Over the past two years, we have heard much talk about the potential (and peril) of AI, and how Singapore should embrace (or take control of) it.

The government announced a national initiative in 2017 with committed investment of S$150 million (US$113 million). A coordinating body, called AI Singapore, was formed.  

AI research in Singapore (Photo: AI Singapore)

A lot of the work by the initiative is on fundamental research, talent and partnerships, rightly so.

There is also a lot of buzz about AI startups in Singapore. Every week, new incubators and accelerators are formed; eager founders give press interviews every other day; and even many IT outsourcing companies change their domain from ‘.com’ to ‘.ai’.

ViSenze’s US$14 million investments led by Rakuten was a further testimony of the capital’s recognition in the sector.

Only that the second tranche of the US$14 million was almost two years ago, and the amount does sound like a lot today.

Fertile ground?

In our daily debates with colleagues and friends in the sector, this topic often comes up: is Singapore the right place for AI startups?

AI Singapore’s approach (Photo: AI Singapore)

Interestingly, there seems to be two camps: one believing that because of all these government support, and the fact that good AI talent actually love to live in Singapore, the country will definitely become a global powerhouse for AI startups; the other camp is more pessimistic, believing that AI entrepreneurship is a game of scale, and Singapore does not have the scale to support that.

Which one is more correct?

Key success factors of AI startups

To answer this, we need to go through the logical process of identifying the key factors that are needed to make AI entrepreneurship successful, and then we can say whether Singapore has (or is able to build) these factors in a sustainable and competitive manner.

In our opinion, for an AI startup to succeed, here are the key factors:

  1. Access to talent
  2. Access to data
  3. Access to funding
  4. Access to business clients

And the following two factors are horizontal (i.e. applying to all four above):

  1. Scalability
  2. Competitiveness

Let’s go through these factors one by one:

Access to talent

The pundits above are actually right – Singapore is an attractive place for AI talent to live. I have actually not met anyone in the sector who would mind living in Singapore, from a personal point of view.

The challenge is, however, that good living is not the most important factor for top AI talent, opportunity to work with other top talent to build great things is. In that regard, Beijing (and to a certain extent Shenzhen) is far ahead.

Even in those cities, startups fight for talent, pushing the prices up. In Singapore, where the pool is much smaller, the fight is also quite evident. For example, a famous deep tech accelerator has been trying to poach top engineers from all the major AI companies in Singapore, while those companies also have difficulties finding new blood.

Access to data

Even if you have the most amazing algorithm, if you can’t train it with data there isn’t much value the algorithm can bring. Singapore is gifted in a way that many of the MNCs have their regional (if not global) headquarters here, hence many decisions on MNCs data can be made here.

That said, however, many of these companies’ data are not ready for serious crunching or feeding into an intelligent system. To further complicate things, many countries now start to see data as a strategic asset, and the ability to process data across border can be hampered if this trend continues.

Singapore’s own Personal Data Protection Act also makes it difficult for businesses to experiment with many datasets. Needless to say, the market of Singapore is pretty small, further limiting the amount of data operators can collect, assess, and derive intelligence from.  

Access to funding

Well, not a problem in Singapore. It was tricky a few years ago, but not an issue now, especially with lots of money coming from outside this region.

Access to business clients

Again, as mentioned above, access to decision makers in Singapore is easy. What kind of decisions they make, and whether organisations are ready to carry out these decisions, is another matter altogether.

Scalability

It could be a big challenge here, mainly because of talent and data. The talent pool is increasing rapidly but still limited. A big challenge is that the big guys are also tapping into this talent pool: Alibaba has recently set up a joint research facility with Singapore’s Nanyang Technological University (NTU).

Some top researchers from NTU (and other institutions) are already working for Alibaba.

The big guys offer more data to good researchers/engineers, something startups simply can’t match. And we are not even talking about the pay package here.

Competitiveness

Everything you do, you need to be aware of the competitive landscape. If the solution you develop is market agnostic, chances are it will be difficult for you to compete against big fish in the world, who have access to more data and more talent.

This we think really depends on your solution and sector.  Quite often, the entry barrier is not as high as many think.

How about algorithms?

It is important to note that while the fundamental research is absolutely essential for the sector’s growth, research itself does not turn into business cases (unless you are DeepMind). In many cases, great research leads to great development in the sector, not necessarily by the original researchers.

One point, though, is that great luminaries actually attract good talent around them; and if that is put into good use, great companies can be created. Think about Element AI or Baidu during the tenure of Andrew Ng.  

It is also worthing noting that algorithms are not as valuable as many would think. The latest major fundamental breakthroughs are pretty much captured in the public domain – with many developers leveraging that in their own model building.

Worse, any deep pocketed rival can easily hire a few good engineers, and recreate whatever algorithm you come out. Especially when the rival also has the data they are willing to let people play with, they can attract good talent to work with them.

What is the conclusion then

In the area of artificial intelligence, Singapore has many advantages to leverage, and it can benefit greatly from being a hub.

However, if you are to develop an AI startup in Singapore, you would really need to analyse carefully about the factors above, and choose your battle wisely.

The last thing you want to do is to embark on something, receive good government support and early stage funding, and then realise your business is going into a dead end.

Thanks for reading The Low Down (TLD), the blog by the team at Momentum Works. Got a different perspective or have a burning opinion to share? Let us know at [email protected].

 

Previous articleHalf of Your Favourite Video Apps are MADE IN CHINA
Next articleIs the market revving up? – Crypto Review Part 5
He has worn many hats in the past - selling advertising space, banking services, and even trading stocks. In 2013, longing for a change of scenery, he joined Rocket Internet’s (now Alibaba’s) Lazada as a online marketer in Bangkok, where he experienced first hand life in a startup. He never looked back since - landing lead roles at Rocket’s EasyTaxi (Singapore), Rocket’s MEIG (Dubai), and Bamilo (Tehran). After that, he launched (and ran) the Thai venture for one of Singapore’s biggest cross-border ecommerce. Last year, Chong put his expertise to work, helping an SGX-listed company relocate to and run operations in Thailand. Nowadays, he’s just chilling by the countryside.