E12: AI's $1.5B Week: Spatial Intelligence Boom, EU Regulation Showdown, Meta vs Data Rules
In this episode, the hosts discuss several developments in the AI space, including the launch of a billion-dollar startup by Fei-Fei Li, the funding round of Anthropic and OpenAI, and the attraction of AI companies to Japan due to lax copyright rules. They explore the importance of spatial intelligence in manufacturing and transportation, the challenges of deploying AI in enterprise settings, and the competition between European startups and overseas giants. The conversation also touches on the financial engineering behind AI investments and the potential for distribution partnerships. In this conversation, the hosts discuss the impact of government regulations on the location decisions of companies, particularly in the context of AI development. They explore the Japanese government's efforts to attract AI companies by offering more relaxed regulations, as well as the potential drawbacks and challenges of such a strategy. They also touch on the upcoming AI Act in the European Union and its implications for companies operating in the region. The conversation highlights the tension between regulation and innovation, and the need for companies to navigate complex regulatory landscapes. Like what you hear? Remember to smash that subscribe button for more insights every week!
Takeaways
- The AI space is experiencing rapid developments, with new models and startups emerging regularly.
- Spatial intelligence is becoming increasingly important in manufacturing and transportation, enabling AI to navigate and optimize real-world environments.
- Deploying AI in enterprise settings requires overcoming challenges related to integration, compliance, and long sales cycles.
- European startups face competition from overseas giants in the AI space, necessitating a focus on specialization and building trust with enterprise customers.
- Financial engineering plays a role in AI investments, with strategic deals and partnerships driving revenue growth and valuation.
- Lax copyright rules in Japan are attracting AI companies to set up base there for training their models. Government regulations can significantly impact the location decisions of companies, particularly in the field of AI development.
- The Japanese government is trying to attract AI companies by offering more relaxed regulations, but there are concerns about the long-term viability of this strategy.
- The upcoming AI Act in the European Union is stirring controversy and discussion, with some viewing it as rushed and vague.
- Companies may need to navigate complex regulatory landscapes and consider factors such as data storage, jurisdiction, and compliance when deciding where to be based.
- There is a tension between regulation and innovation, and finding the right balance is crucial for fostering technological advancements.
Episode Transcript
Introduction
Rod: Welcome to another episode of the Chris Rod Max show, where we discuss trends in AI, the latest developments, and their implications for enterprise and society. I'm joined by my co-hosts, Chris and Max.
Chris: Hello, everyone.
Max: Great to be here.
Rod: This week, we have numerous developments in both the regulatory space and in funding and development of new models. For example, Meta recently announced a new model, Llama 3.1, showcasing how rapidly the field is evolving. We'll be discussing two significant funding rounds, regulatory developments in Europe and Japan, and the situation in China regarding AI model restrictions. Let's get started with our first topic.
Fei-Fei Li's Billion-Dollar Startup
Rod: Fei-Fei Li, a renowned Stanford professor and one of the most prominent figures in AI, has launched a billion-dollar startup called Lore Labs. This is a remarkable achievement, even by AI industry standards. Max, can you explain why this is so significant?
Max: Certainly. Fei-Fei Li's venture into spatial intelligence is fascinating. It reminds me of Arnold Schwarzenegger saying, "Hasta la vista, baby." Jokes aside, this move into the spatial realm is intriguing because it touches the real world, particularly manufacturing and logistics. It enables AI to navigate what we experience in the physical world, unlike many AI models that operate in the digital realm.
This is exciting because it impacts a significant portion of the world's GDP, especially in manufacturing. We're seeing a global trend of countries trying to onshore their supply chains, and having AI models to enhance manufacturing processes is crucial. It could really augment how people work in manufacturing.
Another implication is in transportation. Think about Tesla's AI and LIDAR sensors detecting everything around them. This technology could transform how transport and infrastructure work. From a UK perspective, with our issues around trains and congestion, spatial intelligence could help revolutionize planning for the future.
Rod: You've made an excellent point about the shift from abstract language models to something very tangible like manufacturing. Chris, you're based in Germany, known for its manufacturing prowess. Are you seeing similar developments with companies using AI for manufacturing cases?
