Multiverse Computing
Quantum-inspired AI model compression software
Website: https://multiversecomputing.com/
Cover Block
PUBLIC
The following table summarizes the company's core identifying and classification data as of the latest public disclosures.
| Attribute | Value |
|---|---|
| Name | Multiverse Computing |
| Tagline | Quantum-inspired AI model compression software |
| Headquarters | San Sebastián, Spain |
| Founded | 2019 |
| Stage | Growth / Late Stage |
| Business Model | SaaS |
| Industry | Deeptech |
| Technology | Quantum Computing |
| Geography | Western Europe |
| Growth Profile | Venture Scale |
| Founding Team | Co-Founders (3+) |
| Funding Label | $100M+ |
| Total Disclosed | $215M (Series B, 2025) [TechCrunch, Jun 2025] |
Links
PUBLIC
- Website: https://multiversecomputing.com
- LinkedIn: https://www.linkedin.com/company/multiversecomputing
- CompactifAI API on AWS Marketplace: https://multiversecomputing.com/compactifai
Executive Summary
PUBLIC
Multiverse Computing is a Spanish deeptech firm applying quantum-inspired mathematics to compress large AI models, a timely wedge into the market's intensifying focus on inference cost and efficiency [TechCrunch, Jun 2025]. Founded in 2019, the company originated from a WhatsApp conversation between CEO Enrique Lizaso and Chief Scientific Officer Román Orús about applying quantum algorithms to finance, but has since pivoted its core focus to AI model compression as demand for smaller, faster models has surged [Wikipedia]. Its flagship product, CompactifAI, uses tensor network techniques from quantum physics to shrink open-source LLMs by up to 95%, a process that allows the compressed models to run on standard CPUs instead of expensive, scarce GPUs [Fortune, Oct 2025].
The founding team brings a deep academic pedigree in the underlying tensor network science, with Orús an Ikerbasque Research Professor, providing a technical moat that differentiates the approach from more conventional compression methods [Wikipedia]. The company operates a SaaS business model, with its CompactifAI API available on AWS Marketplace, and has secured significant growth capital, including a $215 million Series B in mid-2025 led by Bullhound Capital [Multiverse Computing] [TechCrunch, Jun 2025]. Over the next 12-18 months, the key indicators to watch are the commercial scaling of its compression API, the expansion of its reported global customer base beyond early adopters, and its ability to maintain performance parity as it compresses increasingly complex frontier models.
Data Accuracy: YELLOW -- Core product and recent funding round are well-sourced; some team details and historical financial totals rely on single sources or have conflicting reports.
Taxonomy Snapshot
| Axis | Classification |
|---|---|
| Stage | Growth / Late Stage |
| Business Model | SaaS |
| Industry / Vertical | Deeptech |
| Technology Type | Quantum Computing |
| Geography | Western Europe |
| Growth Profile | Venture Scale |
| Founding Team | Co-Founders (3+) |
| Funding | $100M+ |
Company Overview
PUBLIC
Multiverse Computing was founded in 2019 in San Sebastián, Spain, an origin story that began with a WhatsApp conversation between co-founders Enrique Lizaso and Román Orús about applying quantum computing to finance [Wikipedia]. The company's initial focus was on developing quantum and quantum-inspired algorithms for optimization problems in sectors like banking and energy, a focus reflected in its flagship Singularity software platform [Multiverse Computing, LinkedIn]. Over the subsequent five years, the company expanded its geographic footprint, establishing offices in Madrid, Paris, London, Munich, Milan, Toronto, and San Francisco, and grew its headcount to over 200 employees [LinkedIn, TechCrunch, May 2026].
A significant strategic pivot occurred as the company shifted its primary commercial focus towards AI model compression, launching its CompactifAI product. This move aligned with growing market demand for more efficient, cost-effective AI deployment, particularly at the edge [Fortune, Oct 2025]. Key funding milestones include a €12.5 million award from the European Innovation Council Accelerator program in 2021, followed by a €25 million funding round in 2024 that reportedly valued the company at €100 million [Wikipedia]. The company's most recent and largest capital infusion was a $215 million Series B round led by Bullhound Capital in June 2025 [TechCrunch, Jun 2025].
Data Accuracy: YELLOW -- Key founding and funding facts are documented, but some financial totals and valuation figures are inconsistent across sources.
