‘Impossible for Any Big Company to Have Enough Imagination in AI’

Kevin Scott, CTO of Microsoft, in a podcast episode on Monday, discussed the AI startup ecosystem and how it will continue to hold immense potential to build products and services that large companies may not prioritise.  

“It’s impossible for any entity, like Microsoft or any other large company, to possess enough imagination and perspective to recognise every interesting possibility,” Scott said.  

One of the primary concerns about building products in the age of AI is the fear that a large company developing AI models will swallow up a startup along with the use case it serves. 

However, over the last few months, startups creating AI-enabled products on the model infrastructure layer have experienced remarkable success and growth. 

Scott indicated that a “vibrant” startup and a product ecosystem will continue to exist while the big entities make innovations in the model layer. 

“This is the best time to be alive if you have an entrepreneurial spirit,” he added. 

When AIM reported the rapid growth of startups, leaders from the industry echoed this success mantra. Besides, Mike Maples, founding partner at Floodgate, said that, unlike big companies, startups have an advantage regarding the high risk profile they can operate with. 

“Not having to be burdened by what could go wrong is a big factor in trying things that could go right,” he said. 

Maples also indicated that specialised domain expertise can create defensible positions where large companies cannot easily compete with their generic AI offerings, especially in complex, multidisciplinary fields.

Despite the growing ecosystem, concerns of big companies killing startups remain, and industry experts often need to reiterate the same. 

For instance, Cursor, an AI-enabled coding platform, earned the tag of the ‘fastest growing SaaS’ company after crossing $100 million in annual recurring revenue just 12 months after earning the first million. 

Soon after, Anthropic, the company behind the Claude family of AI models, released the ‘Claude Code’ agentic coding tool that works directly within the terminal. Several users opined that it is superior in performance as compared to Cursor. Naturally, many are left wondering what it means for the future of Cursor. 

Despite encountering similar threats in the past, what made Cursor successful, at least according to its founders, is the drive to innovate the product’s functionality rapidly. 

Cursor has also successfully defined a product-market fit that benefits from new model releases, enhancing the solution the company is already deeply engaged in providing. 

“You don’t realise [how the AI models] are working for you in every facet of the product, as well as the thoughtfully designed UX in every single feature,” said Aman Sanger, co-founder of Anysphere, while discussing how the company uses the powerful capabilities of foundational models in multiple layers of the product. 

Like Scott mentioned, startups are indeed focusing on numerous ideas that many incumbents aren’t considering. Recently, Y Combinator showcased a startup called Pickle that replaces a user’s Zoom camera with a virtual body double. Similarly, there’s another startup funded by Y Combinator called Miyagi Labs that uses AI to convert YouTube content into complete interactive courses. 

Moreover, CNBC recently reported that Y Combinator startups are the fastest growing and most profitable in the fund’s history, thanks to AI. 

More importantly, Scott highlighted the ease with which powerful foundational models can be accessed and deployed, at quite an inexpensive price. This enables several solo entrepreneurs to often go out on an idea spree, and end up monetising multiple ideas in quick time. For example, Marc Lou, one such ‘solopreneur’, is claiming to earn over $90,000 monthly with over 20 products that he has monetised. 

“AI has made us hyper-creative and hyper-productive. 100 startup ideas are running through my mind, and I can build all of them in 24 hours,” Lou said in a post on X.