Fundamental

What Fundamental does

Fundamental is an AI research company that builds NEXUS, a Large Tabular Model (LTM) foundation model purpose-built for enterprise predictive analytics on structured (tabular) data. Founded in October 2024 by DeepMind alumni, the company positions NEXUS as an alternative to general-purpose LLMs (which fail on tabular data due to tokenization, context-window, and precision issues) and to traditional gradient-boosting pipelines (which require months of bespoke feature engineering per use case). The model uses a proprietary non-transformer architecture, is pre-trained on billions of real-world tables, and produces deterministic outputs with no hallucination risk; it is invoked through a scikit-learn-style Python SDK and deploys either directly from Fundamental, as a single-tenant network-isolated endpoint on Amazon SageMaker inside the customer's own AWS account, or via a hardware-attested confidential computing platform that keeps both model IP and customer data isolated.

The company sells to Fortune 100 enterprises across financial services, insurance, healthcare, energy, mining, telecom, retail, supply chain, and manufacturing, with use cases spanning fraud detection, demand forecasting, predictive maintenance, churn prediction, underwriting, and revenue forecasting. Revenue mechanics combine seven-figure enterprise subscription contracts (sold direct and via AWS Marketplace), usage-based inference consumption on customer-owned SageMaker infrastructure, and Forward Deployed Engineer (FDE) professional services that drive Proof-of-Value engagements and integration. Distribution is anchored by flagship partnerships with AWS and SAP and a Cloudera Enterprise AI Ecosystem seat, supported by a content- and thought-leadership-heavy marketing program.

Fundamental is privately held, headquartered in San Francisco, with research hubs in Barcelona and Tel Aviv and a Houston-based energy vertical presence. As of February 5, 2026 (stealth emergence date), the company had raised a cumulative $255 million across a $30 million seed and a $225 million Series A led by Oak HC/FT at a $1.2 billion post-money valuation, with participation from Salesforce Ventures, Valor Equity Partners, Battery Ventures, and Hetz Ventures. No revenue figure is publicly disclosed.

Name

Fundamental

Legal name

Fundamental Technologies Inc.

Website

https://fundamental.tech

Company type

Private

Founded year

2024

Operating status

Operating

Headcount range

1–10 employees

Short description

Fundamental is a private AI research company that builds NEXUS, a Large Tabular Model foundation model for enterprise predictive analytics on structured data, selling seven-figure contracts to Fortune 100 customers via direct field sales, the AWS Marketplace, and FDE-led deployments.

Ownership category

akta.pro rank

Product category

Enterprise AI Foundation Model Software

NAICS

Computer Systems Design and Related Services (54151), Computer Systems Design Services (541512)

SIC

Services-Computer Integrated Systems Design (7373), Services-Prepackaged Software (7372)

akta.pro primary industry

Foundation Model Developers (LLM/Multimodal Model Labs) (HDAAACAA)

akta.pro secondary industry

Decision Optimization & Prescriptive Analytics (OR, simulation, what‑if) (HDAAAHAI)

Keywords

• Enterprise AI platform
• Foundation model software
• Predictive analytics
• Tabular machine learning
• Large tabular model

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Offices5 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Personnel, Infrastructure, Marketing or Sales, Operations

Revenue model

1. Enterprise software contracts (NEXUS licensing): Seven-figure contracts with Fortune 100 enterprises for NEXUS deployment, sold via direct enterprise sales and the AWS Marketplace. Customers subscribe to the NEXUS model package, deploy it on a SageMaker asynchronous inference endpoint, and connect via the Fundamental Python SDK.
2. Forward Deployed Engineer (FDE) professional services: Embedded FDE teams drive Proof-of-Value engagements, integration, and post-implementation refinement of NEXUS for each customer, generating services and customization revenue alongside the core subscription.
3. Managed / cloud-delivered inference via AWS: Revenue captured through dedicated, single-tenant AWS infrastructure (SageMaker) running inside the customer's environment, with usage-based consumption of model inference capacity tied to customer compute spend.

Pricing tiers

Go-to-market motion4 records

Distribution channels4 records

Marketing channels7 records

Core offering

Fundamental builds and sells NEXUS, a Large Tabular Model (LTM) foundation model that performs enterprise predictive analytics on structured tabular data. Pre-trained on billions of real-world tables, NEXUS replaces months of bespoke feature engineering and model selection with a single pretrained model invoked via one line of code, supporting fraud detection, demand forecasting, predictive maintenance, churn, underwriting, and time series forecasting for Fortune 100 enterprises.