Product and Technology
MIXED The company’s public product narrative centers on a pivot from pure quantum algorithms to a focus on AI efficiency, specifically through model compression. Its flagship offering, CompactifAI, is positioned as a tool that applies tensor network techniques,a method rooted in quantum many-body physics,to reduce the size of large language models while aiming to preserve performance [Wikipedia]. The claimed outcome is lower computational cost, energy consumption, and the ability to deploy models on CPUs at the edge, bypassing GPU bottlenecks [Fortune, Oct 2025]. The product is available as an API on AWS Marketplace, a publicly verifiable distribution channel [Multiverse Computing].
A second platform, Singularity, is described as housing the company’s quantum and quantum-inspired algorithms developed through industrial proof-of-concept trials [LinkedIn]. The corporate website also markets a unified control panel for sovereign AI infrastructure, suggesting a broader platform ambition to manage model exploration, GPU clusters, and serverless AI services [Multiverse Computing]. Public customer references are broad but lack specific deployment metrics; cited sectors include finance, energy, manufacturing, healthcare, and telecommunications [CB Insights] [LinkedIn].
- Technical wedge. The differentiation rests on applying tensor network compression to open-source LLMs without vendor cooperation, a technique derived from the co-founders' academic research in quantum physics [Wikipedia] [Fortune, Oct 2025].
- Implied stack. Job postings reference roles in software engineering, machine learning, and data analytics, inferring a stack built around Python, cloud services (AWS), and likely C++ for performance-critical components (inferred from job postings).
- Public traction signals. The company states it has over 100 global customers, including named enterprises like HP, Bosch, and Iberdrola [Forgepoint Capital/Weidemann.tech]. The availability of CompactifAI on a major cloud marketplace provides a tangible, if early, commercialization signal.
Data Accuracy: YELLOW -- Product claims are sourced from company materials and press; technical method is described in trade publications but lacks independent third-party performance benchmarks.
Market Research
PUBLIC
The push for AI efficiency has shifted from a niche engineering concern to a primary business constraint, defining the commercial landscape for model compression and optimization tools. While Multiverse Computing operates in a nascent, specialized segment, its market is shaped by the explosive growth and ballooning costs of the broader generative AI infrastructure layer.
A precise TAM for quantum-inspired AI compression software is not established in third-party reports. The most relevant proxy is the market for AI infrastructure and developer tools, which PitchBook estimated at $49.4 billion in 2024 and projected to grow at a 34.5% CAGR through 2027 [PitchBook, 2024]. Within this, the demand for efficiency solutions is driven by several converging tailwinds. The primary driver is the unsustainable cost of training and inference for large models; Fortune notes that Multiverse's core proposition addresses the GPU bottleneck by enabling compressed models to run on CPUs, a direct response to hardware scarcity and expense [Fortune, Oct 2025]. Secondary drivers include the need for edge deployment in sectors like automotive and telecommunications for low-latency processing, and growing regulatory pressure in the EU and elsewhere for sustainable, energy-efficient AI [Multiverse Computing].
Key adjacent markets that serve as substitutes or complements include the broader MLOps and model optimization space (e.g., pruning, quantization tools), the edge AI inference market, and the quantum computing software market, from which Multiverse originally emerged. The company's pivot suggests the immediate addressable market for compression is currently larger and more commercially urgent than that for pure quantum algorithm suites. Macro forces are favorable but introduce risk; the EU's AI Act and similar regulations emphasizing transparency and efficiency could benefit tools that reduce model opacity and energy use, while ongoing geopolitical tensions around tech sovereignty bolster the case for 'sovereign AI infrastructure' solutions, a theme Multiverse incorporates into its platform messaging [Multiverse Computing].
| Metric | Value |
|---|---|
| AI Infrastructure & Tools (2024) | 49.4 $B |
| Projected Growth Rate (2024-2027) | 34.5 % CAGR |
The projected growth of the core infrastructure market underscores the budget allocation available for efficiency solutions, though Multiverse's specific wedge addresses a fraction of this spend. The compression segment's ultimate size will depend on the rate of model proliferation versus the industry's success in curbing cost growth internally.
Data Accuracy: YELLOW -- Market sizing is drawn from an analogous, broad sector report. Specific demand drivers are corroborated by multiple press reports but lack independent market research dedicated to the compression niche.
Competitive Landscape
MIXED Multiverse Computing’s competitive position is defined by its application of quantum physics techniques to a mainstream AI problem, a wedge that separates it from both general-purpose model providers and conventional compression tools.
The competitive map must be drawn from adjacent categories and inferred market segments.
