• AI Emergence
  • Posts
  • NVIDIA Revenue Up 94%, Profits More Than Double But Shares Fall!

NVIDIA Revenue Up 94%, Profits More Than Double But Shares Fall!

Along with: Microsoft’s AI driven agents!!!

Hey there, 

Today, we got a glimpse of what to expect in the coming quarters. Let me explain!

Remember “With great power comes great responsibility”? The equivalent proverb for AI is “With huge investments comes massive expectations.

NVIDIA announced its quarterly results. The revenue grew 94%, and profits more than doubled. 

But - the share price - well, it fell! Because this is not good enough for NVIDIA.

I think we are going to see this more often in the future. Results that otherwise look great are falling short of expectations.

Let’s look at the developments this week!

What would be the format? Every week, we will break the newsletter into the following sections:

  • The Input - All about recent developments in AI

  • The Tools - Interesting finds and launches

  • The Algorithm - Resources for learning

  • The Output - Our reflection 

Table of Contents

NVIDIA smashed records this quarter, reporting a jaw-dropping $35.1 billion in revenue- a 94% year-over-year surge!

But here’s where it gets interesting: despite this stellar performance, NVIDIA forecasted its slowest revenue growth in seven quarters. This left some investors underwhelmed, causing shares to dip 5% post-announcement before recovering slightly to a 2.5% decline after hours.

What’s behind the slowdown? It may be the new Blackwell Chips: These AI powerhouse chips, essential for high-performance workloads, hit a snag with overheating concerns. NVIDIA has since collaborated with TSMC to resolve the design issues and requested suppliers to redesign server racks.

Despite these challenges, demand remains sky-high. Major players like AWS, Meta, Microsoft, and SoftBank have already secured supplies through 2025, reaffirming NVIDIA’s dominance in the AI chip market. (source)

Microsoft 365 Copilot is becoming a daily habit for organizations worldwide, with nearly 70% of Fortune 500 companies relying on it to power their workflows. But Microsoft isn’t stopping there. At Ignite 2024, the tech giant unveiled new Copilot agents that promise to revolutionize business processes and amplify employee productivity.

The Spotlight: Intelligent Agents

These agents, built on Copilot Studio, are designed to tackle routine tasks and complex workflows alike, turning them into seamless, automated processes. Here are the standout innovations that caught everyone’s attention:

  • SharePoint Agent: Each SharePoint site now includes an agent to streamline document access and project updates. Employees can quickly retrieve files, summarize memos, and access key resources without searching through multiple locations.

  • Interpreter Agent in Teams: This agent offers real-time speech-to-speech translation during meetings, even simulating the user’s voice for a more personal experience. It’s designed to make multilingual communication more natural and seamless.

  • Employee Self-Service Agent: Focused on improving internal workflows, this agent handles common HR and IT queries, like understanding benefits or requesting equipment. It can also be customized to meet specific organizational needs.

  • Facilitator Agent: Designed to improve productivity, it takes real-time notes during team meetings and chats.

  • Project Manager Agent: Automates task creation and management in Microsoft Planner, helping teams stay organized and efficient.(source)

Microsoft’s announcements also included updates to Copilot’s overall capabilities, such as automated workflows and collaborative tools. With the introduction of these agents. (source)

Google’s latest experimental AI model, Gemini-Exp-1114, has claimed the top spot on the Chatbot Arena leaderboard, achieving superior performance in mathematics, creative writing, and visual understanding with a score of 1344. 

However, deeper analysis highlights limitations in traditional benchmarking - as controlling for response formatting and length dropped the model to fourth place. 

This disparity reveals how current evaluation metrics often inflate AI capabilities by focusing on surface-level outputs rather than genuine advancements in reasoning, reliability, or safety.

Concerns over Gemini’s real-world reliability persist, with previous versions producing harmful outputs. (source)

The figure has announced major advancements in its Figure 02 humanoid robots, now boasting a 400% increase in speed and a sevenfold improvement in task success rates. 

These robots, designed for commercial and domestic use, are achieving 1,000 autonomous placements daily, with performance refined through digital twin simulations. 

Founder Brett Adcock envisions deploying millions of these robots globally, likening their impact to living 30 years into the future. (source)

Mistral AI introduces Pixtral Large, a 124-billion-parameter multimodal model based on Mistral Large 2, designed for advanced image, document, and chart understanding while maintaining exceptional text comprehension. 

