Beyond Code: Why Pre-IPO Investors are Chasing the "HALO" Tech Trend

Over the last decade, the classic venture capital playbook has been to seek out capital-light software-as-a-service (SaaS) companies with sky-high gross margins and predictable, recurring revenues. But a profound structural shift is rewriting the rules of the private markets. The traditional software moat is experiencing intense disruption anxiety driven by AI.

Generative AI tools are shifting the unit economics of traditional software development. Because of this, general IT and software multiples have faced downward compression, sliding to a global median of 12.1x EV/EBITDA¹. The response from the world's most sophisticated institutional allocators? Re-architecting their portfolios around a new framework coined earlier this year: HALO (Heavy Assets, Low Obsolescence).

The private tech market is rapidly transitioning into a hardware-heavy, energy-intensive industrial race. Venture-capital investments in physical AI and robotics grew by over 5x from 2019 to 2025, according to Pitchbook. This year alone, companies in those sectors have raised over $23 billion, set to far outpace last year’s record of $26 billion. Advanced computing startups, chips and quantum computing companies, saw a similar growth trajectory having already raised $20 billion this year versus $28 billion in 2025. For late-stage private market investors, the potential big opportunities are no longer just pure software plays. They are anchored in physical, real-world infrastructure that AI cannot automate away.

The Private Market Leaderboard: Heavy Asset Companies are Reshuffling Demand

The macro rotation toward heavy, asset-backed technology is explicitly visible in EquityZen’s proprietary platform data. While Artificial Intelligence retained its crown as the #1 most sought-after sector amongst investors in Q1 of 2026, the rest of the leaderboard underwent a physical reshuffling as capital shifted toward asset-heavy, capital-intensive sectors.

• The Industry Leaderboard Shift: Frontier tech and physical infrastructure took center stage, driving Aerospace to the #2 spot and pushing Manufacturing up steadily to #4 from its #6 ranking in late 2025. Meanwhile, traditional capital-light software and financial technology categories gave up ground, with Fintech dropping to the #5 spot.
• The Defense + Robotics Renaissance: Driven by national security needs and physical automation, AI-powered drone provider XTEND grew more in popularity than any other private company, leaping +334 spots quarter-over-quarter. Defense tech innovators Mach Industries (+156 spots) and autonomous vessel maker Saronic (+111 spots) also saw massive interest spikes, alongside commercial robotics platforms Gecko Robotics (+111 spots) and Miso Robotics (+102 spots).
• The Energy + Resource Squeeze: To power the computing evolution, investors are also targeting the raw materials supply chain. Machine-learning mining explorer KoBold Metals saw the second largest spike in popularity, climbing +263 spots, while battery innovator EnergyX logged a top-five popularity surge of +117 spots.
• AI Infrastructure Backend: Meanwhile, foundation model scaling drove AI data storage platform VAST Data up +85 spots. AI computing clusters are scaling at such a clip that tech giants are projected to spend over $650 billion in capital expenditures in 2026 to build out computing infrastructure. Players like VAST are growing in popularity as potential solutions to growing computing infrastructure needs.
  

The Valuation Barbell

An analysis of transactions on EquityZen’s platform in the first quarter of 2026 revealed a distinct bifurcation between the valuations assigned to HALO companies versus legacy software tech. Industrials and Traditional Tech (Software/Hardware) led the market, with 67% of companies in both sectors executing trades at a premium to a company’s last funding round. Conversely, sectors that dominated the previous market cycle are seeing more notable discounts. The Enterprise SaaS, Fintech, Healthcare, and Consumer Tech sectors saw all companies trade at a discount in Q1 of 2026.

This data illustrates that investors remain highly selective, demanding discounts for traditional software while assigning value to physical infrastructure, AI, and energy.

Private Pioneers: Building the New Physical Stack

So who are the private companies aiming to solve today’s “hard tech” problems? For investors looking to allocate capital in the private markets, the AI infrastructure landscape includes a variety of well funded companies spanning hardware accelerators, quantum mechanics, silicon photonics, and sustainable data center operations. As hyperscalers seek to build out their technologies, substantial private valuation shifts and large-scale financing rounds are redefining the sector. Here are some of the key private players attempting to alter this ecosystem.