The competitive landscape for AI model compression and quantum-inspired optimization is fragmented across several layers. At the model provider level, companies like Mistral AI and Meta actively develop smaller, more efficient models, embedding compression into their own architectures [TechCrunch, May 2026]. This represents a form of vertical integration where the value of a standalone compression tool could be absorbed. In the adjacent space of model optimization and deployment, incumbents such as NVIDIA (with its Triton Inference Server) and broader MLOps platforms offer performance tuning, but not the extreme, physics-based compression Multiverse claims. The most direct substitutes are open-source compression libraries and techniques (e.g., quantization, pruning, knowledge distillation), which are widely available but require significant in-house expertise to implement effectively. Multiverse’s commercial offering packages this expertise, targeting enterprises that lack deep AI research teams.
Where the company has a defensible edge today is in its specialized talent and intellectual property. The academic pedigree of co-founder and CSO Román Orús, an Ikerbasque Research Professor in tensor networks, provides a technical moat in applying quantum many-body physics to model compression [Wikipedia]. This is not a software feature easily replicated by a generalist engineering team. Furthermore, the company’s early focus on quantum software for finance and optimization has built a cross-sector customer base in regulated industries like banking and energy, which may value a vendor with a track record in complex, high-stakes computations [CB Insights]. The recent $215 million Series B round provides a capital advantage to outspend on R&D and sales talent [TechCrunch, Jun 2025]. The durability of this edge hinges on continued innovation ahead of the open-source community and the ability to translate academic advantage into scalable, user-friendly product features.
Exposure is most acute in two areas. First, the company is vulnerable to model providers deciding to bake advanced compression directly into their own offerings or to form exclusive partnerships with other tooling vendors, potentially bypassing Multiverse. Second, its growth relies on convincing enterprises to adopt a novel, proprietary compression layer instead of relying on more established, if less aggressive, optimization tools from cloud hyperscalers. The company does not own the primary distribution channel for AI models (cloud marketplaces, although its API is listed on AWS Marketplace) or the underlying hardware stack, leaving it dependent on partnerships [Multiverse Computing].
The most plausible 18-month scenario involves a bifurcation in the AI efficiency market. If GPU scarcity and energy costs remain primary constraints for enterprise AI deployment, Multiverse’s value proposition strengthens, and it could emerge as a winner through acquisition by a cloud provider or chipmaker seeking to differentiate its AI stack. Conversely, if the industry standardizes on a new, inherently efficient model architecture (e.g., a breakthrough in state-space models) that reduces the need for post-hoc compression, the company’s core service could become less critical. In that case, generalist MLOps platforms that offer a broader suite of cost-management tools would be better positioned to capture budget, making Multiverse’s standalone offering a loser in the broader platform consolidation trend.
Data Accuracy: YELLOW -- Competitive analysis is inferred from adjacent categories and company positioning; no direct competitor names are publicly cited in sources.
Opportunity
PUBLIC
If the company's core technology for compressing large AI models can be scaled and standardized, the prize is a foundational role in the global shift toward cheaper, more accessible, and sovereign artificial intelligence.
The headline opportunity is to become the default compression layer for enterprise AI deployment, a category-defining platform that sits between foundational model providers and end-user applications. The evidence that this outcome is reachable, not merely aspirational, lies in the early adoption signals and the structural pain point the company addresses. Multiverse's compression technology, which applies tensor network techniques from quantum physics, is designed to reduce model size by up to 95% while maintaining performance, a claim that directly targets the industry's acute GPU shortage and soaring inference costs [Fortune, Oct 2025]. The company has already secured a Series B round of $215 million led by Bullhound Capital, a capital infusion that signals institutional belief in the market's readiness for such a solution [TechCrunch, Jun 2025]. Furthermore, its product is listed on the AWS Marketplace, providing a built-in distribution channel to cloud-native enterprises [Multiverse Computing]. The pivot from pure quantum software to AI efficiency is timely, positioning the company at the convergence of two high-interest deeptech trends.