AI in Manufacturing and Industry
Chris: The space of spatial intelligence isn't entirely new. Even five or six years ago, we saw pitches around using this technology for navigation in complex environments like shopping malls or airports. There have been startups developing technologies around virtual reality, augmented reality, and machine learning for spatial understanding.
Regarding Fei-Fei Li's startup, I couldn't find much information beyond their branding website. I'd be really interested in understanding exactly how their technology works. My guess is that it's coming from a computer vision angle, similar to what I worked on a few years ago in a startup focused on understanding and modifying objects in images for commercial real estate.
Rod: Indeed, this combination of spatial AI and the concept of the metaverse, which was a buzzword until recently, is interesting. It involves navigating 3D environments, which could potentially connect these technologies.
Max, from an investor's perspective, how do you view the rollout and go-to-market strategy for these types of companies? They require much more than just deploying a website; they need to be installed or available in remote factories around the world.
Max: In the enterprise space, what I've seen work successfully is having design partners. A startup will work with a few large manufacturers to design the product according to their specifications. This approach entices both the investor and the potential customer. The risk is over-indexing on one customer's problems, but from a go-to-market perspective, this has worked well for early-stage startups.
Anthropic's $500 Million Funding Round
Rod: Let's move on to our second development. Anthropic, one of the better-known AI model development companies, announced that it raised 5.5 billion valuation. Chris, can you put this into perspective?
Chris: Anthropic's ChatGPT is known for general question-answering, while Claude excels at summarization, and Perplexity is great for specific knowledge questions. Each of these chatbots has strengths in specific tasks and domains due to the data they're trained on.
Anthropic seems to be focusing on helping businesses succeed by "speaking the language of business." For example, they might extract more detailed financial data from reports, such as Compound Annual Growth Rate (CAGR), which is more valuable than a high-level summary.
However, from my conversations with enterprises, the real challenge isn't necessarily the model itself, but the workflows to implement it. This is something we've discussed often on our show.
Rod: That's an interesting point about the specialization of German and European companies in catering to business customers. Now we're seeing these giant players from overseas trying to do similar things but with much larger financial resources. How do you see this competition playing out?
Chris: In the German context, companies like Anthropic would be competing with established players like SAP, Microsoft, or Google. There's often hesitancy to let smaller companies or startups access core company data. Anthropic would need to build trust and a strong brand to crack the enterprise and B2B market.
The recent failure of Aleph Alpha, the German AI competitor to ChatGPT, doesn't bode well for other startups trying to enter the German and European markets. We're talking about compliance issues, integration challenges, and very long procurement cycles where security and functionality testing are crucial.
Valuations in AI Startups
Rod: Max, Anthropic reportedly has 5.5 billion. How do you view this valuation from an investor's perspective?
Max: Valuation is relative and depends on overall demand and the amount of capital chasing these companies. What's interesting about Anthropic is that they have many corporate investors on their cap table, including Cisco, Salesforce, Oracle, AMD, NVIDIA, Fujitsu, and SAP. This suggests they're taking the design partner route with large corporates, which has direct implications for valuation.
If these larger corporates decide to use Anthropic over ChatGPT, it will directly impact revenue and overall valuation. In contrast, OpenAI's investors are primarily VCs, creating a different dynamic.
From a valuation perspective, it's relative. The fact that Anthropic has several corporates on their cap table signals demand, not just from an investment standpoint but also from a sales perspective, which has more direct implications for valuation.
Rod: How much financial engineering do you think is taking place here? Are these companies providing cash resources or computing credits for their hardware and data centers?
Max: From a financial engineering perspective, I don't think you can play much with the P&L or balance sheet. However, you can structure more strategic deals operationally, such as offering certain pricing to large corporates. These deals often focus on increasing revenue, reducing costs, or reducing risk for the corporate partners.
I haven't seen much changing of balance sheets or revenue recognition, but more of helping to enhance the overall corporate's offering to their core constituency. These partnerships could lead to integrations with existing products and services, potentially increasing revenue for both parties.
AI and Copyright in Japan
Rod: Moving on to our third topic, there have been reports that laxer copyright rules in Japan are attracting AI companies to set up base there for training their models. However, creators in Japan are concerned about the potential threat to their industries. Chris, how do you see this strategy of using relaxed regulations to attract innovation?