Product overview

Fundamental offers a single core foundation model — NEXUS, a Large Tabular Model (LTM) — supported by a small set of named developer and deployment assets that together form one unified enterprise predictive-AI platform rather than a suite of separate SaaS products. At the center is NEXUS itself (the pretrained LTM, accessible at fundamental.tech/nexus), which is invoked through the Fundamental Python SDK and exposed to enterprise customers through a scikit-learn-style fit()/predict()/predict_proba() interface. NEXUS is available through three deployment surfaces that work as a single product family: (1) the NEXUS on Amazon SageMaker option (a single-tenant, network-isolated SageMaker endpoint distributed via AWS Marketplace), (2) the NEXUS Confidential Computing Platform (a hardware-attested in-VPC deployment for regulated buyers), and (3) the direct-from-Fundamental channel. Capability-wise, NEXUS bundles general tabular predictive analytics with a dedicated NEXUS Time Series Forecasting module that automatically selects and combines forecasting architectures. A public-facing showcase, the NEXUS World Cup Prediction app at soccer.fundamental.tech, demonstrates the model's capabilities on real-world match data. Together, these constitute a tightly integrated product portfolio anchored on one foundation model.

Differentiator

Problem solved

Functional benefit

Brands

• NEXUS: Fundamental's Large Tabular Model (LTM) — a foundation model purpose-built for enterprise tabular/structured data analysis and predictive analytics.

Products and services

• NEXUS Large Tabular Model Fundamental's flagship Large Tabular Model (LTM) foundation model, pre-trained on billions of real-world tables to perform enterprise predictive analytics on structured data. Replaces months of feature engineering and bespoke ML pipelines with a single pretrained model accessed via one line of code, supporting fraud detection, demand forecasting, predictive maintenance, churn, underwriting, and time series forecasting. Uses a non-transformer deterministic architecture with no hallucination risk, deployable directly from Fundamental, via Amazon SageMaker, or within customer VPCs using Fundamental's confidential computing platform.
• Fundamental Python SDK Python SDK providing a scikit-learn-style interface (fit(), predict(), predict_proba()) to the NEXUS model. Handles serialization, endpoint invocation, asynchronous polling, and model management when connected to a NEXUS SageMaker endpoint. Designed for enterprise-scale workloads including gigabyte-scale datasets, dedicated GPU instances, asynchronous inference, and S3-native data movement.
• NEXUS on Amazon SageMaker NEXUS distributed via AWS Marketplace as a model package that deploys as a dedicated, single-tenant, network-isolated Amazon SageMaker asynchronous inference endpoint inside the customer's own AWS account. Datasets remain in the customer's S3 buckets; inputs, outputs, and model artifacts never transit a Fundamental-controlled environment, and the container runs network-isolated with no outbound calls during inference.
• NEXUS Confidential Computing Platform Hardware-backed deployment platform that allows NEXUS models to run inside a customer's VPC with cryptographic attestation. The verified software-stack fingerprint is embedded in the key-release policy so the encrypted model only decrypts if the runtime matches the approved fingerprint. Designed for regulated industries (banking, healthcare, energy) where neither data nor model IP can leave its respective environment.
• NEXUS World Cup Prediction (soccer.fundamental.tech) Public demonstration application that uses NEXUS to predict outcomes of the 2026 FIFA World Cup match-by-match, drawing on 42,000 historical international matches plus recent form. Reports title-win probabilities and correct-call rates, serving as a public showcase of NEXUS applied to sports prediction.

Quantifiable outcome

• 81% of decisive 2026 World Cup group-stage matches called correctly by NEXUS; Argentina flagged as 17.2% title favorite vs market's 12% (Polymarket).
• +4 more outcomes

Named customers3 records

Segments5 records

Ideal customer profiles1 record

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration5 records

AI capability9 records

Feature6 records

Partnerships

Five partnerships are on record, tiered flagship and core.

