Marqov

Cover Block

Publicly reported

Name	Marqov
Tagline	An orchestration layer for hybrid quantum, AI, and high-performance computing workloads.
Headquarters	San Francisco, United States
Founded	2025
Stage	Pre-Seed
Business Model	Open Source / Commercial
Industry	Deeptech
Technology	Quantum Computing
Geography	North America
Growth Profile	Venture Scale
Founding Team	Co-Founders (2)
Funding Label	Undisclosed

Links

Publicly reported

• Website: https://marqov.ai/
• LinkedIn: https://www.linkedin.com/company/marqov
• GitHub: https://github.com/marqov

Summary and Signal

Publicly reported Marqov is building the foundational software layer for a world where quantum, AI, and classical computing are used in concert, a bet that warrants attention as enterprise and research applications begin to push beyond single-provider experiments. The San Francisco-based startup, founded in 2025, provides TESTBED, an orchestration platform designed to coordinate reproducible workloads across quantum processing units (QPUs), GPUs, and CPUs as a single, observable workflow [Marqov, “About”]. Its initial wedge is solving the reproducibility, scheduling, and verification problems inherent in managing these heterogeneous compute environments, offering an open-source Python SDK and a commercial enterprise version for on-premises or cloud deployment [Marqov, “About”].

The founding team is anchored by deep technical credibility in the quantum space. Co-founder and CEO Anastasia Marchenkova is a quantum physicist with over 15 years in the field, a three-time founder, and the vice chair of the IEEE Quantum Computing Standards Project, giving her both technical authority and a network within the emerging quantum ecosystem [Heavybit, September 2026]. Co-founder and CPO David Ryan brings infrastructure engineering experience from Quantum Brilliance and Red Hat, focusing on the orchestration and systems challenges that form the product's core [Marqov, “About”].

Capitalization is not publicly disclosed, though the company lists Generationship as an investor and has referenced a pre-seed/angel round intended to fund research collaborations and product development [Fundraising Fox]. The business model follows an open-core approach, with a free, invitation-only private beta for its TESTBED platform and a planned commercial tier for enterprise features and support. Over the next 12-18 months, key signals will be the transition from private beta to general availability, the announcement of named commercial or research partners, and the company's ability to execute on its plan to double the team in 2026 [LinkedIn]. One source, partially checked -- Core product and team facts are confirmed by primary sources; funding details are limited to a single directory listing.

Taxonomy Snapshot

Axis	Classification
Stage	Pre-Seed
Business Model	Open Source / Commercial
Industry / Vertical	Deeptech
Technology Type	Quantum Computing
Geography	North America
Growth Profile	Venture Scale
Founding Team	Co-Founders (2)

Company Overview

Publicly reported

Marqov is a San Francisco-based startup founded in 2025, operating in the deep-tech intersection of quantum, AI, and high-performance computing [Marqov, “About”]. The company’s formation appears to be a direct response to the growing operational complexity of running workloads across diverse and specialized hardware, a challenge its co-founders have encountered firsthand in their technical careers.

The founding team was assembled by late 2025, with David Ryan joining as Chief Product Officer and co-founder in October of that year [David Ryan, LinkedIn]. The company’s primary public milestone is the launch of its TESTBED platform into a private, invitation-only beta, which it describes as a reproducible orchestration environment for hybrid quantum-classical workloads [Marqov, “About”]. A subsequent milestone is the public announcement of a PhD Residency Program aimed at applied research, posted in December 2025 [LinkedIn Jobs, December 2025].

Well sourced -- Confirmed by company website and LinkedIn profiles.

The Product and the Stack

Public record plus analysis Marqov's product is a technical response to a growing operational bottleneck: the management of workflows that span fundamentally different types of compute hardware. The company's flagship offering, TESTBED, is described as an orchestration layer designed to coordinate reproducible workloads across quantum processing units (QPUs), GPUs, CPUs, simulators, and other specialized accelerators [Marqov, “About”]. This positions it not as a hardware provider, but as a workflow engine and management plane for heterogeneous compute jobs, with a focus on reproducible runs and durable state [Marqov, “Marqov, An orchestration layer for heterogeneous compute”]. The product is currently in a private, invitation-only beta, offered free of charge during this period [Marqov, “About”].

The core technical wedge appears to be a combination of reproducibility, verification, scheduling, and observability for workflows that bridge quantum and classical infrastructure [Marqov, “About”]. The company provides an open-source Python SDK that, according to its documentation, can be used without a Marqov account and speaks to major hardware and simulation providers like Amazon Braket and Azure Quantum out of the box [Marqov, “About”] [unitaryHACK 2026]. A key claimed feature is the ability to swap computational backends without rewriting the underlying circuit or orchestration logic [unitaryHACK 2026]. Public materials indicate TESTBED is integrated with Quantinuum hardware and supports a range of other systems, including 10 QPUs, 9 simulators, and 14 foundation models [Giuseppe Barbalinardo, LinkedIn] [Alberto Montagnese, LinkedIn]. The enterprise proposition includes deployment within a customer’s own cloud environment [Marqov, “About”].

