EN

Google’s Gemini 3 Receives Rave Reviews

#AI Tools
2025-11-21 20:338 min read

Content Introduction
Ask Questions

The video features a discussion on the 'Gemini leap' in tech, with Mandep Singh, a senior tech analyst at Bloom Intelligence. They explore the advancements brought by the Gemini 3 model in multimodal capabilities, highlighting its ability to perform visual reasoning, image generation, and coding without relying on Nvidia GPUs. The conversation reflects on Google's transition to using TPUs (Tensor Processing Units) for their computing needs, indicating a significant shift in their operational strategy. They further speculate on how this could enhance Google Cloud's capacities and overall revenue. The analysts also mention the previous failures of Google's Bard launch and the improvements seen over two years, positioning the Gemini 3 as a leading frontier model. They conclude by discussing the implications of these advancements for cloud computing and external customer relationships, suggesting a potential uplift in cloud revenue due to increased availability of Nvidia GPUs.

Key Information

  • Discussion focuses on the so-called Gemini leap and its implications.
  • Mandep Singh, a senior tech analyst, indicates that this represents a significant advancement, particularly in visual reasoning.
  • The Gemini model is highlighted as a true example of multimodality, capable of performing various tasks including coding and image generation.
  • The analysis points out that the Gemini 3 model has been executed well, utilizing TPUs rather than traditional Nvidia GPUs.
  • It is noted that many still rely on Nvidia for training, which could change with the rise of Google's TPUs.
  • The conversation emphasizes the potential for Google's cloud and GCP to repurpose Nvidia resources for clients, enhancing their cloud business.
  • The panel recognizes Google's ongoing investment in Nvidia chips, revealing their status as one of Nvidia's top customers.
  • Reports highlight how Alphabet's internal operations rely heavily on TPUs and Google Cloud for machine learning and AI applications.
  • Conjecture is made on the increased revenue potential for cloud services due to the greater availability of Nvidia GPUs.

Timeline Analysis

Content Keywords

Gemini Leap

Discussion on the implications of the so-called 'Gemini leap' in AI technology, particularly in visual reasoning and multimodal models. Senior tech analyst Mandep Singh provides insights into how benchmarks demonstrate significant advancements.

Gemini 3 Model

The Gemini 3 model is highlighted as a frontier model that effectively executes multimodal tasks, including image generation and code creation, showcasing remarkable capability in AI performance.

TPUs vs Nvidia GPUs

The conversation points out that, unlike many AI models that rely on Nvidia GPUs, the Gemini model is trained on Google's TPUs. This indicates a significant shift in AI training strategies.

Impact on Google Cloud

The discussion explores how Google's successful use of TPUs could free up their Nvidia allocation for other customer uses, potentially enhancing Google's cloud services and revenue.

AI Image Generation

The model discusses capabilities of generating images and engaging in visual reasoning, illustrating the depth of AI's applications in creative and technical fields.

Cloud Revenue Growth

Predictions suggest that the availability of more Nvidia GPUs may lead to an increase in cloud revenue, indicating a positive outlook for Google's cloud business expansion efforts.

Related questions&answers

What is the so-called Gemini leap?

The Gemini leap refers to significant advancements in visual reasoning and multimodality capabilities in AI, specifically with the introduction of the Gemini 3 model.

Who is Mandep Singh?

Mandep Singh is a senior tech analyst with Bloom Intelligence.

Did they use Nvidia GPUs for training the Gemini model?

No, the training for the Gemini model was done entirely on TPUs without Nvidia GPUs.

What are the implications of Google not relying on Nvidia for training?

This could free up Google Cloud resources to allocate more to external-facing customers, enhancing their cloud business.

How successful is the VHMO in terms of AI?

The VHMO has been successful due to their extensive experience and capabilities in AI, as reflected in their recent performance advancements.

What is the significance of Google's TPUs?

Google's TPUs (Tensor Processing Units) provide a customized solution for running AI workloads efficiently, which has been pivotal in their cloud deployments.

What has been the trend in Google's use of Nvidia chips?

Despite the advancements with TPUs, Google remains one of the top customers for Nvidia chips.

How does the Gemini 3 model differ from previous models?

Gemini 3 model showcases improved execution in handling multimodal tasks compared to earlier models, making it a frontier model in AI capabilities.

What predictions were made regarding cloud revenues?

There is an expectation that cloud revenue could increase due to the enhanced availability of Nvidia GPUs and the overall advancements in AI capabilities.

What role does Google Cloud play in utilizing Nvidia's allocation?

Google Cloud utilizes Nvidia's allocation to support various AI workloads, including training and inference systems.

More video recommendations

Share to: