AI companies in focus
Who is shaping the AI market — from infrastructure to application. Every figure with source and date, no share prices, no forecasts.
AI companies in focus
Not investment adviceWho drives AI development — and how do these companies make money? Context instead of quotes: every fact dated and sourced.
- Infrastructure
NVIDIA
publicly listed · NVDAUSA
NVIDIA builds graphics and compute accelerators along with the CUDA software platform. Data-centre hardware for training and running AI models now accounts for most of its business.
The company supplies a large share of the compute on which current AI models are trained, which makes its quarterly figures a widely watched indicator of investment across the AI ecosystem.
- Quartalsumsatz Q1 Geschäftsjahr 2027:
- 81,6 Mrd. USD, +85 % gegenüber dem Vorjahresquartal Source 2026-05
- Umsatz Rechenzentrums-Segment Q1 Geschäftsjahr 2027:
- 75,2 Mrd. USD, +92 % gegenüber dem Vorjahresquartal Source 2026-05
- Infrastructure
Microsoft
publicly listed · MSFTUSA
Microsoft operates Azure, one of the major cloud platforms, and embeds AI assistance under the Copilot brand across its productivity and developer products. It is also an investor in and infrastructure partner to OpenAI.
For many organisations Azure is the route through which AI models reach existing IT estates. Microsoft now reports AI revenue separately, which makes adoption easier to track.
- Konzernumsatz Q3 Geschäftsjahr 2026:
- 82,9 Mrd. USD, +18 % gegenüber dem Vorjahresquartal (+15 % währungsbereinigt) Source 2026-04
- Wachstum Azure und andere Cloud-Dienste:
- +40 % gegenüber dem Vorjahresquartal (+39 % währungsbereinigt) Source 2026-04
- Annualisierte Umsatzrate des KI-Geschäfts:
- über 37 Mrd. USD, +123 % gegenüber dem Vorjahr Source 2026-04
- Models
Alphabet (Google)
publicly listed · GOOGLUSA
Through Google DeepMind, Alphabet develops the Gemini model family and distributes it via Google Cloud, the Gemini app and Search. It also designs its own AI accelerators, the TPUs.
Alphabet is one of few companies covering the full chain from in-house silicon through models to end-user products. Its cloud results indicate how far enterprises have taken these models into production.
- Konzernumsatz Q1 2026:
- 109,9 Mrd. USD, +22 % gegenüber dem Vorjahresquartal (+19 % währungsbereinigt) Source 2026-04
- Applications
Palantir Technologies
publicly listed · PLTRUSA
Palantir provides data integration and analysis platforms for government and commercial customers. Its Artificial Intelligence Platform (AIP) connects language models to operational data and workflows.
The company represents the application layer: the product is not the model but its embedding into existing processes and data. Its US revenue trajectory shows how quickly that market is currently growing.
- Umsatz US-Geschäftskundengeschäft Q1 2026:
- 595 Mio. USD, +133 % gegenüber dem Vorjahresquartal Source 2026-05
- Infrastructure
AMD (Advanced Micro Devices)
publicly listed · AMDUSA
AMD develops EPYC server processors and Instinct AI accelerators, with the open ROCm stack covering the software side.
AMD is the largest direct competitor in the AI accelerator market. How quickly a credible second source establishes itself matters to data-centre operators.
- Umsatz Rechenzentrums-Segment Q1 2026:
- 5,8 Mrd. USD, +57 % gegenüber dem Vorjahresquartal Source 2026-05
- Applications
SAP
publicly listed · SAPDeutschland
SAP builds enterprise software for finance, supply chain and HR, increasingly delivered from the cloud. Under the Business AI label, AI capabilities are embedded directly into these business processes.
SAP is Europe's largest software vendor and, for many companies, the place where AI features meet actual master data. Its cloud backlog indicates how firmly customers have committed to that shift.
- Cloud-Umsatzwachstum Q1 2026:
- +19 % nominal, +27 % währungsbereinigt gegenüber dem Vorjahresquartal Source 2026-04
- Infrastructure
ASML
publicly listed · ASMLNiederlande
ASML builds the lithography systems chipmakers use to pattern circuits onto wafers. It is the sole global supplier of EUV technology for the most advanced process nodes.
Without ASML systems, the chips that run AI models cannot be manufactured at current geometries. Since 2025 the company has also been an anchor investor in Mistral AI.
- Data & platforms
Snowflake
publicly listed · SNOWUSA
Snowflake runs a cloud data platform where organisations store, combine and analyse their data. Cortex and Snowflake Intelligence let language models operate directly on that data.
Enterprise AI projects rarely fail on the model but on access to clean data. Snowflake discloses how many accounts actively use its AI features, making adoption measurable.
