AI Servers & Systems
AI Servers & Systems covers companies that design, manufacture, or integrate complete rack-scale AI compute systems, spanning system leaders such as Dell Technologies and Super Micro Computer, electronics manufacturing services and ODM partners such as Sanmina, Celestica, Jabil, Foxconn, and Wistron that build and assemble these systems, and the distribution and component layers that support them. Dell and Super Micro Computer integrate Nvidia, AMD, and Intel silicon into branded rack and server platforms sold to hyperscalers, enterprises, and specialized infrastructure operators, while contract manufacturers such as Sanmina and Celestica have built dedicated AI-systems manufacturing relationships tied to AMD and to named hyperscale customers including Meta Platforms, Dell, and Hewlett Packard Enterprise.
Active institutional exposure flipped from a net outflow of $1.8 billion in Q1 2026 to a net inflow of $60.8 billion in Q2 2026. 61.4% of companies in this sub-theme saw increased active positions in Q2 2026.
Overview
AI Servers & Systems consists of companies that design, manufacture, or integrate complete rack-scale AI compute systems, as distinct from the underlying chip and component vendors that supply into those systems. Dell Technologies and Super Micro Computer sit at the center of this value chain as system leaders, building branded server and rack platforms that integrate GPUs from Nvidia and AMD alongside CPUs from Intel and AMD. Hewlett Packard Enterprise and Lenovo compete in the same data-center and high-performance-computing infrastructure market as system vendors. Beneath these OEMs sits a layer of electronics manufacturing services and original design manufacturer partners — including Sanmina, Celestica, Jabil, Foxconn, Wistron, Flex, Benchmark Electronics, and Plexus — that assemble, test, and package servers and racks on behalf of OEMs, chipmakers, and hyperscale customers. Distribution of finished systems into enterprise channels runs through IT distributors and resellers such as Arrow Electronics, TD SYNNEX, Ingram Micro, and CDW. A further layer of component and interconnect suppliers — including Amphenol, Astera Labs, Broadcom, Arm, and MACOM — provide connectors, interconnects, and silicon building blocks that feed into system designs but are treated as component-level enablers rather than system builders.
Investment Thesis
Investment thesis details are part of the Pro plan.
Bottleneck Tracker
Bottleneck details are part of the Pro plan.
Key Companies
Related Events
A large fintech reports about a 10x inference cost cut by defaulting to self-hosted open-weights GLM instead of commercial coding services such as Claude Code and Codex. This is a data point against frontier-lab token pricing power, and may pressure Anthropic and OpenAI coding revenue if other enterprises follow. Nvidia benefits from the on-prem hardware demand, since 56 B200 GPUs were deployed and the serving optimizations ran on NVFP4. Z.ai gains credibility for Chinese open-weights models in regulated Western enterprises, though data-residency concerns are mitigated only by on-prem or regional hosting. The model-router pattern, with vision tasks sent to Gemini or GPT, suggests a hybrid enterprise architecture with frontier models used selectively.
Brookfield and its Radiant neocloud platform plan a UK AI campus at Drax Power Station that could exceed £20 billion and 1 GW of capacity. Drax shares rose about 8.8% as investors priced in monetization of its grid-connected site. Phase 1 (up to 100 MW from Q1 2028, £2.5 billion) still needs planning, regulatory and network approvals, and a tenant, and no final investment decision has been made. The project tests whether the UK can host large-scale AI compute despite high power costs and grid connection delays, which pushed Microsoft out of the Nscale deal in October. If delivered, it would add sovereign-style UK compute capacity and demand for power, cooling, chips and construction, though execution risk remains high.
SK Hynix's warning that the memory shortage will persist through 2030, alongside record profit, signals prolonged pricing power for HBM and DRAM suppliers. Long-term agreements covering more than half of 2027-2028 capacity give the company unusual earnings visibility. Electronics makers such as smartphone, PC and console vendors face continued margin pressure and price hikes. Korean memory shares rallied, with SK Hynix up 10.9%, Samsung up 9% and the KOSPI up 6.6%. The upside is tempered by investor fears of an AI spending slowdown: SK Hynix remains about 50% below its mid-year peak, it postponed the Solidigm IPO, and the Bank of Korea flagged up to $100 billion of export downside.
TikTok USDS Joint Venture is exploring multi-gigawatt compute procurement, with one source citing a four-gigawatt target by the early 2030s. The scale adds to the global scramble for AI infrastructure and could create sizable demand for data center capacity, power and chips. Oracle, which is both a managing investor and the primary cloud provider, remains the likely anchor beneficiary, though sourcing from other providers is possible. Localized retraining of the recommendation algorithm on U.S. data and a planned B2B AI product point to rising inference and training needs. Sources are anonymous and the spokesperson only affirmed Oracle's primary role, so the size and timing of any deals remain unconfirmed.
Nvidia is moving further up the stack by selling a pre-integrated, containerized building block rather than chips or single racks. Shrinking data center deployment from six months to 14 days could speed customer capex conversion into revenue for Nvidia and its manufacturing partners. Nvidia's content per gigawatt has risen from about $18 billion to more than $40 billion, and the Pod reinforces that trend. Server ODMs and OEMs such as Foxconn, Quanta, Pegatron, Lenovo, HPE and Gigabyte gain volume but also face a tighter, Nvidia-defined reference design that may limit differentiation. Pre-integrated liquid cooling and cabling shift some value from on-site integrators and traditional data center construction toward factory-built systems, and the pods are pitched to European customers facing deployment constraints.