Pixtral Large excels in benchmarks like MathVista, ChartQA, and DocVQA, outperforming leading models such as GPT-4o and Gemini-1.5 Pro. It showcases capabilities in multilingual OCR, chart reasoning, and document understanding, making it suitable for real-world applications. Available under research and commercial licenses, it provides cutting-edge performance for research, testing, and production. (source)

Perplexity has entered the e-commerce space with a new shopping feature for its U.S.-based paid customers. 

This feature offers shopping recommendations, visual product cards with pricing, pros and cons, and one-click checkout without redirecting users to retailer websites. 

The system integrates with platforms like Shopify and allows users to search by image, competing with giants like Google and Amazon. Merchants can enroll in Perplexity's program to enhance visibility and access free APIs for their sites. (source)

LangGraph has introduced significant updates to enhance interoperability and usability in multi-agent frameworks. 

The announcement includes the Agent Protocol, a standardized interface for agent communication that supports various frameworks like AutoGen and CrewAI. 

LangGraph Studio now connects locally, eliminating the need for Docker and enabling seamless debugging across platforms. 

Additionally, guides have been released to integrate LangGraph with other frameworks and deploy these integrations using the LangGraph Platform, which offers scalable infrastructure and memory support for agentic applications. (source)

OpenAI has expanded its Advanced Voice mode to the ChatGPT desktop web version. 

This allows users to engage in natural, emotive voice interactions with nine distinct voice profiles, including five new ones inspired by nature, such as Arbor and Ember. 

Currently available to paying subscribers (Plus, Enterprise, Teams, and Edu), the feature includes daily usage limits, with free users expected to get access in the coming weeks. 

Notably, the controversial "Sky" voice, allegedly resembling Scarlett Johansson and removed earlier due to legal action, remains unavailable.(source)

Ben Affleck calls AI a "cross-pollinator" that enhances creativity by simplifying technical challenges and lowering filmmaking costs. Speaking on CNBC, he highlights how AI empowers filmmakers to create more easily while fixing mistakes or tweaking details. However, he insists that AI cannot replace human artistry, as it lacks taste, consistency, and the ability to decide when to stop.(source)

The Apple Intelligence notification summary feature, available on the latest iPhones and devices with M1 chips or newer, attempts to condense missed notifications into concise summaries. While the feature provides quick overviews, its inconsistent performance often produces bizarre or inaccurate results, leading to frustration and unintended humor. 

The system uses predefined prompts to generate summaries but struggles to deliver reliable outputs, making its practical use limited in its current form.(source)

Elon Musk's, xAI, is reportedly raising $5 billion at a $50 billion valuation, with $5 billion expected from Middle Eastern sovereign funds and $1 billion from other investors. 

The funds will be used to acquire 100,000 Nvidia chips, critical for Tesla’s Full Self Driving capabilities, leveraging the Memphis supercomputer.(source)

Tool: Jammable

Create AI covers from your favorite voices.

How to Access:

  1. Sign in to the platform.

  2. Click on "Create."

  3. Select "Duets."

  4. Choose the voices you want to include.

  5. Add the song link.

  6. Generate your AI cover!

Here’s our version of the trial.

  • In the recent episode of Leading with Data, I had an amazing conversation with Mark Landry, Director of Data Science & Product at H2O.ai, about his views on advancements in GenAI, agentic AI, and the evolution of AI technologies, and his experience in competitions, multimodal training, and developing award-winning AI applications.

  • This course by DeepLearning.AI provides hands-on experience in building LLM-based applications, guiding you to create an AI-powered text-based game from scratch while learning app development fundamentals, hierarchical content generation, and integrating safeguards with tools like Llama Guard.

  • This introductory course by Analytics Vidhya provides a concise overview of Agentic AI systems, covering their evolution, current state, and practical applications.

To be fair, it is almost impossible to tell what is a reasonable expectation and what is inflated in today’s world!

There are days when I feel every piece of software will be re-written in the next 3 years and AI will be at the center of this new creation. But then, there are days when I get reminded that changes of this magnitude take time.

Would love to hear how you guys are feeling about the expectations from AI and Big Tech companies!

How do you rate this issue of AI Emergence?

Would love to hear your thoughts

Login or Subscribe to participate in polls.

Reply

or to participate.