- Produktumsatz Q4 Geschäftsjahr 2026:
- 1,23 Mrd. USD, +30 % gegenüber dem Vorjahresquartal Source 2026-02
- Models
OpenAI
privately heldUSA
OpenAI develops the GPT model family and operates ChatGPT, one of the most widely used AI applications. Revenue comes from consumer subscriptions, enterprise licences and a developer API.
The launch of ChatGPT in late 2022 shaped today's market. Despite substantial revenue, the company is not yet profitable.
- Finanzierungsrunde:
- 122 Mrd. USD eingesammelt bei einer Post-Money-Bewertung von 852 Mrd. USD Source 2026-03
- Models
Anthropic
privately heldUSA
Anthropic develops the Claude model family, focused primarily on enterprise customers and developers. Safety and interpretability research is a stated part of its work.
It is one of few frontier model providers and holds a strong position in the API and coding segments. Its funding rounds illustrate the scale of capital required for model training.
- Series-H-Finanzierungsrunde:
- 65 Mrd. USD eingesammelt bei einer Post-Money-Bewertung von 965 Mrd. USD Source 2026-05
- Models
Mistral AI
privately heldFrankreich
Paris-based Mistral AI develops language and multimodal models, releasing some under open licences. It also offers the La Plateforme developer platform and dedicated compute under the Mistral Compute brand.
Mistral is Europe's best-known developer of frontier models and is frequently cited in debates on digital sovereignty. ASML's role as lead investor ties it to the European semiconductor industry.
- Series-C-Finanzierungsrunde:
- 1,7 Mrd. EUR eingesammelt bei einer Post-Money-Bewertung von 11,7 Mrd. EUR Source 2025-09
- Lead-Investor der Series C:
- ASML Holding NV, weitere Beteiligung u. a. von DST Global, Andreessen Horowitz, Bpifrance, General Catalyst, Index Ventures, Lightspeed und NVIDIA Source 2025-09
- Models
xAI
privately heldUSA
xAI develops the Grok model family and runs its own Colossus I and II data centres in Memphis. The models are delivered via an API and through the X platform.
xAI has chosen to build compute rather than rent it. The scale of these clusters shows the infrastructure effort behind training current frontier models.
- Series-E-Finanzierungsrunde:
- 20 Mrd. USD eingesammelt; das ursprüngliche Ziel von 15 Mrd. USD wurde überschritten. Strategische Investoren sind unter anderem NVIDIA und Cisco Investments. Source 2026-01
- Rechenkapazität Colossus I und II:
- über eine Million H100-GPU-Äquivalente zum Jahresende Source 2026-01
- Data & platforms
Databricks
privately heldUSA
Databricks runs a data platform that merges warehouse and lake architectures (the lakehouse approach), on which organisations can train, adapt and operate their own models.
The company addresses the layer between raw data and AI application. The share of AI in its revenue shows how much productive AI work actually happens on enterprise data.
- Run-Rate-Umsatz aus KI-Produkten:
- über 1 Mrd. USD (bei einem Gesamt-Run-Rate von 4,8 Mrd. USD zu diesem Zeitpunkt) Source 2025-12
- Data & platforms
Hugging Face
privately heldUSA / Frankreich
Hugging Face runs the largest open platform for machine-learning models, datasets and demo apps, alongside widely used open-source libraries such as Transformers and Datasets.
The platform is the main clearing house for open AI models and a useful gauge of what is built outside the large commercial labs. Its platform metrics are published openly.
- Applications
Helsing
privately heldDeutschland
Munich-based Helsing develops AI software for defence applications and autonomous systems, including the HX-2 drone. It positions itself as a software-defined supplier to European armed forces.
Helsing is currently Germany's highest-valued AI company and shows that defence has become a distinct field of AI application, with its own ethical and regulatory questions.
- Series-E-Finanzierungsrunde:
- 1,8 Mrd. USD bei einer Bewertung von 18 Mrd. USD; Investoren unter anderem Dragoneer, Lightspeed, Goldman Sachs Alternatives, JPMorganChase und CPP Investments Source 2026-07
- Applications
DeepL
privately heldDeutschland
Cologne-based DeepL builds neural translation and writing tools and has extended its range with an AI assistant for workflows. It operates its own GPU infrastructure.
DeepL is one of few German AI companies with an established international product business, showing that a tightly defined use case can be viable without a general frontier model.
- Finanzierungsrunde:
- 300 Mio. USD bei einer Bewertung von 2 Mrd. USD, angeführt von Index Ventures Source 2024-05
This overview provides market context and does not constitute investment advice or a recommendation to buy or sell. We deliberately show no prices and no forecasts. Figures are dated as stated and may have changed.