Tuesday 30 May 2023
By /
main news image

In March, Aemulus Holdings Bhd was at its lowest ebb. The company, which specialises in semiconductor testing equipment, had a problem collecting payments from its customers in China and it was running out of money.

“Back in November 2019, the first thing I told my chief financial officer was to go and borrow as much as we could while we still had enough credibility to get financing,” founder and CEO Ng Sang Beng tells Enterprise.

But when he named a figure (RM20 million), his CFO shook his head. Things had already deteriorated badly and it would be a challenge to raise this amount. The CFO put up an Excel sheet on the Google shared drive to let management know just where the company stood in terms of cash flow and it did not make for pleasant reading.

Ng had to make a decision quickly. Either the company closed its loss-making operations in China or he had to go to the ground to see what the problem was and why customers were not paying.

First, he looked at the data. He realised that over the past 12 months, the company had spent RM1 million to RM2 million on things like sales and marketing, travelling and application engineering expenses in China alone. That was the equivalent of what it spent on the same things for the rest of the world.

“But the revenue generated by the rest of the world was about 8 to 10 times more than what we made in China. So at one point, we thought maybe this was how Chinese companies work and maybe we should just close that office. I mean, if I cut out this expense, I would have been profitable on the spot,” says Ng.

But he was not ready to give up yet. It was just not in his personality. In his first job at Altera, when he was assigned the mind-numbing task of comparing graphs visually on the computer and told that he could not use software to do the job because the two graphs had different bases, he promptly found a way to do so. And when he left Altera to set up his own company and landed a project to build a semiconductor tester (something he knew nothing about), he managed to not only build that tester but also built one that was the cheapest and fastest in the world.

Before it was listed, Aemulus was viewed as one of the most promising third-generation companies in the electrical and electronics (E&E) sector in Penang. That was why Khazanah Nasional Bhd decided to take a stake in the company. This year, just as Aemulus was going to turn things around after years of non-performance, the sovereign wealth fund divested its shareholding.


Taking the bull by the horns

Ng is something of a whizz-kid when it comes to solving problems, and he was not about to let this setback floor him.

Before this, he depended on his team in China — a general manager from Taiwan and two sales people — to do the work, pushing them to perform from Malaysia. But they were just not delivering the goods.

“I realised that I needed to put my boots on the ground, look at things and talk to customers over there. If I went there just as a visitor, I would never be able to understand what was going on,” Ng says.

First, he appointed a sales agent to visit Aemulus’ customers in China. “The feedback was just terrible,” he says.

Ng found out that when customers had a problem, his team was unable to deal with it. “We had managers who were taking a very high salary but who would not appear at a customer’s site when they were called. Instead, they would try to solve the problem by logging in remotely. It took us hours to convince them to go to the customer’s site,” he says.

Ng took away the perks given to these non-performing staff and they eventually left. He moved his general manager back to Taiwan, where he started to perform again, and hired a whole new team for its operation in China.

Because of the poor after-sales service, Aemulus’ reputation was badly affected in China. Ng had to visit every customer to personally apologise. “It was purely to pacify them. This would not bring us any revenue, but I needed to do it.”

He remained in China and Taiwan for three months — from November last year to January, at the height of the Covid-19 pandemic and with the US-China trade war in full swing — visiting customers and identifying opportunities.


A problem of perception

One solution would have been to form a joint venture with a local company, which would then take care of the Chinese market. In fact, a joint venture had been on the cards with Tangren Microtelligence Co Ltd (linked to the largest foundry in China, Semiconductor Manufacturing International Corp), but the plans fell through. Ng decided that one of his top priorities would be to revive it.

The proposed joint venture was particularly important because it would have meant that Aemulus’ entire business in China would be handled by a local entity, which would lower barriers to entry and the cost of doing business in the country. The company would also have become part of the domestic ecosystem and be aligned with China’s plans to localise 70% of semiconductor imports by 2025.

“At the centre of the trade war is Huawei. The centre of Huawei is 5G, and the centre of 5G is RF (radio frequency technology),” say Ng.

“The reason our partners looked for us was because a very influential person in the semiconductor world told them that we were the only non-US, non-European, non-Japanese company doing RF. And we have been doing RF for 13 years.”