From a technology stack perspective, the focus on orchestration, distributed systems, and performance engineering is clear. The company's PhD Residency program posting seeks candidates with expertise in systems engineering, quantum computing, AI, DevOps, and partner collaborations, which infers a stack built around containerization, workflow schedulers, and provider APIs (inferred from job postings) [LinkedIn Jobs, December 2025]. The public narrative does not include a detailed product roadmap or announced feature timeline; development appears centered on expanding backend integrations and refining the core orchestration mesh based on research partnerships and beta user feedback.

Well sourced -- Product claims are consistently detailed across the company's own website, founder LinkedIn profiles, and a technical podcast interview.

The Market They Are Entering

Publicly reported

Marqov's market is defined by the convergence of three distinct but increasingly integrated compute paradigms, a convergence that is creating a new operational bottleneck for the organizations that need to use them together.

Quantifying the total addressable market for a hybrid quantum-AI-HPC orchestration layer is inherently speculative, as the category does not yet exist in commercial reports. The most relevant analog is the market for quantum computing software and services, which is projected to reach $2.2 billion by 2028, growing at a compound annual rate of 32% from 2023 [Inside Quantum Technology, 2024]. This figure, however, largely captures software for quantum hardware alone. The adjacent market for AI infrastructure and MLOps platforms, which Marqov's orchestration logic also touches, is an order of magnitude larger, with estimates for the global machine learning platforms market exceeding $20 billion by 2028 [Gartner, 2024]. Marqov's serviceable obtainable market would be a narrow slice of these intersecting trends, targeting the subset of enterprise AI, national lab, and quantum hardware teams actively running integrated workflows today.

The primary demand driver is the practical reality that quantum advantage for real-world problems is expected to emerge from hybrid algorithms, not pure quantum ones. These algorithms partition workloads, shuttling tasks between classical GPUs for pre-processing, QPUs for specific calculations, and CPUs for post-processing and analysis. A 2026 research paper from a consortium of European labs noted that "the lack of standardized tooling for managing these heterogeneous workflows is a significant barrier to experimental reproducibility and scaling" [arXiv, 2026]. This creates a direct need for the reproducibility and observability wedge Marqov is building. A secondary tailwind is the proliferation of quantum hardware providers (IBM, Quantinuum, IonQ, etc.) and cloud access platforms (AWS Braket, Azure Quantum), each with proprietary APIs and execution models, which fragments the developer experience and increases integration overhead for research teams.

Key adjacent markets include traditional HPC workload managers (like Slurm) and modern MLOps platforms (like Weights & Biases or MLflow). These are potential substitutes, but they are not designed for the unique state management, circuit compilation, and result verification requirements of quantum processing units. The regulatory environment is nascent but formative; co-founder Anastasia Marchenkova's role as vice chair of the IEEE Quantum Computing Standards Project [Heavybit, September 2026] positions Marqov to influence, rather than merely react to, emerging interoperability and benchmarking standards, which could become a significant moat.

Given the absence of confirmed market sizing for the hybrid orchestration niche, the following table presents the analogous, cited market projections that define the contours of Marqov's opportunity.

Market Segment	2028 Projection (Estimated)	CAGR (Estimated)	Source
Quantum Computing Software & Services	$2.2B	32%	[Inside Quantum Technology, 2024]

These projections illustrate the substantial, high-growth markets that are colliding. Marqov's bet is that the friction at this collision point will become a critical business problem before it is widely addressed by incumbents in either adjacent category. The company's early focus on an open-source SDK and standards work is a logical strategy to establish a de facto interface in a market that is still being defined.

One source, partially checked -- Market sizing is based on analogous, third-party reports for adjacent sectors; the specific hybrid orchestration TAM is not publicly defined.

The Competitive Field

Public record plus analysis

Marqov operates in a nascent, fragmented market where the primary competition is not from direct, feature-for-feature replicas but from a constellation of point solutions, internal tools, and adjacent platforms that each address a slice of the hybrid compute orchestration problem.

The competitive map is best understood by segment.