Ayar Labs

• EquityZen Popularity Score: 87
• What They Do: Develops co-packaged optics (CPO) and optical I/O chiplets that aim to replace traditional electrical connections with light, reducing interconnect latency and power consumption in AI data centers.
• Last Funding Round: Closed a $500 million Series E funding round in March 2026 led by Neuberger Berman, valuing the company at $3.75 billion.
• Recent News: In March 2026, announced a strategic partnership with Wiwynn, a cloud IT infrastructure provider for data centers, aiming to combine Ayar Labs’ CPO solution with Wiwynn’s rack-level system design and manufacturing capabilities.
• Risks: The company faces steep operational hurdles as a result of high-volume back-end packaging, testing, and complex supply chain synchronization.

Crusoe Energy

• EquityZen Popularity Score: 88
• What They Do: Operates vertically integrated, sustainability-aligned data centers designed to leverage clean energy resources and innovative microgrids to power massive GPU clusters.
• Last Funding Round: Completed a $1.375 billion Series E round in October 2025 led by Mubadala Capital and Valor Equity Partners, securing an implied baseline valuation of $10 billion. It subsequently added $300 million in debt from Goldman Sachs in February 2026.
• Recent News: In March 2026, Crusoe announced the development of a new AI factory campus in Texas, intended to support large scale AI workloads for Microsoft.
• Risks: Capital intensity remains incredibly steep, requiring billions to fund physical facilities, while the company faces intense competition from scale-advantaged, specialized AI cloud providers like CoreWeave.

d-Matrix

• EquityZen Popularity Score: 63
• What They Do: Provides a generative AI inference compute platform utilizing Digital-In Memory Compute (DIMC) and 3D DRAM-based chiplets to break through the memory wall and accelerate inference workloads.
• Last Funding Round: Raised a $275 million Series C round in November 2025, reaching a post-money valuation of approximately $2 billion.
• Recent News: In April 2026, d-Matrix acquired GigaIO's data center business to boost its rack-scale capabilities. This acquisition gives them access to data center technologies such as SuperNODE and the FabreX PCIe-based memory fabric to accelerate the deployment of low-latency AI inference at scale.
• Risks: d-Matrix must overcome the significant hurdle of competing directly against entrenched AI chip leaders like Nvidia, requiring substantial capital to commercialize their DIMC platform.

Etched

• EquityZen Popularity Score: 65
• What They Do: Develops an ASIC accelerator called “Sohu”, designed specifically to execute transformer-based AI models rather than general-purpose compute.
• Last Funding Round: Raised $500 million in January 2026 led by Stripes and Peter Thiel, valuing the company at $5 billion.
• Recent News: Etched is actively preparing its first major chip through a manufacturing partnership with Taiwan Semiconductor’s Emerging Businesses Group.
• Risks: The company is taking a massive architectural gamble by hardwiring its silicon exclusively for transformer models; if the fundamental architecture of leading AI models shifts, their highly specialized chips could face immediate obsolescence.

Lambda

• EquityZen Popularity Score: 96
• What They Do: Operates specialized, high-density cloud infrastructure and supercomputing clusters tailored for public cloud rentals, enterprise operations, and large language model training.
• Last Funding Round: Raised $1.5 billion in a Series E round led by TWG Global in late 2025.
• Recent News: In May 2026, Lambda named former Sprint CEO Michel Combes as its new Chief Executive Officer, closing a $1 billion senior secured credit facility arranged by J.P. Morgan in the same month aimed squarely at accelerating Lambda's gigawatt-scale AI factory deployment.
• Risks: Lambda’s business is heavily dependent on the availability and pricing of GPUs, particularly from NVIDIA. This reliance exposes Lambda to supply chain risks, potential price increases from NVIDIA, and the threat of NVIDIA's own strategic decisions, such as investing in other cloud partners or developing its own direct-to-customer services.

