI order bookRoughly $60 billionManagement cited it as a foundation for fiscal 2027 guidance.
• F5 AI-related bookingsRoughly $50 million in the first halfAcross AI data delivery, AI runtime security and AI-factory load balancing.

Impact & implications

The report indicates that the AI buildout is becoming broader and more operationally complex, favoring companies that can supply integrated networking, compute, storage, security, services and financing. It also stresses that component availability, pricing recovery, customer concentration, demand timing and execution will determine whether strong revenue momentum translates into sustained margin expansion.

Risks

• Component shortages and inflation, especially in memory and networking silicon, may pressure gross margins or delay deployments.
• Cloud and enterprise IT spending may weaken, reducing demand for AI infrastructure and refresh projects.
• Competition from incumbent vendors, Chinese providers, ODMs and white-box suppliers may pressure share and pricing.
• Customer concentration and credit exposure, particularly in neocloud markets, can increase revenue volatility.
• Hardware-heavy revenue mix and execution on services, security, campus or enterprise strategies may limit margin expansion.

What to watch

• Conversion of AI orders, backlog and deferred revenue into recognized revenue as data-center power, cooling, construction and cabling constraints are resolved.
• The pace of enterprise, sovereign and inference-oriented AI deployments relative to hyperscaler and neocloud demand.
• The transition from 400G to 800G and 1.6T networking, and the expected scale-up-Ethernet ramp from late 2027 into early 2028.
• Component availability, memory pricing and vendors' ability to pass cost inflation through pricing or product mix.
• Company guidance updates, including NetApp's INSIGHT product announcements and Ingram Micro's September 15 analyst-day financial update.