Author: Melanie Stuetz

NVIDIA is using CES 2026 to showcase how the Rubin platform, open AI models and autonomous driving are reshaping computing and business models across every industry. For innovation leaders, startups and mid‑market firms, this is the perfect moment to rethink their innovation pipeline with a data‑driven approach.​

CES 2026: Rubin as a new AI engine for enterprises

NVIDIA founder and CEO Jensen Huang opened CES 2026 in Las Vegas with the message that AI is scaling into every domain and every device, modernising around 10 trillion US dollars of legacy computing into accelerated computing. At the centre stands the new NVIDIA Rubin platform, the company’s first extreme co‑designed six‑chip AI platform, now in full production.​

Rubin is architected from the data centre outward and includes:​

• Rubin GPUs delivering 50 petaflops of NVFP4 inference performance for gigascale models.​
• Vera CPUs engineered for high‑efficiency data movement and agentic processing.​
• NVLink 6 for scale‑up networking and Spectrum‑X Ethernet Photonics for scale‑out networking.​
• ConnectX‑9 SuperNICs and BlueField‑4 DPUs as accelerators for network and storage workloads.​

By designing chips, racks, networking, storage and software together, Rubin removes bottlenecks and dramatically lowers training and inference costs. NVIDIA adds an AI‑native storage layer (Inference Context Memory Storage Platform) that boosts long‑context inference with up to 5x more tokens per second, 5x better TCO performance and 5x higher power efficiency.​

Open models: a global AI innovation fabric

On top of Rubin, NVIDIA positions itself as a frontier AI model builder, deliberately focusing on open models to enable companies, industries and countries to join the AI revolution. These models are trained on NVIDIA supercomputers and form a global fabric of intelligence that developers and enterprises can build upon.​

The open model portfolio spans six domains:​

• Clara for healthcare.
• Earth‑2 for climate science and simulation.​
• Nemotron for reasoning and multimodal AI.​
• Cosmos for robotics and simulation.​
• GR00T for embodied intelligence.​
• Alpamayo for autonomous driving.​

These models are open to the world: they can be created, evaluated, guardrailed and deployed in bespoke applications. For corporate innovation teams this means they can stop reinventing the wheel and instead test new products and services on top of proven base models, accelerating time‑to‑market.​

Physical AI & autonomous driving: Alpamayo, Cosmos & Mercedes‑Benz CLA

A key theme of the CES keynote is physical AI – intelligence embedded in real products, vehicles and factories. Huang highlighted how such systems are trained with synthetic data in virtual worlds long before they interact with the real world.​

Core building blocks include:​

• Cosmos, which generates realistic video from a single image, synthesises multi‑camera driving scenarios, models edge cases and drives closed‑loop simulation.​
• Alpamayo, an open portfolio of vision‑language‑action models, simulation blueprints and datasets designed for level‑4‑capable autonomy.​
• Alpamayo R1, the first open reasoning VLA model for autonomous driving, which not only acts on sensor input but can reason about its next action.​

The first production car to feature Alpamayo on the NVIDIA DRIVE full‑stack autonomous vehicle platform will be the new Mercedes‑Benz CLA, bringing AI‑defined driving to US roads this year. At the same time, the open, modular and level‑4‑ready DRIVE Hyperion platform is gaining traction among leading OEMs, suppliers and robotaxi providers.​

Gaming, creators & personal agents: AI on every desk

Beyond data centres, NVIDIA uses CES to show how AI reaches the desktop and everyday devices. One example is DGX Spark, a desktop “supercomputer” capable of running personalised AI agents locally, embodied for instance via a Reachy Mini robot using Hugging Face models.​

For gaming and content creation, several updates were announced:​

• DLSS 4.5 with Dynamic Multi Frame Generation, a new 6X mode and a second‑generation transformer for DLSS Super Resolution, now supported by over 250 games and apps.​
• RTX Remix Logic expands modding by triggering graphical effects based on real‑time game events.​
• NVIDIA ACE brings AI companions with long‑term memory to titles like Total War: PHARAOH and PUBG: BATTLEGROUNDS.​
• G‑SYNC Pulsar monitors deliver tear‑free gameplay with perceived 1,000Hz+ motion clarity and adaptive technology.​

In addition, NVIDIA is expanding GeForce RTX gaming to more devices via new GeForce NOW apps for Linux PC and Amazon Fire TV, while NVIDIA RTX accelerates 4K AI video generation on PCs with LTX‑2 and ComfyUI upgrades.​

Turning CES 2026 signals into a concrete own roadmap

CES 2026 clearly signals how fast AI hardware, open models and physical AI are reshaping competitive landscapes – from manufacturing plants to hospitals and entertainment platforms. Many business models that worked yesterday risk being outpaced by AI‑first challengers that build directly on platforms like Rubin and open model ecosystems.​

This is where IDEASCANNER creates tangible value:

• IDEASCANNER quantifies the strength of innovation and product portfolios, linking technological trends like Rubin, Cosmos or Alpamayo to business value.
• Using a structured scoring methodology, organisations can benchmark ideas, prioritise investments and align innovation projects with AI opportunities revealed at CES.
• Leadership teams gain a clear view on which existing offerings can evolve into AI‑defined solutions and where new, data‑driven business models should be explored first.