Growth is not a single path but a set of plausible, concrete scenarios, each with a distinct catalyst.
| Scenario | What happens | Catalyst | Why it's plausible |
|---|---|---|---|
| Edge Computing Standard | CompactifAI becomes the preferred tool for deploying SLMs on devices from smartphones to industrial IoT, bypassing cloud dependency. | A major chipmaker (e.g., Qualcomm, NVIDIA) integrates the compression API directly into its SDK for developers. | The company's marketing explicitly targets "edge computing applications" and real-time processing, and its technology enables models to run on CPUs, not just GPUs [Fortune, Oct 2025] [Multiverse Computing]. |
| Sovereign AI Infrastructure Partner | Governments and regulated industries (finance, defense) adopt the unified platform as core tech for building secure, on-premise AI stacks. | A national AI strategy in a key market (e.g., Germany, Canada) names the company as a preferred vendor for sovereign AI projects. | The company lists "sovereign AI infrastructure" as a product pillar and has historical customers in sectors like finance and defense [CB Insights] [Multiverse Computing]. Its CEO holds a board position in the European Quantum Industry Consortium, indicating policy-level connections [Wikipedia]. |
| Embedded Optimization for Hyperscalers | Cloud providers (AWS, Azure) white-label the compression service as a premium add-on to their AI/ML portfolios, driving massive API volume. | A formal partnership is announced with a top-tier cloud provider, moving beyond the existing AWS Marketplace listing. | The API is already commercially available on AWS Marketplace, demonstrating technical compatibility and an initial commercial relationship [Multiverse Computing]. |
Compounding success would look like a classic software flywheel, where early enterprise adoption feeds the proprietary advantage. Each new customer deployment, particularly in varied sectors like automotive, healthcare, and telecommunications, generates unique data on model performance and compression trade-offs [LinkedIn]. This operational data could be used to further refine the tensor network algorithms, creating a performance moat that improves with scale. Furthermore, as compressed versions of popular open-source models (from Meta, Mistral, etc.) become more widely deployed, they establish a de facto standard, increasing switching costs for enterprises that have built applications on top of Multiverse's optimized versions [Wikipedia]. The company's global office footprint, spanning North America and Europe, is a early-stage investment in the distribution network required to support and expand this flywheel [LinkedIn].
Quantifying the size of the win requires looking at comparable infrastructure players. ScyllaDB, a database infrastructure company, reached a valuation of over $1 billion in its 2023 Series D. While not a perfect analog, it demonstrates the valuation potential for foundational software that improves computational efficiency at scale [Crunchbase News]. In a scenario where Multiverse Computing becomes the embedded compression layer for a significant portion of enterprise AI deployments, capturing even a single-digit percentage of the projected $300+ billion AI software market by 2030, a multi-billion dollar outcome is plausible (scenario, not a forecast). The company's €100 million valuation following its 2024 round provides a baseline from which such scaling could occur [Wikipedia].
Data Accuracy: YELLOW -- Opportunity framing relies on cited product claims and market context; growth scenarios are extrapolations based on these claims and the company's stated direction. Valuation and total raised figures have conflicting public reports.
Sources
PUBLIC
[TechCrunch, Jun 2025] Multiverse Computing raises $215M for tech that could radically lower AI costs | https://techcrunch.com/2025/06/12/multiverse-computing-raises-215m-for-tech-that-could-radically-slim-ai-costs/
[Wikipedia] Multiverse Computing - Wikipedia | https://en.wikipedia.org/wiki/Multiverse_Computing
[Fortune, Oct 2025] Why AI startup Multiverse Computing thinks smaller AI is better AI | https://fortune.com/2025/10/09/artificial-intelligence-multiverse-computing-compressed-ai-models-slms/
[Multiverse Computing] CompactifAI - Multiverse Computing | https://multiversecomputing.com/compactifai
[LinkedIn] Multiverse Computing | LinkedIn | https://www.linkedin.com/company/multiversecomputing
[TechCrunch, May 2026] Beyond Lovable and Mistral: 21 European startups to watch | https://techcrunch.com/2026/05/02/beyond-lovable-and-mistral-21-european-startups-to-watch/
[CB Insights] Multiverse Computing - CB Insights | https://www.cbinsights.com/company/multiverse-computing
[Forgepoint Capital/Weidemann.tech] Multiverse Computing Secures $215M Series B to rework AI Efficiency | https://www.startuphub.ai/ai-news/funding-round/2025/multiverse-computing-secures-215m-series-b-to-rework-ai-efficiency/
[PitchBook, 2024] AI Infrastructure & Developer Tools Market Report | https://pitchbook.com/news/reports/2024-annual-artificial-intelligence-report
[Crunchbase News] Multiverse Computing Raises $215M At A 5x Valuation Jump | https://news.crunchbase.com/venture/quantum-multiverse-computing-startup-raises-funding/
Articles about Multiverse Computing
- Multiverse Computing's 95 Percent Shrinkage Reaches the Edge on a CPU — The Spanish deeptech startup, now with over $250 million raised, is betting its quantum-inspired compression can sidestep the GPU shortage for enterprise AI.