Chris: Traditionally, companies might relocate for tax relief or to be closer to potential customers. In this case, Japan is advertising the ability to use their data to train AI models. From a Japanese point of view, I understand the reasoning. We've heard a lot about AI development in the US and Europe, but not much about Asia beyond China.
It's a smart move to think about how to attract talent and perhaps build something more Asian or Japanese in terms of AI models. We've spoken about language barriers and how well ChatGPT performs with local dialects. This could be Japan's intention – to foster AI development that's more tailored to their region.
However, I'm skeptical if this is the right long-term strategy. The data might be consumed quickly, and then these AI companies will need more fuel. Being strategically located in Japan might not be advantageous in the long run, considering the high costs and language barriers.
The EU AI Act
Rod: Let's discuss the AI Act in the European Union, which is coming next month and stirring controversy. It's being criticized as imprecise, rushed, vague, and potentially stifling to innovation. Chris, as someone in the EU, have you heard reactions from companies about this upcoming legislation?
Chris: There are actually a couple of startups addressing this, helping companies comply with AI regulations using AI itself, which I found quite ironic. In general, I believe the AI Act serves a good purpose and has good intentions. The challenge lies in the details of implementation and compliance.
Rod: Do you see a risk of companies hiring more lawyers than coders just to navigate this legislation?
Chris: Absolutely, 100%. We used to have data protection officers, and I'm sure we'll see similar roles for AI compliance. Conceptually, having regulations in place makes sense, but whether this is the best approach remains to be seen.
Max: From a regulation perspective, I believe it's pretty rushed. My view is that we should let the market do its thing and then regulate based on what emerges. Rushing to regulate could stifle innovation. We need the ability to experiment and try different things to get innovative results.
Many of these regulations are based on our understanding of AI today, but AI will likely be very different in 5-10 years. If we had heavily regulated the internet in its early days, we might not have the capabilities we have now.
My concern is that this will put an extra burden on AI companies, especially startups. It could divert resources from development to compliance, which might not be the best use of capital.
Chris: On the other hand, I think it's quite positive that the EU was able to come together quickly and produce this act despite the different interests of member countries. The intention is good – they want to ensure safety and prevent exploitation. I hope and trust that the decision-makers will also consider how to keep the EU competitive in this field.
AI Regulation in China
Rod: We also have news from China, where AI companies like ByteDance, Alibaba, and Tencent need to go through regulatory approval for their AI models. This raises questions for multinational companies that need to comply with different local rules. Chris, are you seeing large organizations planning to segment their AI deployments to adjust to these various restrictions?
Chris: Definitely. This is on people's minds in terms of data storage, company headquarters location, and under which jurisdiction they operate. These are crucial questions for B2B companies. From customer interviews, we've learned that one of the biggest factors they consider is the location of the company and its data centers.
Another important factor is how recognized and popular a company is, especially in the US market. It serves as a proxy for how many people use the service and the ability to complain if something goes wrong. The recent incident with CrowdStrike affecting aviation and financial systems illustrates the importance of company location, jurisdiction, and customer base size in decision-making.
Meta's Complaint About EU Regulations
Rod: Lastly, Meta is complaining about EU regulations, particularly the request to stop training their language models with European data. They argue this will stifle innovation. Max, how do you see this from an investor's perspective? Could this affect portfolio companies or those you're considering working with?
Max: If companies aren't allowed to use data, it raises the question of why they're in that location at all. Any AI company needs to use data. This will directly impact the competitiveness of setting up AI companies within the EU. It might be good for the UK or the US, as companies might move there for both market size and ease of doing business.
I find the idea of not using European data hard to enforce. Europe is like the world's library, with a wealth of historical and cultural information. Training AI models will inevitably touch on EU-related data. Differentiating between EU and non-EU data seems challenging and could create more problems than solutions.
We are seeing startups move if it becomes harder to do business in a particular location. They're already trying to do something difficult, and additional regulations make it even harder.
Conclusion
Rod: Thank you both for these insights. We've covered massive funding rounds in spatial intelligence and language models, regulatory developments worldwide, and the potential impact on innovation and company locations. To our listeners, remember to like, subscribe, share, and provide feedback. We're here to listen and adjust. Looking forward to next week's show!
Chris: Thank you, everyone. See you next time!