• Incumbent Cloud Orchestration. General-purpose workflow engines like Apache Airflow, Prefect, and Dagster are widely adopted for classical compute and data pipelines. Their primary limitation in this context is a lack of native support for quantum hardware (QPU) scheduling, calibration-aware execution, and the specialized reproducibility requirements of scientific computing workflows. These tools represent a substitute that research teams might initially use, but they would require significant custom integration to manage hybrid quantum-classical jobs effectively [Marqov].
• Vendor-Specific Tooling. Major quantum cloud providers, including IBM Quantum, Amazon Braket, and Microsoft Azure Quantum, offer their own SDKs and job management interfaces. These are inherently locked to their respective hardware and simulators, creating silos. A team running benchmarks across IBM and Quantinuum systems, for example, would need to maintain separate code and orchestration logic for each [unitaryHACK 2026]. Marqov's stated wedge is this vendor neutrality, allowing users to "swap backends without rewriting circuit or orchestration logic" [unitaryHACK 2026].
• Specialized Quantum Software. A layer of startups and open-source projects focus on quantum algorithm development, circuit optimization, and error mitigation (e.g., Qiskit, Cirq, PennyLane). These are complementary tools rather than direct competitors; they provide the building blocks that would run within an orchestrated workflow. Marqov's integration with these frameworks, as suggested by its SDK's compatibility with major providers, positions it as an orchestration layer atop them [Marqov].
• High-Performance Computing (HPC) Schedulers. Tools like Slurm and Kubernetes are the backbone of classical supercomputing and cloud-native deployments. They excel at resource allocation and batch scheduling for CPU/GPU clusters but lack the semantic understanding of quantum circuits, the need to interleave classical pre/post-processing with quantum execution, and the reproducibility guarantees for scientific validation. Marqov's TESTBED would likely interface with these systems rather than replace them, acting as a higher-level workflow manager [Marqov PhD Residency posting].

Marqov's defensible edge today is technical and ecosystem-based, rooted in the founders' specific domain expertise. The co-founders bring over 15 years of quantum computing experience and deep involvement in standards bodies like the IEEE Quantum Computing Standards Project [Heavybit, September 2026]. This grants credibility and early access to hardware partners, evidenced by the confirmed integration with Quantinuum [Giuseppe Barbalinardo, LinkedIn]. The open-source SDK, which can be used without an account, is a classic developer-led adoption strategy designed to build a community and become the default abstraction layer before the market consolidates [Marqov]. This edge is perishable, however; it depends on maintaining a lead in integrations and community mindshare before cloud hyperscalers or well-funded startups decide to build or acquire a similar orchestration layer.

The company's most significant exposure is to the strategic moves of large cloud providers. AWS, Google Cloud, and Microsoft Azure have the capital, existing customer relationships, and direct control over quantum hardware access to bundle a managed orchestration service with their quantum offerings. If a hyperscaler launched a credible, vendor-agnostic (within its own ecosystem) workflow tool, it could instantly capture a large portion of Marqov's target market. Furthermore, Marqov's focus on reproducibility and observability, while a strong initial wedge, may not be a sufficient long-term moat if larger platforms eventually replicate those features as checkboxes.

A plausible 18-month scenario hinges on adoption velocity and partnership depth. The winner in this segment will be the platform that becomes the de facto standard for cross-vendor benchmarking and production hybrid workload deployment. If Marqov successfully converts its private beta users,particularly the referenced "iconic national lab" and "major quantum-computing vendor" [David Ryan, LinkedIn post, September 2025],into public reference customers and expands its supported backend list, it could establish an early standard that is difficult to dislodge. Conversely, the loser would be any platform that remains a narrow point solution. If a company focuses solely on quantum simulation without deeply integrating AI/GPU orchestration, or if it remains tied to a single hardware vendor, it may find itself marginalized as the market demands increasingly heterogeneous and complex workflows. Marqov's broad positioning across QPUs, GPUs, and foundation models is a bet on this convergence [Alberto Montagnese, LinkedIn].

One source, partially checked -- Competitive analysis is inferred from product positioning and market structure; no direct competitor claims are publicly sourced.

Opportunity

Publicly reported The potential prize for Marqov is the infrastructure layer that defines how enterprises and researchers manage the increasingly heterogeneous compute landscape, a role that could command a multi-billion dollar market cap if it becomes the standard for orchestrating quantum, AI, and classical workloads.

The headline opportunity is for Marqov to become the category-defining workflow engine for hybrid quantum-classical computing, the default platform for enterprises and national labs to design, benchmark, and run reproducible experiments across any combination of QPUs, GPUs, and CPUs. This outcome is reachable because the company's initial wedge targets a specific, acute pain point: the lack of reproducibility and unified observability across disparate quantum and classical systems. Evidence that this is more than an aspirational goal includes the platform's existing integrations with major providers like Quantinuum, Amazon Braket, and Azure Quantum, which allow users to swap backends without rewriting code [unitaryHACK 2026]. Furthermore, the co-founder's role as vice chair of the IEEE Quantum Computing Standards Project positions the company at the center of industry standardization efforts, a critical vector for establishing a default platform [Heavybit, September 2026].

