Thank you, Alan. And thank you to all of the board members of C3AI who are here at this show today. Could I ask the board members to stand up so we can see that you're here? I don't know of any other company where when the CEO says I have a travel restriction due to health issues, the board then comes along and everyone comes to the event to be with the customers. This is the essence of resilience of a company, and resilience does matter these days because we live in a crazy world where being resilient is absolutely critical. Now, I've met many leaders in my career, but I have to say that Tom Siebel is by far the most resilient leader I have ever seen. I wanted to use this moment to thank you, Tom, for inspiring us all with the kind of leadership and imagination that drives the future and the conversation these two days. You may not be here physically, but you certainly are in spirit. Thank you, Tom.
Now, I also learned about resilience quite early in my life. When I was two years old, I moved to Greenland – it's getting a very popular place these days. I remember Greenland as a very tough environment: cold, dark, and full of snow and ice, like you see in this picture. It was not green at all. I actually believe the name 'Greenland' is the first example in history of greenwashing. The good news is I might get a green card really soon – you never know. One of the things I learned during seven years of childhood in Greenland was how to overcome a snowstorm. Some people think the best way is to dig in and wait for it to finish – you'll probably survive. But experienced explorers don't just survive; they move on based on three important leadership principles. Number one: clarity of direction. In a snowstorm, everything turns white; you have no idea where you're going. Without clarity, you may go in circles and not survive. Number two: focus on what really matters. You don't spend energy on things that don't matter; you focus all your energy on the few things that get the job done. Number three: team up for success. If you're alone, chances of survival are slim; together, your chances grow dramatically.
So I believe the world today is a little bit like a snowstorm on Greenland. Of course, you can dig in and wait for it to pass, or you can lead with clear direction. The same is true for AI. I agree with Tom that we may be at a moment where the snowstorm around AI needs to settle so we can focus on what really matters. The hype is over. AI is no longer about hiring a few data scientists for a project in a corner of the company. While ChatGPT is impressive at writing poems and emails, it cannot run your company. I believe this is a moment for a clear direction: how do we make AI not a project or experiment but the critical infrastructure of your company at speed and scale? To do that, I try as a leader not to get bogged down by all the noise. I spend very little time on the present – it's a distraction. I spend most of my time on the past and the future: learning from the past to constantly improve, and imagining the future so I can participate in shaping it instead of becoming a victim. I often say to my teams, you need to be at the table, not on the table.
To do that, I went back in history to find technologies that had a profound impact on civilization. One of the most profound inventions was the printing press. In the 15th century, if you wanted to share an idea, you either talked to someone or wrote a book by hand – meaning knowledge was limited to the few. Then Johannes Gutenberg imagined a future where books could be printed at low cost and knowledge distributed as never before. He invented the printing press, and everything changed. Suddenly knowledge and ideas were available to everyone everywhere at very low cost. It democratized knowledge. This changed governance, businesses, societies, and even inspired revolutions. What can we learn from that? First, some technologies have the power to change everything – the printing press was one, and I believe AI will have the same opportunity. In the case of the printing press, the limitation was not the invention itself but the learning that needed to happen: everyone had to learn to read and write. Entire countries transformed education systems from being for the few to being for the many. If you didn't learn, you were left behind.
I believe the same characteristics will happen with AI. Like the printing press democratized knowledge, we are now on the edge of AI democratizing intelligence. Imagine what that will do to the world. But there is one big difference: speed. It took 400 years from the invention of the printing press until 80% of the US population could read and write, because we had to educate people to use the technology. The same was true for computers and the internet – it took decades to democratize. With AI, ChatGPT reached a million people in five days. Why? Because this technology is different. So far, we invented technology and then taught users how to use it. With AI, we teach AI how humans act. We taught it to be human. It communicates with us through vision, sound, gestures. This is the first technology in history where usage will be instant. The limitation is no longer educating the user – the limitation is now on us: our ability to deploy this technology. That is our challenge.
To do that, we need to imagine the future like Gutenberg did. With AI, we need to imagine a future where intelligence is available to everyone. It's hard to imagine the future, so I ask 'what if' questions. Number one: what if AI became super intelligent – not just average with hallucinations, but super deep expertise, better than humans in multiple domains? That's already happening: in healthcare, AI is better at diagnosing cancer from images; in protein folding, AI outperforms humans; it plays better chess and Go. Imagine if that happens in most areas of expertise. Number two: what if AI is not just about answering questions but can take action – perform tasks? Instead of asking when an asset will fail and what to do, the AI will analyze the asset, identify needed maintenance, order parts, and execute the repair. You will see examples of that at this event. Number three: what if AI is everywhere – not just on our screens and phones, but in our watches, glasses, cars – embedded in every piece of hardware, intelligent itself, making autonomous decisions? That is already happening in autonomous vehicles. What if that becomes true everywhere in our physical world?
If these three 'what ifs' become true – which I think is likely – you have to ask: what would that do to my business? Taking Siemens as an example, that future would be pretty cool. If intelligence is better than humans, can take action, and is in every physical product we make, Siemens will become dramatically more productive, more predictive, and more innovative. On productivity: think about all the time we waste on emails and expense reports. Soon, one AI will talk to another AI in email and we won't even need to be involved. 80% of what we do in a normal business day could be done by AI agents. Imagine a company with half of its 'employees' being AI agents – doing all the low-value work so humans can focus on what really matters. On predictiveness: think of all the waste when things don't go as planned. I chaired Maersk, the world's largest shipping company – no plan ever went exactly as planned. We constantly had to replan. Imagine if 80% of unexpected events could be predicted and prevented – maximizing asset utilization, creating supply chain resilience, detecting fraud in real time. And I still hope boardrooms will become more predictive, so we spend time on the future instead of reporting the past.
