This interview was extracted from an interview with Jensen Wu during the Impulso Podcast E112 “How AI humans are killing marketing.”
[00:00:56]
You spent 13 years at Alibaba before founding Topview.AI. What major technological shifts have you seen shape the AI industry during that time?
Jensen:
If we look at AI development, we can divide it into two eras—before and after 2022. Before 2022, AI was focused on narrow, specialized applications using deep learning to solve specific problems. Models were trained for singular tasks, such as facial recognition or speech-to-text. The paradigm shifted dramatically after 2022 with the rise of large language models (LLMs). Suddenly, AI was no longer just about performing individual tasks—it became a universal tool that could generate text, images, and even videos.
ChatGPT’s launch marked a turning point, accelerating both public awareness and industry investment in generative AI. The most significant change has been the rapid iteration cycle. Before, AI advancements took years. Now, every six months—sometimes even three months—a new foundational model is released. This has led to an explosion in AI-generated content, pushing applications in e-commerce, marketing, and creative industries to a whole new level.
But while capabilities have improved, the challenge now lies in integrating AI effectively. Companies that fail to adapt risk being left behind, while those that leverage AI strategically are gaining massive competitive advantages.
[00:02:39]
The AI industry is becoming more competitive. Do you feel the pressure to keep up with the competition, or do you see it as a chance to innovate and stay ahead?
Jensen:
It’s both. On the one hand, AI applications are benefiting tremendously from the rapid advancements in foundational models. A year ago, using OpenAI’s API cost $10,000 per month. Today, that cost has dropped to $1,000 for equivalent performance. This has democratized AI, making it accessible to more startups and businesses.
However, on the other hand, the fast-moving landscape means companies must constantly adapt. For example, last year, we heavily relied on OpenAI’s models. Then, we found that Gemini performed exceptionally well, so we switched. But as soon as GPT-4.5 was released, we had to reconsider. AI companies can no longer rely on a single provider. Instead, we have to continuously evaluate performance, cost, and capabilities to stay ahead.
Another aspect is market pressure. Investors, clients, and even employees all feel the urgency to move fast. If we don’t integrate the latest advancements, someone else will. This constant state of flux creates stress, but it also means that new opportunities emerge every few months. The companies that thrive are the ones that can navigate this cycle effectively.
[00:12:50]
Will AI contribute to making the Metaverse more realistic?
Jensen:
Yes, and not just realistic—AI will make the Metaverse feasible in ways that were previously impossible. In the past, creating virtual environments required manual 3D modeling, motion capture, and scripting. This made development expensive, slow, and often lacking realism.
AI is changing that. We’ve already seen AI evolve from generating simple digital humans with basic facial movements to creating full-body animations with natural gestures. The next step is integrating AI into fully interactive 3D environments. Once AI can generate high-quality 3D spaces in real-time, the barriers to Metaverse adoption will dramatically decrease.
AI-generated 3D Animation Sample
Right now, most AI-generated content is still in 2D. But within a few years, we will see AI that can create entire virtual worlds on demand. This will revolutionize gaming, social interactions, and even e-commerce. Instead of looking at a product image online, you’ll be able to walk into a virtual store, interact with an AI salesperson, and see how an item looks in different lighting conditions.
Some people ask whether AI will replace traditional Metaverse development. I don’t think it will replace everything, but it will augment and accelerate it. The true Metaverse won’t just be manually built—it will be AI-generated, adaptive, and deeply personalized.
[00:14:04]
The AI industry is becoming increasingly competitive. Do you feel the pressure to keep up with this fast-paced evolution, or do you see it as an opportunity to innovate? Additionally, many businesses have rushed to adopt AI, but what do you think is the biggest challenge in successfully integrating AI?
Jensen:
Both pressure and opportunity coexist in this environment. AI is advancing at an unprecedented pace—what was state-of-the-art six months ago is now outdated. This constant evolution means we must stay agile, continuously testing new models and technologies. For example, just last year, OpenAI was the clear leader, but now Gemini and other models have emerged as strong competitors. We don’t just passively wait for new tools; we actively experiment, switching between different models based on cost, efficiency, and capabilities.
At the same time, the biggest challenge in AI adoption is expectation management. Many companies jumped into AI expecting instant transformation, investing heavily in proof-of-concept projects without a clear strategy. When results didn’t immediately match the hype, disappointment followed. The reality is that AI is not a magic wand—it’s a tool. Successful companies are the ones that integrate AI where it truly enhances workflows rather than treating it as a vague, futuristic solution.
Another major challenge is talent. There’s a huge gap between the demand for AI expertise and the available workforce. Many businesses don’t have in-house AI specialists, which makes adoption difficult. This is why AI-as-a-service models are growing in popularity, allowing businesses to leverage AI without needing to build complex systems from scratch.
Ultimately, AI presents immense opportunities, but only for those who approach it strategically. The winners in this space won’t necessarily be those who adopt AI first, but those who adopt it wisely—aligning it with real business needs, continuously iterating, and staying adaptable in an ever-changing landscape.
Another challenge is talent. Many companies want to use AI, but they don’t have in-house expertise. Hiring AI specialists is expensive, and many businesses struggle to build the right teams. That’s why companies that offer AI as a service—providing plug-and-play AI solutions—are seeing massive demand.
The businesses that succeed with AI aren’t the ones that simply adopt it for hype. They are the ones that carefully identify where AI adds value, invest in the right talent, and iterate their AI strategies based on real-world feedback.
