Hiverge
Hiverge builds the Hive, an AI program synthesis platform that automatically discovers and optimizes algorithms for AI training, industrial planning, and quantum chemistry. Founded in Cambridge, UK by ex-Google DeepMind researchers behind AlphaTensor and FunSearch, it serves enterprise customers via an early-access waitlist.
- Company typePrivate
- Founded2024
- HeadquartersCambridge, United Kingdom
- Headcount1–10
- GTM typeB2B
- OfferingSoftware
What Hiverge does
Hiverge is a Cambridge, United Kingdom-based artificial intelligence company founded in April 2024 by former Google DeepMind researchers Alhussein Fawzi, Bernardino Romera-Paredes, and Hamza Fawzi — the team behind AlphaTensor, FunSearch, and early-stage AlphaEvolve. The company has built the Hive, an AI-driven program synthesis platform that combines large language models with advanced search algorithms to automatically discover and optimize algorithms for complex optimization problems. The Hive deploys multiple agents in parallel that propose, evaluate, and evolve code against concrete performance metrics, producing human-interpretable, deployable algorithms that can span hundreds to over a thousand lines of code, rather than opaque black-box solutions.
The platform has demonstrated quantifiable results across multiple domains: over 20% reduction in CIFAR-10 image classification training time (achieving a 1.99-second training run on a single NVIDIA A100 GPU), approximately 1.25% improvement on GPT-2 training on top of prior state-of-the-art, up to 70% reduction in two-qubit gate requirements for quantum chemistry algorithms compared to state-of-the-art VQE, and planning solutions approximately 1,000 times faster than existing approaches for Airbus operations — winning first place in both scalability tracks of the Airbus Beluga AI Challenge. Some discovered optimizations have shown cross-domain generalization, with techniques developed for CIFAR-10 (e.g., optimizer vectorization) transferring to GPT-2 training.
Hiverge operates a waitlist-based early access go-to-market strategy targeting selected enterprise customers, with pricing not publicly disclosed and revenue model inferred as subscription or usage-based platform access augmented by custom algorithm development services. The company raised $5 million in seed funding in September 2025 led by Flying Fish Ventures with participation from Jeff Dean and Ahren Innovation Capital, among others. Its target segments span AI/ML organizations seeking training acceleration, industrial enterprises with complex planning and logistics problems, and quantum computing researchers — the latter segment expanded via strategic partnerships with Quantinuum (InQuanto), NVIDIA (CUDA-Q, Grace Blackwell), and AWS (cloud infrastructure for simulation), with planned broader distribution through AWS and Google Cloud marketplaces.
Hiverge firmographics
Firmographics- Name
- Hiverge
- Legal name
- Hiverge
- Website
- https://hiverge.ai
- Company type
- Private
- Founded year
- 2024
- Operating status
- Operating
- Headcount range
- 1–10 employees
- Short description
- Hiverge builds the Hive, an AI program synthesis platform that automatically discovers and optimizes algorithms for AI training, industrial planning, and quantum chemistry. Founded in Cambridge, UK by ex-Google DeepMind researchers behind AlphaTensor and FunSearch, it serves enterprise customers via an early-access waitlist.
- Ownership category
- akta.pro rank
Hiverge industry classification
Industry- Product category
- AI Algorithmic Discovery Software
- akta.pro primary industry
- LLMOps & Generative AI Platforms (Prompt/Agent Orchestration, RAG) (HDAEANAD)
- akta.pro secondary industry
- Prompt Engineering, Orchestration & LLMOps Tooling (HDAAACAD)
Keywords
Where Hiverge is headquartered
LocationHeadquarters
- HQ city
- Cambridge
- HQ country
- United Kingdom
- HQ region
- Europe
Offices1 record
Markets served
Hiverge business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Personnel, Technology or R&D, Infrastructure, Operations, Marketing or Sales
Revenue model
- Hive Platform Access: The company offers early access to its Hive algorithmic discovery platform to selected enterprise customers, suggesting a subscription or usage-based model for platform access and algorithmic discovery services.
- Algorithm Development Services: Custom algorithm development for specific enterprise optimization challenges, including AI training acceleration and industrial planning applications.
Go-to-market motion1 record
Distribution channels2 records
Marketing channels6 records
Hiverge product offering
Product offeringCore offering
Hiverge sells access to its Hive platform, an AI discovery engine that automatically writes and optimizes algorithms for critical optimization problems using program synthesis technology. The platform combines large language models with advanced search algorithms and deploys multiple agents working in parallel to optimize code based on concrete metrics. Target applications include AI model training acceleration, sequential planning (logistics, aerospace, supply chain), and quantum algorithm discovery for quantum chemistry.