On innovation: the biggest value will come from generative AI creating ideas. Companies using digital twins in product design already see a 50% reduction in time to market. Now imagine adding generative AI to the innovation process – having a conversation with your PLM system: 'Help me design an EV sports car with four seats, 500-mile range, drag coefficient of 0.3, that looks a little better than a Porsche 911.' That would revolutionize innovation. These examples are not just talk – they are happening now, and many are in this room. I thank all of you sharing your successes with AI in productivity, predictability, and innovation at this conference, because we all need to learn from each other faster. The key question is: how do we deploy AI much faster to get the real business value, not just hype?
To do that, we need to focus on what really matters: moving AI from projects to infrastructure – thinking of AI as a utility that is intelligent, can take action, and is everywhere in the company and its physical presence. We need to focus on three things. First, master AI ourselves. Yes, we can buy large language models, but for domain-specific knowledge unique to your company, you need to train your own AI – that is your differentiation. Second, orchestrate across multiple AIs to get consistency and relevance. Third, accelerate deployment because we, not the user, are the limitation on AI's value.
To master AI, we need to understand what's happening. I remember the launch of ChatGPT – I was on stage here talking about how my son and I were inventing a song about destrates (deserts?) when making pizza. A lot has happened since. Last year people argued there would only be four or five LLMs because training costs hundreds of millions. That sounded like IBM in the 50s saying there would only be four or five mainframes. Then DeepSeek came along claiming training costs of five to six million dollars – a factor of ten cheaper – and it's open source. Suddenly every company can consider building their own LLMs. This doesn't mean we need less compute; we'll need even more because everyone will do it. I call this the language layer – it interacts with humans but cannot run your company because it's not intelligent enough. Three weeks ago, I was doing my expense report. I uploaded five receipts to ChatGPT and asked it to write an email converting everything to US dollars. The sum didn't add up; I had to go through five iterations and still had to adjust by ten dollars. You want to run your company with that? Good luck. LLMs learn math through language – that's an expensive way to learn math. What you need are specialized AIs that train only on math, or protein folding, or predictive maintenance – domain-specific systems that are deep and accurate. That is the knowledge layer – your company's unique data becomes true differentiation.
Now, if you master your own domain-specific AIs, you end up with many AIs – language models from the market and your own specialized ones behind the firewall. Now you need orchestration. The user asks a question through an LLM, but the answer needs to come from the right specialized AI – perhaps multiple ones working in parallel. This orchestration layer – which C3 calls the agentic AI layer – acts as a conductor, understanding what each AI can do and distributing tasks accordingly. I believe this will be the crucial infrastructure of any AI future.
The last thing is accelerating adoption at speed and scale. You can decentralize for speed but pilots never scale; or centralize for scale but lose speed. You need both, through a platform approach. Be a dictator on choosing one enterprise AI platform so you can decentralize the use of the platform to create specialized AIs while orchestrating across them. The platform must be open (connect to any AI technology), have an agentic layer for orchestration, and be productive – meaning low software-to-services ratio. In the ERP world, $1 of software requires $10 of services. With C3, it's the opposite: $10 of software, $1 of services – because you learn to use the platform with your own skills. You cannot outsource AI for the enterprise – it is your destiny and competitive advantage.
So in summary, to come out of the AI snowstorm better and faster: number one, train your own AI in domain-specific areas for differentiation. Number two, orchestrate across all AIs for consistent, relevant infrastructure that supports your people and takes actions. Number three, accelerate deployment by choosing one coherent platform that lets you build and master your AI landscape. The last recommendation from the explorers is teaming up for success. You can do it all yourself and become a technology company, or team up with someone who has world-class technology so you can focus on using AI. At C3, we believe in teaming up – with partners and with you, our customers – to enable you to master your AI infrastructure.
Why is this so important? Like the printing press, AI has incredible potential to change everything in society, business, and governance. But it must be done consistently. We have the biggest opportunity ever to reinvent entire value chains. Look at automotive: after 100 years of optimization, we use a car 8% of its life; 92% it sits parked. With AI and intelligence everywhere, we can create an ultimate transportation system that is convenient, affordable, and sustainable. The same for energy: from centralized power plants to decentralized AI-driven systems on every roof, making energy nearly free. Healthcare is the biggest opportunity – 18% of US GDP on healthcare. With AI, we can individualize therapy, predict and prevent disease, creating healthier longer lives at lower cost. This is a leadership moment, an inflection point in history. For the first time, we have the opportunity to make intelligence available to everyone – every human on the planet, every person in your company. AI is about making businesses and societies better. We must have the courage, like Gutenberg, to imagine a future where intelligence makes our companies and societies better. That is the most important leadership skill in this time in history.
Let me finish with a quote that has inspired me my entire career. George Bernard Shaw said, 'Most people see things the way they are and ask why. I imagine things that never were and ask why not.' Thank you.