• Amazon Web Services (AWS)flagshipStrategic or Co-development Partner · 5 February 2026Strategic partnership announced at stealth emergence on February 5, 2026, enabling direct enterprise deployment of NEXUS through the AWS dashboard and via the AWS Marketplace. AWS CEO Matt Garman publicly endorsed the partnership, stating that Fundamental's breakthrough in structured data prediction complements AWS's mission to make advanced AI accessible to enterprises. NEXUS is now available on Amazon SageMaker as a single-tenant, network-isolated instance inside the customer's own AWS account, with training performed on Amazon SageMaker HyperPod.
• SAPcoreTechnology or Integration · 1 January 2026Integration partnership enabling NEXUS to work directly with SAP enterprise data, described by the company as making 'the world's enterprise data just got a lot smarter'. A dedicated 'NEXUS + SAP' blog post on the Fundamental site frames SAP as a key data source for the model.
• ClouderacoreStrategic or Co-development Partner · 14 October 2025Fundamental joined Cloudera's Enterprise AI Ecosystem as one of four new partners (alongside ServiceNow, Pulse, and Galileo.ai), delivering a predictive engine for structured data to Cloudera's enterprise customers. Cloudera's Chief Strategy Officer described the ecosystem as a cornerstone of its AI-native customer strategy.
• Gaël Varoquaux (scikit-learn co-creator)coreStrategic or Co-development PartnerFounding Advisor at Fundamental and Chief Science Officer of Probabl. Featured as the inaugural guest on the company's 'Ground Truth' video series, where he argued that researchers should pick tabular AI over the crowded LLM field. His work on TabPFN and TabICL is cited as adjacent open-source LTM research.
• Wojciech Marian CzarneckiflagshipStrategic or Co-development PartnerFounding Advisor at Fundamental, co-author of the company's research whitepaper, and featured inaugural guest on the 'First Principles' research interview series. Best known for his work on multi-agent reinforcement learning, including DeepMind's StarCraft project.

Scale indicators10 records

Recent moves8 records

Expansion highlights7 records

Direct peers

• H2O.ai: Provider of an enterprise AI platform (H2O AI Cloud, Driverless AI) for predictive modeling on structured and tabular data, with strong Fortune 100 penetration in financial services. Closely comparable in target use cases and customer base, even though it uses a more traditional AutoML / open-source ML approach rather than a dedicated tabular foundation model.
• dotData: Enterprise AutoML and predictive analytics platform focused on automating feature engineering and model deployment for Fortune 1000 customers. Directly comparable to NEXUS in target customers (large enterprises), use cases (forecasting, fraud, churn), and value proposition (collapsing manual data science work).
• DataRobot: Enterprise AI/ML platform focused on predictive analytics, AutoML, and time-series forecasting for Fortune 100 buyers. Directly comparable to NEXUS in use cases (fraud, demand forecasting, churn, predictive maintenance) and customer profile, though it relies on a classical ML/AutoML stack rather than a purpose-built foundation model for tabular data.

Broad incumbents

• Databricks: Unified data and AI platform (including Mosaic AI) used by enterprises to build ML and predictive models on large tabular datasets. Comparable as both a partner (Fundamental integrates with Databricks-resident data) and a broad incumbent that could extend into tabular foundation models.
• Google Vertex AI: Google Cloud's managed ML platform offering foundation models and AutoML capabilities for tabular and other data, used by large enterprises. Comparable as a broad incumbent enterprise AI platform serving similar predictive use cases, even though it does not yet offer a dedicated Large Tabular Model.
• Salesforce Einstein: Salesforce's enterprise AI platform, including predictive scoring, forecasting, and Einstein GPT for CRM and tabular data. Comparable as a broad incumbent in enterprise predictive AI for structured data, and an investor (Salesforce Ventures) in Fundamental, reflecting both partnership and competitive overlap.
• Palantir Technologies: Enterprise data and AI platform (Foundry, AIP) widely used by Fortune 100 and government customers for predictive and operational use cases on large structured datasets. Comparable in customer profile and forward-deployed delivery model (Fundamental explicitly hires ex-Palantir FDEs), and an indirect alternative to NEXUS for enterprise predictive workloads.
• Amazon SageMaker: AWS's managed ML platform on which NEXUS deploys as a single-tenant endpoint. Directly comparable as a deployment surface and a channel for enterprise tabular ML workloads, and a potential competitor if AWS extends its own built-in models for structured data.

Emerging players

• Prior Labs (TabPFN): Research-driven startup behind TabPFN, one of the leading open-source tabular foundation models. Directly comparable as a tabular foundation-model effort and an explicit adjacent research effort to NEXUS, with Founding Advisor Gaël Varoquaux contributing to the open-source LTM line of work.
• AI21 Labs: Foundation model company building enterprise AI systems (Jamba, Jurassic). Comparable as a foundation-model vendor with enterprise focus; Fundamental's Founding Engineer Yuval Azoulay previously worked at AI21 Labs on training and serving AI models at scale.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/ltmpredict
• YouTubeyoutube.com/live/R7EchZ1ITD8?si=QigYOhvjF0QNVeua&t=8355
• LinkedInlinkedin.com/company/fundamentalhq

Compliance4 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles15 records

Funding overview

Funding rounds3 records

Investors16 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Fundamental