Excellence Intelligence

DeepSeek's recent developments have ignited significant discussion in the AI community,?and I wanted to take a minute to share some thoughts.?If you haven’t heard,?the company's latest model, R1, showcases a reasoning capability comparable to OpenAI's o1, but with a notable distinction: DeepSeek claims that their model was trained for much less cost. ? It isn’t clear yet if DeepSeek is the real deal or a DeepFake, but regardless of what we learn in the coming days, it’s clear that this is a wake up call -- the path of bigger and bigger LLMs that rely on ever-increasing GPUs and massive amounts of energy is not the only path forward. In fact, it’s clear there is very limited upside to that approach, for a few reasons: ? ???First, pure scaling of LLMs at training time has reached the point of diminishing or maybe near zero returns. Bigger models trained with more data are not resulting in meaningful improvements. ? ???Further, enterprises don’t need huge, ask-me-anything LLMs for most use cases. Even prior to DeepSeek, there's a noticeable shift towards smaller, more specialized models tailored to specific business needs. As more enterprise AI use cases emerge, it becomes more about inference -- actually running the models to drive value.?In many cases, that will happen at the edge of the internet, close to end users.?Smaller models that are optimized to run on commodity hardware are going to create more value, long-term, than over-sized LLMs. ? ???Finally, the LLM space is ripe for optimization. The AI models we have seen so far have focused on innovation by scaling at any cost. Efficiency, specialization, and resource optimization are once again taking center stage, a signal that AI’s future lies not in brute force alone, but in how strategically and efficiently that power is deployed.

Last year, I took part in an AI agent hackathon in San Francisco. Reflecting on that experience, here’s why I’m not surprised markets are bouncing back after Monday's selloff: ? ???DeepSeek's Breakthrough: #DeepSeek, a new LLM startup, has demonstrated that high-performing models can be trained with less compute than we previously thought. However, this doesn't disrupt the existing AI investment paradigm. ?? ? ???The Bigger Picture: Last-mile training costs are just one piece of the puzzle. DeepSeek hasn't disclosed the costs of previous model iterations or pre-training runs. With 196 contributing engineers on their paper, the investment in talent alone is significant. ?? ? ???Next-Gen Models & Agents: While DeepSeek has developed a functioning current-gen model, they're not yet on the path to creating next-gen models. Consumers want to be surprised and delighted by what tech can do, and the bar keeps raising higher. The race is still on to build AI AGENTS that do more than just generate text—they'll take on complex tasks—I’m excited to see real solutions shipping this year. ?? ? ???Infrastructure Needs: Building these advanced tools requires more than just an LLM. It demands corporate partnerships, robust security and privacy measures, and substantial investment in both people and compute power. ??? ? ???Explosion of AI Tools: At last year’s hackathon, my 4-person team spent 6 hours building a sales training agent using a tech stack including LLM, RAG, voice APIs and lots of quick & dirty Python code. Today, we could do probably do it in 3 hours with twice as many tools at our disposal. The pace of innovation is staggering. Venture capital funding is allowing an explosion of ideas to take shape in real time. Some of the most exciting startups use proprietary AI models (and not just LLMs) that need serious computer power to develop and train. ? ? ???Powering the Future: The demand for electricity in AI is driven by the need to train next-gen models, as well as wider adoption. Even with compute efficiencies, more power is crucial for staying at the cutting edge. AI data center demand is massively driving US electricity consumption right now. ?? ? I'm excited about what cheaper models mean for everyone in the venture ecosystem. Cheaper models = Better productivity for all. ?? ? You can also access this slide in our #GuideToAlternatives and more of our insights on AI -- linked in the comments!

To our clients and partners all around the world, we wish you a happy and healthy holiday season! ? As we reflect on this past year, we're grateful for the continued collaboration—it’s essential for driving progress. Together, we've achieved great things and pushed the boundaries to deliver real results—from driving impact with AI and modernizing environments to zero trust for maximum security or preparing for the post-quantum future. And to our incredible DXC team, THANK YOU. We work with some of the largest brands in the world and they trust us to take them to the next chapter of their transformation journey. The difference is in the doing! #WeAreDXC

The data that's being collected, now more than ever needs to be connected. In our final installment of AI Insights, @KathleenWalker from Elastic and I wrap up our discussion and emphasize the power of connecting the right data with the right technology. With the precision of search and the power of AI, you get your secret sauce, Search AI, that helps you do exactly that. Tune in to discover how Search AI empowers you to find the right answers within a proprietary data set to mitigate security risk, improve operational resilience, and enhance customer experiences. #AIInsightsSeries #GenerativeAI #AI #Search #PowerofSearch #LinkedInSeries #SearchAI

Eva Tukdarian, CPA is #hiring. Know anyone who might be interested?

OpenAI last week removed Aleksander Madry, one of OpenAI's top safety executives, from his role and reassigned him to a job focused on AI reasoning, sources familiar with the situation confirmed to CNBC. Madry was OpenAI's head of preparedness, a team that was "tasked with tracking, evaluating, forecasting, and helping protect against catastrophic risks related to frontier AI models," according to a bio for Madry. The decision came less than a week before a group of Democratic senators sent a letter to OpenAI CEO Sam Altman concerning "questions about how OpenAI is addressing emerging safety concerns." The letter, sent Monday and viewed by CNBC, also stated, "We seek additional information from OpenAI about the steps that the company is taking to meet its public commitments on safety, how the company is internally evaluating its progress on those commitments, and on the company's identification and mitigation of cybersecurity threats." (CNBC)