However, this very important joint venture was called off. “Things were very bad and I needed to find a turning point. At the time, something my wife told me really helped. She was watching a China series about an emperor and empress, and the empress was a very smart woman who said, ‘We are dealing with so many issues. The thing to do is solve one issue at a time’. It is actually very simple and that was the solution,” he says.

“So I asked myself, ‘What exactly is my problem now?’ There were so many at the time — employee morale, customer confidence, customers not buying from us — which did I want to solve first? I decided it was the joint venture because by being a local company, we would have local personnel in China dealing with customers directly and who could probably provide better support and produce better results.”

What was the main problem impeding the joint venture? “Someone had bad-mouthed us to our partners, so it was a perception issue that I had to solve.”

First, Ng pointed out to his counterpart at Tangren that Aemulus had been dealing with US companies that were the top two or three suppliers in the world. Surely, it could not be that bad? “The partner agreed but asked if there was anyone he could talk to. I called one of our customers in China, a local who worked for a US company, and arranged a dinner. I would not be present.”

The dinner was a huge success. The customer told the potential partner that Aemulus’ product would be given a score of 100 while most of the newcomers in China in the same product category would only manage to score about 50 to 60. “He told my partner that if he thought Aemulus could test the US company’s product and not theirs, there must be something wrong. His words carried weight because he worked for such a reputable company,” says Ng.

Then, the guy from Tangren asked Ng to have lunch with a friend of his who worked for a very large RF integrated circuit (IC) company in China. “He wanted to use his friend to gauge our product. So, I talked to his friend and by the end of the lunch, this guy reported back to my partner that our product was good; we just didn’t know how to sell it.”

This was the turning point. “Our JV partners knew that first of all, the JV must happen because we [Aemulus] have to be here. Second, our product was good and we were just not doing well in selling it. He was so happy that he said we would push forward. So, that problem was solved,” says Ng.

The new JV company, Tang Ming Sheng Shi (Jiashan) Technology Co Ltd (TMSS), was incorporated a few weeks ago. “It is located in Jiashan, a two-hour drive from Shanghai,” he says.

“Our facility there will produce our AMB 7600-SR 5G testers and AMB 7300 RF filter testers. We believe these products will be the major revenue contributor to TMSS. We believe that TMSS will be able to serve the China-based RF and 5G-related design houses and OSATs (offshore assembly and test) companies in the country that are aligned with the Made in China 2025 initiative.”

Ng is very excited about the prospects of TMSS. “The market potential is from 100 to 300 testers a year at US$70,000 each, which works out to about US$7 million to US$21 million,” he says.

The second problem Ng had to tackle was the unhappiness of his customers in that market. Having apologised to his customers and gotten rid of then his non-performing employees and hired new staff, he appointed a new sales agent who started to generate sales almost immediately.  “During the Movement Control Order period, we were talking to him over the phone and through WeChat, and we closed US$1.3 million in booking something we had just launched this year, the AMB 7300 RF filter tester,” he says.

“Every phone has multiple RF filters. Filters are crucial to making sure that our phones work in different countries and in different bands of frequencies. That is the job of the filters. And it is a huge market.”

Ng says the two key players in this market were Murata and Broadcom. “But the Chinese want to replace 70% of the US$300 billion import of semiconductors by 2025. There are now easily 100 companies in China designing filters to replace these two companies in the country.”

This is part of China’s retaliation against the US in the trade war. “And when they get rid of the Americans on the IC itself, they will also want to get rid of the US test instruments — we strongly believe so. And who else can provide these? We are probably the best candidate,” he says.

Ng saw the opportunity and seized it. “We got the purchase order for US$1.3 million in bookings during the MCO period, after no sales for the longest time. And we shipped one of the testers in June. The order was for 20 RF filter testers.”


Not a company for generic products

The company’s sales had flagged for the longest time because it had designed a generic tester, the AMB 5600, which had been all but impossible to move. “Last year, when we were crunching the data, we also looked at what had happened with this product. All the money we put into R&D did not materialise in revenue,” says Ng.

Why? “For so many years, the company experienced a compound annual growth rate of 70%, but our approach was different. We always went to the customers and talked to them about their pain points. Then, we customised a product that would become the standard for their product. That was how we did it in the first 11 years before we were listed.”

