## The rise of AI Inference: An Interview with Thomas Sohmers
This week on Archyde, we delve into the burgeoning world of AI Inference with Thomas Sohmers, CEO of PositronAI.
**Archyde:** Mr. Sohmers, yoru recent predictions suggest a monumental shift in the AI computing landscape towards inference. could you elaborate on this trend and its importance?
**Thomas Sohmers:** Absolutely. While training AI models has been the primary focus, we’re now witnessing a surge in demand for inference, which is essentially the “thinking” part of AI. By 2024,we anticipate most AI computing spending will be directed towards inference,and this growth will be exponential.
**Archyde:** This paradigm shift naturally presents opportunities for emerging players in the AI chip market. How do you see smaller companies like PositronAI positioning themselves to capitalize on this trend?
**Thomas Sohmers:** Companies like ours are laser-focused on developing specialized hardware and software solutions optimized for inference workloads. We believe this niche focus will allow us to deliver unprecedented performance and efficiency, attracting developers and businesses seeking to leverage the full potential of AI.
**Archyde:** Nvidia, the current market leader, acknowledges this shift and anticipates benefits for all inference chip suppliers by 2025. Do you see the market evolving into a more diverse landscape with increased competition?
**Thomas sohmers:** Absolutely. The arrival of specialized inference chips will undoubtedly create a more competitive and dynamic market. This is ultimately beneficial for innovation and will drive the progress of even better AI solutions.
**Archyde:** While NVIDIA currently dominates the AI market, this evolving landscape presents an interesting possibility for other players to challenge the status quo. do you believe we are on the verge of a more decentralized AI ecosystem?
**Thomas Sohmers:** Time will tell, but the signs certainly point in that direction.
**Archyde:** What are your thoughts on the ethical implications of this rapid advancement in AI inference? Should there be increased regulations and oversight?
**Thomas Sohmers:** This is a crucial discussion. As AI becomes more powerful and pervasive,we need to ensure its responsible development and deployment.Transparently communicating the capabilities and limitations of AI, addressing bias, and promoting inclusive development practices are paramount.
**Archyde:** Thank you for your insights, Mr. Sohmers. It’s clear that the AI landscape is undergoing a interesting transformation. we encourage our readers to share their thoughts on the future of AI inference and its potential impact on our lives.
## Teh Rise of AI Inference: An Interview wiht Thomas Sohmers
**Archyde**: Welcome to the show, Thomas. Today we’re diving into the exciting world of AI inference and how it’s changing the game for big tech.
**Thomas Sohmers**: Thanks for having me. It’s an exciting time for the industry.
**Archyde**: You’ve stated that by 2024, most AI computing spending will shift towards inference. This seems like a dramatic prediction. Can you elaborate on why you’re seeing this trend?
**Thomas Sohmers**: Absolutely. Essentially, while training large AI models requires immense computational power, the real value lies in using those trained models to deliver results. That’s where inference comes in. We’re seeing a huge increase in applications that require real-time, on-demand responses from AI—think chatbots, autonomous vehicles, personalized recommendations, and more.
This translates to a surge in demand for inference processing power, driving the spending shift. [[2](https://iris.who.int/bitstream/handle/10665/371097/9789289059466-eng.pdf)]
**Archyde**: This shift seemingly creates opportunities for smaller players in the AI chip market. How are companies like Positron AI capitalizing on this?
**Thomas Sohmers:** You’re exactly right. The traditional players like Nvidia have focused largely on training. now, the emergence of specialized inference chips allows us to offer solutions that are optimized for speed, efficiency, and cost-effectiveness. We’re seeing a real opening for innovation and competition in this space.
**Archyde**: Nvidia’s CEO, Jensen Huang, has acknowledged the positive impact of these advancements. He even stated that post-training strategies will benefit all inference chip suppliers by 2025. What’s your take on Nvidia’s position in this evolving landscape?
**Thomas Sohmers:** I see Nvidia’s acknowledgment as validation of the trend towards inference. they recognize the coming wave and are adapting accordingly. However,while they have a strong presence in the market,the demand is so vast that there is room for multiple players to thrive.
**Archyde**: Looking ahead, what are the biggest challenges and opportunities you foresee in the AI inference market?
**Thomas Sohmers:**
One major challenge is ensuring the security and reliability of AI systems as they become more ubiquitous.We need to address concerns about bias, explainability, and data privacy.
On the chance side, I see massive potential for AI inference to transform industries like healthcare, finance, and manufacturing. We’re just scratching the surface of what’s possible.
**Archyde:** Thank you for your insightful outlook, Thomas.This has been a fascinating discussion. We look forward to seeing how the AI inference landscape unfolds.
**Thomas Sohmers:** The pleasure was all mine.
