Business Case Study

As an Investor, Analyst or Portfolio manager what would be your No.1 priority & expectations from AI ?

  

Summarization vs. Abstraction

Since the introduction of Generative AI, AI driven companies and technologies have been popping up everywhere with promises of AI influencing everyday life, from hiring, eduction, healthcare, law, financial, ...

  

Financial companies are incorporating qualitative data from Earnings Call Transcripts to improve their investment decisions. With more than 10,000 publicly traded companies on the U.S stock exchange and growing, the voluminous amount of earnings call transcripts to review is overwhelming for manual analysis. Several AI companies are coming out with products analyzing earnings transcripts with AI.

This past January, Bloomberg introduced a tool derived from Generative AI that promises to do just that effectively.

 

Generative AI uses algorithms such as word embedding, self attention and beam search in conjunction with backward regression to summarize narratives such as Earnings Call Transcripts. If it works, it would be a big win for investors. However, Generative AI does not understand the words that it uses in backward regression, nor can it understand the context and semantics of words that it generates. For it to summarize professional opinions and bring values to the summarization process remains a subject that is yet to be proven.

In recent days, there is other offering such as the one called "Knowledge" from TODADDA.IO, which begins to offer similar service. The following is an image from its website:

The approach to use Generative AI for summarization is directly challenging the foundation of Aristotelian logic where understanding requires deductive reasoning as expressed in syllogisms.

We believe human derived simplicity from complexity by means of abstraction; a process that is different from summarization. Deductive reasoning is required for explicit understanding and is essential to abstraction. On the other hand, summarization can be mimicked with stochastic processing using Large Language Model where no explicit understanding is required.

  

SiteFocus ELAINE is an implementation of symbolic logic. It applies deductive reasoning to perform abstraction similar to human's approach to abstraction.

 

We invited Dinakar Rajasekaran, a technologist who is familiar with the latest development in this type of application, to evaluated the solution from DOTADDA.IO and SITEFOCUS.

Source document for analysis: Nvidia Corporation (NVDA) presentation at Morgan Stanley Technology, Media & Telecom Conference Transcript

  

The following is the feedback from Dinakar:


Result from ELAINE


ELAINE zeroed in on 1 item H200 as focus areas in the Logical Analysis Report - 03/03/24 NVDA Morgan Stanley's Technology ...

"Key Focus H200 is that example, adding on to H100 and that part of it. What we see time and time again is when you are in a certain architecture and staying with the architecture, you have worked in terms of qualifying it within your systems, qualifying it within your software, within your security ..."

and identified a key challenge:

"Unknown Analyst, We are big fans of Jensen and NVIDIA being leading this AI revolution. I have two questions. One is regarding the longer-term prospects, right? So for example, some of your competitor, AMD and both TSMC has commented a long-term future and been quoting about $400 billion by '29."

 

Result from DOTADDA.IO


"Summary The discussion at Morgan Stanley's Technology event with NVIDIA's CFO, Colette Kress, covered NVIDIA's remarkable growth, particularly in data center revenue, driven by the demand for AI and accelerated computing, and strategies for managing supply chain challenges and future product launches."

"Metric/TopicValue/DescriptionDiscussion
Data Center Revenue GrowthFrom $4 billion to approaching $20 billion per quarter. Highlighted the significant increase in demand for NVIDIA's data center solutions, largely driven by the adoption of AI technologies like generative AI and large language models. "

"Supply Chain Management
Multiple strategies for ramping up supply. Discussed the importance of resiliency and redundancy in the supply chain, expanding capacity with existing suppliers, seeking new suppliers for redundancy, and improving manufacturing cycle times "

"Demand Fulfillment
Challenges in meeting the surging demandDespite efforts to ramp up supply, NVIDIA continues to face challenges in fully meeting the demand for its products, particularly the H100 GPU, due to the unprecedented interest in AI technologies. "

"Product Innovation Cadence
Acceleration from a 2-year to a 1-year cadence NVIDIA has sped up its product innovation cycle, allowing for quicker responses to market demands and technological advancements. This strategy is seen as a competitive advantage. Networking Revenue$13 billion run rate."