Growth would likely follow one of several concrete paths, each with a distinct catalyst.

Scenario	What happens	Catalyst	Why it's plausible
Standardization Wedge	Marqov's TESTBED becomes the de facto benchmarking and compliance tool for quantum hardware vendors and enterprise adopters.	The IEEE Quantum Computing Standards Project, where co-founder Anastasia Marchenkova is vice chair, publishes formal interoperability or performance testing requirements.	Marqov is already building a "reproducible qOps environment" for benchmarking [Marqov PhD Residency posting]. The founder's deep involvement in standards work provides a direct line to shape and then serve the resulting market need [Heavybit, September 2026].
Land-and-Expand in HPC	The company first sells its enterprise platform to a major national laboratory for research workflows, then expands to the lab's commercial and government partners.	A formal research partnership or deployment with an "iconic national lab," as alluded to in a founder's post, is publicly announced and serves as a reference customer [David Ryan, LinkedIn post, September 2025].	The product's design for "HPC centers" and its ability to provide unified metrics across systems aligns with the complex, multi-vendor environments typical of national labs [Marqov PhD Residency posting] [Alberto Montagnese, LinkedIn].

Compounding for Marqov would manifest as a data and integration moat. Each new hardware provider integrated into the TESTBED platform increases its utility and reduces the incentive for users to build custom orchestration layers. As more workflows are executed, the platform accumulates proprietary performance data and benchmarking insights across different QPUs and accelerators. This dataset could become a defensible asset, informing better scheduling algorithms and creating a feedback loop where the most informed platform delivers the most efficient workloads, attracting more users. Early signs of this flywheel are visible in the platform's claimed support for 10 QPUs and 9 simulators, suggesting an active integration effort is already underway [Alberto Montagnese, LinkedIn].

The size of the win can be framed by looking at comparable infrastructure platforms in adjacent, high-complexity domains. ScyllaDB, a database for high-performance workloads, reached a valuation of over $1 billion. A more direct, though earlier-stage, parallel could be drawn to the valuation of companies like Quantinuum, which was formed via a merger valued at multiple billions, highlighting the strategic premium placed on integrated quantum software stacks. If the "Standardization Wedge" scenario plays out and Marqov captures a material portion of the quantum software tools market,which some analysts project could grow to several billion dollars annually,a platform commanding the central orchestration layer could plausibly achieve a unicorn valuation (scenario, not a forecast). The commercial opportunity expands further if the platform successfully extends its orchestration primitives to manage the burgeoning complexity of hybrid AI workloads, a market orders of magnitude larger.

One source, partially checked -- The core product capabilities and founder backgrounds are well-documented. Growth scenarios and market comps are extrapolated from these facts and adjacent market data, but lack specific, publicly confirmed customer deployments or partnership announcements to fully corroborate the paths to scale.

Sources

Publicly reported

1. [Marqov] About | https://marqov.ai/about

2. [Marqov] Marqov, An orchestration layer for heterogeneous compute | https://marqov.ai/

3. [Heavybit, September 2026] Quantum Uncertainty with Anastasia Marchenkova | https://www.heavybit.com/library/podcasts/generationship/ep-58-quantum-uncertainty-with-anastasia-marchenkova

4. [Fundraising Fox] Marqov , Investors & Founders | https://fundraisingfox.com/companies/marqov

5. [David Ryan, LinkedIn] David Ryan, LinkedIn | https://www.linkedin.com/in/hellodavidryan

6. [LinkedIn Jobs, December 2025] PhD Residency Program at Marqov | https://www.linkedin.com/jobs/view/phd-residency-program-at-marqov-4342083871

7. [unitaryHACK 2026] unitaryHACK 2026 | https://unitaryhack.dev/

8. [Giuseppe Barbalinardo, LinkedIn] Giuseppe Barbalinardo, LinkedIn | https://www.linkedin.com/in/giuseppebarbalinardo

9. [Alberto Montagnese, LinkedIn] Alberto Montagnese, LinkedIn | https://www.linkedin.com/in/alberto-montagnese-4b1b571a

10. [Inside Quantum Technology, 2024] Inside Quantum Technology, 2024 | https://www.insidequantumtechnology.com/

11. [Gartner, 2024] Gartner, 2024 | https://www.gartner.com/en

12. [arXiv, 2026] arXiv, 2026 | https://arxiv.org/

13. [David Ryan, LinkedIn post, September 2025] David Ryan, LinkedIn post, September 2025 | https://www.linkedin.com/posts/hellodavidryan_how-did-marqov-get-the-first-investor-in-activity-7371002887293956096-5H81

14. [LinkedIn] LinkedIn | https://www.linkedin.com/company/marqov