Product overview
Hiverge is building algorithmic superintelligence through the Hive platform - an AI discovery engine that automatically writes breakthrough code and algorithms. The platform specializes in algorithmic discovery, finding novel computational approaches to solve complex optimization problems across AI training acceleration, sequential planning, and quantum computing. Users can access the Hive through a waitlist for early access to the platform.
Differentiator
Problem solved
Functional benefit
Brands
- Hive: The Hive is Hiverge's AI discovery engine that automatically writes algorithms for optimization problems using program synthesis technology.
Products and services
- Hive Platform AI discovery engine that automatically writes and optimizes algorithms for critical optimization problems. Uses program synthesis combining large language models with advanced search algorithms to deploy multiple agents working in parallel to tailor algorithms to specific challenges including AI training acceleration, sequential planning, and quantum computing. Targeted at enterprise customers via early access waitlist.
Quantifiable outcome
- Over 20% reduction in CIFAR-10 training time, achieving training in 1.99 seconds on single A100 GPU (first time below 2-second mark)
- +4 more outcomes
Companies that use Hiverge
Customer profileNamed customers1 record
Segments3 records
Ideal customer profiles3 records
Hiverge technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- No
- API docs
- API detail
Core technology
AI maturity
App detail
Integration5 records
AI capability5 records
Feature5 records
Hiverge partnerships and signals
Strategic signalPartnerships
Three partnerships are on record, tiered core.
- Amazon Web Services (AWS)coreAWS provides cloud infrastructure for Hiverge's quantum algorithm discovery. Quantum algorithms generated by the Hive are simulated on powerful NVIDIA GPUs running on AWS before being validated on Quantinuum's quantum hardware.
- NVIDIAcoreNVIDIA collaborated with Hiverge as part of hybrid quantum-AI supercomputing initiatives. NVIDIA's Grace Blackwell platform and CUDA-Q framework are used together with Hiverge's Hive platform. NVIDIA GPU-based decoders improved logical fidelity of quantum operations by more than 3% on Quantinuum's Helios quantum computer.
- QuantinuumcoreQuantinuum partnered with Hiverge to use AI-driven automated algorithm discovery for quantum chemistry. The collaboration uses Hiverge's Hive platform with Quantinuum's InQuanto quantum chemistry platform to develop quantum algorithms for molecular electronic structure problems. Algorithms are simulated on NVIDIA GPUs before validation on Quantinuum's quantum hardware. Achieved up to 70% reduction in two-qubit gate requirements compared to state-of-the-art VQE algorithms.
Scale indicators6 records
Recent moves6 records
Expansion highlights5 records
Hiverge competitors and assessment
Company assessmentMarket position
Competitive moat4 records
Key highlights6 records
Customer concentration
Hiverge social profiles
Digital presenceHiverge financial estimates
Financial estimateRevenue estimate
Valuation estimate
Hiverge leadership team
Management profileNumber of profiles
Profiles3 records
Hiverge funding detail
Funding detailFunding overview
Funding rounds1 record
Investors3 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Hiverge M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Hiverge
What does Hiverge do?
Hiverge sells access to its Hive platform, an AI discovery engine that automatically writes and optimizes algorithms for critical optimization problems using program synthesis technology. The platform combines large language models with advanced search algorithms and deploys multiple agents working in parallel to optimize code based on concrete metrics. Target applications include AI model training acceleration, sequential planning (logistics, aerospace, supply chain), and quantum algorithm discovery for quantum chemistry.
Is Hiverge a public or private company?
Hiverge is a private company. It is classified as venture growth investor backed and is currently operating.
When was Hiverge founded?
Hiverge was founded in 2024. It employs 1 to 10 people.
Where is Hiverge based?
Hiverge is headquartered in Cambridge, United Kingdom, in the Europe region.
How does Hiverge make money?
Two revenue lines are on record. Hive Platform Access are the primary driver. The others are algorithm Development Services.
Does Hiverge have an API?
No public API is recorded for Hiverge.
What industry is Hiverge in?
Hiverge's product category is AI Algorithmic Discovery Software. Its primary akta.pro industry code is HDAEANAD, LLMOps & Generative AI Platforms (Prompt/Agent Orchestration, RAG), with a secondary code of HDAAACAD, Prompt Engineering, Orchestration & LLMOps Tooling.