PsiQuantum

• EquityZen Popularity Score: 97
• What They Do: Develops large-scale, fault-tolerant quantum computing systems and quantum chips, aiming to commercialize utility-scale quantum computation for advanced AI workloads.
• Last Funding Round: Secured a massive $1 billion Series E round in September 2025 led by BlackRock at a post-money valuation of $7 billion.
• Recent News: In January 2026, PsiQuantum announced a collaboration with Airbus, seeking to advance applications in aerospace for fault-tolerant quantum computers.
• Risks: Quantum computing remains a highly capital-intensive, largely unprofitable technology with prolonged development horizons that face significant execution and deployment timeline risks.

SambaNova Systems

• EquityZen Popularity Score: 94
• What They Do: Designs specialized AI chips and integrated hardware-software systems utilizing proprietary Reconfigurable Dataflow Units (RDUs), aiming to handle massive workloads optimized for enterprise and agentic AI.
• Last Funding Round: Raised a $350 million Series E round in February 2026, led by Vista Equity Partners and Cambium Capital.
• Recent News: In early 2026, the company launched the latest iteration of its ultra-fast chip tailored for agentic AI, the SN50 AI chip. SambaNova claims that scale and distribution efforts will occur in collaboration with Intel.
• Risks: The company's recent valuation reflects a drop from its peak private valuation of $5 billion in 2021, while reported acquisition talks with Intel are said to have collapsed in late 2025.

SiFive

• EquityZen Popularity Score: 56
• What They Do: Acts as a supplier of high-performance RISC-V processor IP and custom silicon solutions, building open-source instruction set architectures for edge AI, automotive, and data center processors.
• Last Funding Round: Raised $400 million in an oversubscribed Series G in April 2026, valuing the company at $3.65 billion.
• Recent News: In May 2026, SiFive launched its third-generation Performance P570 out-of-order processor IP. Earlier in the year, the company announced it is integrating NVIDIA NVLink Fusion into its data center platform to enable coherent connectivity with NVIDIA GPUs.
• Risks: As a pure IP provider, SiFive's growth depends heavily on third-party silicon designers adopting RISC-V at scale across data centers before proprietary architectures entirely lock up the market.

Tenstorrent

• EquityZen Popularity Score: 86
• What They Do: Tenstorrent specializes in AI processors created to enhance algorithm training and adaptability. Their processors utilize software-programmable application-specific integrated circuit chips. The company aims to lead a new era in AI and deep learning through their architecture and software.
• Last Funding Round: Raised a $800 million Series E in November 2025, at a $3.2 billion post-money valuation.
• Recent News: In May 2026, the company demonstrated production-ready LLM inference on its Tenstorrent Galaxy architecture, which aims to run any model, giving users flexibility.
• Risks: As an alternative AI chip designer utilizing open-source architecture, Tenstorrent faces intense competition and must successfully scale its custom architecture against both legacy incumbents and well-funded start-ups.

The Investor Playbook: Navigating HALO Opportunities

For private market secondary investors looking to allocate capital, the emergence of the HALO framework requires a shift in the vetting process:

• Audit for Real-World Moats: When evaluating a late-stage company, look past top-line growth. Ask yourself: Does this company own a specialized hardware layer, a custom physical deployment infrastructure, a regulatory moat, or a unique data loop that cannot be replicated by a generic large language model or deemed obsolete?
• Capitalize on Corporate Capex Tailwinds: Companies focusing on specialized compute factories and modular physical deployments are scaling rapidly because their business models plug directly into the multi-hundred-billion-dollar infrastructure buildout required by global tech firms. Backing the essential "pick-and-shovel" providers of this physical landscape may create a stable market opportunity.
• Position for Multi-Channel Exits: Traditional IPO execution remains highly selective, but corporate cash buyouts and strategic private-equity integrations are showing they can deliver high-valuation exit events. Late-stage companies that own their entire operational stack are uniquely positioned to command maximum value from incoming corporate buyers looking for turn-key scale.

The era of tech investing relying solely on capital-light code is taking a breather. In 2026, the physical, asset-backed foundation is proving its value. By incorporating hard tech, specialized compute infrastructure, and energy-innovative private companies into a diversified portfolio, individual investors can align their portfolios with the structural forces reshaping the future of technology.

To explore current opportunities in Hard Tech, AI Infrastructure, and Energy, visit EquityZen.