For organisations that do not just want to observe CES announcements but translate them into action, now is the time to rescan their idea portfolio.

With IDEASCANNER, decision‑makers can systematically identify which ideas are AI‑ready, where strategic partnerships with model or platform providers make sense, and how to shorten time‑to‑market for new AI‑enabled offerings.

Next steps:

1. Book a non‑binding IDEASCANNER session to discuss how CES 2026 themes like NVIDIA Rubin, open models and autonomous driving can be mapped to your specific business – and which ideas should be scanned first.
2. Follow IDEASCANNER for more actionable expert insights.
3. Contact us and let’s think smarter with AI together.

Gemini 3 is Google’s most powerful AI model to date and is available in Google AI Studio at https://aistudio.google.com/models/gemini-3, making it an ideal choice for anyone building AI-powered products, services and innovations. Combined with IDEASCANNER, organisations can not only prototype faster, but also assess whether the underlying business idea is truly worth pursuing – before committing to costly development and scaling.

What is Gemini 3?

Gemini 3 is the new model generation from Google DeepMind and is positioned as a model to “help with anything”: learning, building and planning – from the first idea to a finished AI product. It unifies cutting-edge text, image, audio, video and code processing in a single multimodal model and sets new benchmarks in understanding, reasoning and interaction.

The first available variant, Gemini 3 Pro, is integrated into Google AI Studio, Vertex AI, the Gemini API and other developer tools, and is optimised for demanding reasoning, coding and agent workflows. In addition, Google introduces “Gemini 3 Deep Think”, a mode designed for particularly complex tasks such as scientific questions or strategic planning.

Key features in AI Studio

In Google AI Studio, users can try out Gemini 3 Pro directly in the browser, iterate on prompts, use templates and export ready-made setups into their own applications. For developers, product teams and innovators this significantly accelerates the journey from an idea to a working prototype, including front end, logic and API integrations.

Gemini 3’s strengths in AI Studio include advanced vibe coding for UI/UX, agentic workflows for multi-step automation and a context window of up to one million tokens – ideal for long documents, research material or complex multimodal data. Moreover, Gemini 3 Pro Image enables the generation and editing of images with improved text and diagram rendering, for example for infographics, product visuals or learning content.

Gemini 3 vs predecessors and peers

*Generic positioning based on published benchmarks and market reports.

In practice, this makes Gemini 3 particularly suitable for applications with multiple media types, demanding workflows and long contexts – such as interactive learning platforms, complex dashboards, dynamic knowledge assistants or AI-driven product configurators.

“Learn, Build, Plan Anything” in action

Google clearly positions Gemini 3 around the use cases “Learn anything”, “Build anything” and “Plan anything”, each with a strong multimodal focus. Examples span from analysing scientific papers including interactive visualisations and evaluating sports videos with personalised training plans to translating handwritten family recipes into structured cookbooks.

In an enterprise context, teams can use Gemini 3 to turn internal knowledge assets (documents, presentations, videos) into training and onboarding materials or to simulate complex project and investment plans with real-time data. Through integration into Google Search and AI Mode, new interactive research interfaces emerge that make expertise more accessible – for instance via visual layouts, simulations or infographics generated directly from a query.

Why combine Gemini 3 with IDEASCANNER?

Gemini 3 makes it easier than ever to build AI prototypes and digital products with state-of-the-art reasoning and multimodality – but technology alone does not guarantee a viable business model. This is where IDEASCANNER comes in: Using a structured scan approach, it systematically evaluates business ideas and existing models, identifies scalable levers and reveals innovation potential – before capital is poured into expensive AI development.

Anyone using Gemini 3 in Google AI Studio to create new applications should scan their idea in parallel: Where is the real customer value, how differentiated is the approach compared to competitors and how can AI be used to create measurable value rather than just another “feature”? IDEASCANNER helps answer exactly these questions in a data-driven way – turning Gemini 3 into a catalyst for the best ideas, not just the loudest ones.​

Take action now

Use Gemini 3 in Google AI Studio to rapidly build powerful prototypes and AI services – and combine it with an IDEASCANNER to invest only in ideas that truly pay off. Book a call with IDEASCANNER today to scan your current or planned AI project and gain a solid decision basis for your strategy, product roadmap and investments.