"NVIDIA's networking solutions, bolstered by the Mellanox acquisition, have seen strong demand, emphasizing the importance of networking in maximizing data center computing efficiency. "

"Software and Services Revenue
Reached an annualized $1 billion levelNVIDIA's growing focus on software and services, including AI platforms and SaaS offerings, contributes significantly to its revenue and is expected to be a key growth area moving forward"

"Gross Margin Forecast
Expected to normalize in the mid-70sAfter achieving high gross margins in the upper 70s, NVIDIA anticipates a return to mid-70s margins due to a mix of products being brought to market.AI at the EdgeSignificant opportunities identifiedNVIDIA sees considerable potential in AI at the edge, including in autonomous vehicles and robotics, leveraging platforms like Jetson for robotics and focusing on AI capabilities in PCs and workstations. "

"Government Spending on AI
Focus on training, with inference expected to follow. Initial government spending is primarily on training large language models specific to their regions, with inference applications anticipated as the next step."

"Multimodal Inference
Identified as a key growth areaNVIDIA views multimodal inference, which includes applications like text to video, as a significant opportunity for driving demand for inference solutions."

"Competition
NVIDIA's platform approach differentiates from competitors focusing on silicon. NVIDIA views its comprehensive platform approach, including computing infrastructure, networking, memory, and a full stack of software, as a key differentiator from competitors that may focus on individual components or chips "

" Accelerated Computing Market Size
Potential to exceed $1 trillionNVIDIA views the market for accelerated computing as potentially exceeding $1 trillion, driven by the transition from traditional computing to AI and accelerated computing solutions. Inference "

"Revenue Contribution 40% of revenues Highlighted the significant and growing contribution of inference to NVIDIA's revenues, underscoring the importance of efficient inferencing solutions in AI applications."

" Strategic Insights Rapid Innovation Cycle as a Competitive Edge:
NVIDIA's acceleration of its product innovation cycle from approximately two years to about one year positions it uniquely in the market. This rapid pace not only allows NVIDIA to respond more swiftly to technological advancements and market demands but also sets a challenging pace for competitors. The ability to innovate and launch new products at this speed could deter competitors and solidify NVIDIA's market dominance, especially in AI and accelerated computing sectors."

"Expanding Beyond Data Centers:
NVIDIA's focus on AI at the edge, including autonomous vehicles, robotics, and AI capabilities in PCs and workstations, represents a strategic expansion beyond traditional data center applications. This diversification could open new revenue streams and growth opportunities, positioning NVIDIA to capitalize on the increasing demand for AI and computing power across various industries and applications. "

"Software and Services as Key Growth Drivers:
NVIDIA's emphasis on growing its software and services revenue, which has reached an annualized $1 billion level, indicates a strategic shift towards offering comprehensive solutions that include both hardware and software. This approach not only enhances customer value but also creates recurring revenue streams and strengthens customer relationships. As AI and accelerated computing continue to evolve, NVIDIA's software offerings, including AI platforms and SaaS, could become increasingly critical to its overall growth strategy."

"Government Spending on AI:
The focus on government spending for training large language models specific to regions highlights a significant market opportunity for NVIDIA. As governments worldwide seek to develop AI capabilities that reflect their unique cultural and linguistic characteristics, NVIDIA's technology and expertise could be in high demand. This trend may also indicate a broader shift towards nationalized AI strategies, where countries prioritize the development of homegrown AI technologies for strategic and economic reasons."

  

Point to note : https://knowledge.dotadda.io Summary did not capture anything related H200.

 

Conclusion


As an Investor, Analyst or Portfolio manager what would be your No.1 priority & expectations from AI ?

  • A concise abstract providing key focus attributes like H200 here, to build on with your own reasoning and analytical models (or)
  • A summary enabled through LLMs/GenAI which appears as a trimmed down version of the transcript
  • A logical summary report with specifics connecting the key underpinning building blocks like Market - Software - Nvidia here

  •  

    Complimentary Membership is free to SiteFocus followers on the LinkedIn platform. Experience symbolic logic with your own documents. Click to join.

    More discussions on use cases:


    Managing Information in a Post LLM World