All these customised products have a life cycle. “For example, we invested RM500,000 to design the AMB 4600, which was one of our products for Broadcom. But we generated RM40 million to RM50 million in revenue from that product. Will that revenue continue? No. Then, we would design another product based on the same principles and start another life cycle. That’s what we had been doing all this while,” says Ng.

But he decided that this was the wrong approach and if the company designed a generic product, it could see even more growth. It was not to be. “We spent RM2 million to RM3 million on R&D for the AMB 5600 and were not able to sell even one.”

There was nothing wrong with the product itself. “It was a very high pin count, high digital concentration tester, which means it could test for a lot of channels. It was a pure digital tester and you could add an RF link to it. We were supposed to target something called baseband chips,” says Ng.

But it didn’t sell. “The problem with generic products is not the product itself but the documentation, support, day-to-day answering of questions and inquiries of what customers want. Because it is generic, you can serve different types of customers, which sounds good. But instead of spending 2.5 engineers per customer, we were left with 0.5 engineers per customer. That explains why the service was terrible and customers were very unhappy with us.”

Ng gives the analogy of a high-performance car that is sold along the same lines. “Both the sales office and the service centre are in Penang. So, if you wanted to buy this car, you would have to pick up the car in Penang. And every time you wanted to service it, you would have to go there. Would you buy it? No, because for this type of generic product, you would also need to have sales offices in Kuala Lumpur and Johor. And you would need to open service centres in all these areas.

“But we didn’t have the capability to support such a generic product. It doesn’t work for our business model.”

The effects of introducing a product that did not sell were far-reaching. “Our share price tanked all the way to eight sen in March. Khazanah (which had a 15% stake in the company) started selling in December 2019, our people were leaving and nobody knew what was going on. We never explained to the market what actually went wrong. And that whole thing, coupled with low cash flow, made everything so complicated and difficult,” he says.

Morale was at an all-time low. “Many of the managers were not prepared to face such terrible malaise. I myself had a hard time thinking about how to solve this problem. But the strategy of looking at one thing at a time was very important.”

Ng brainstormed with his vice-president of sales and marketing and chief technology officer (CTO) to come up with solutions. They came up with several strategies and tested them. If they failed, they tried something else.

They decided to modify the non-performing AMB 5600 tester into a specialised tester to serve the CMOS (complementary metal oxide semiconductor) image sensor market. This tester is three times faster than conventional image sensor testers in the market because Aemulus engineers design based on FPGA architecture while its competitors use PC architecture.

“When you do coding based on PC architecture, most of the time, it is serialised, which means you execute one thing at a time. Whereas when you use FPGA design, a lot of things can be done in parallel,” Ng explains.

That is why the modified AMB 5600 tester can test the functionality of image sensors three times faster than what is available on the market right now. The company received immediate orders for this product, which was launched during the MCO period in both Taiwan (US$450,000) and China (US$270,000).


Not sweating the small stuff

One thing Ng was adamant about was that senior management should be concerned about strategy rather than wasting their time looking at every minor detail. The middle managers would have to step up.

Most of them had not been trained to do the work that was now expected of them, so Ng decided that what they needed was coaching. Then, the team moved into the company’s spanking new headquarters. Morale improved immediately.

Everyone was given more responsibility. “I told my very senior management that each of them had to manage more people. I told my CFO to take on procurement and then manage the operations. I know there will be no issues because we now have all the right managers in place,” he says.

“Then, I put my sales and marketing VP to manage the front-end — that is, customer programme management, customer support, engineering support, application engineering and sales and marketing — to have an efficient joint action force.”

Ng told his CTO and software architect to simply produce the technology. “I said, ‘You need to give me wonders. Don’t ask me what I want. Tell me what you are going to offer me.’”

He says the technologists came up with very good ideas, which would be very appealing to TMSS. “China is not able to source RF equipment above a certain criteria because the producers are being prohibited from selling to the country. So, we are going to design those parts in China through the JV company. We have just started. We expect the first product to roll out in 12 months.”

In Taiwan, he asked the general manager there (who had been transferred from China) to concentrate on the CMOS image sensors. “I said, ‘Forget about RF. The CMOS image sensor is our best bet’.”


Shareholding changes

As Ng mentioned earlier, Khazanah — which had been the second largest shareholder in the company since its IPO — has divested its stake. It bought into the company at 28 sen per share, but after factoring in the bonus issues along the way, this worked out to a total investment of 23 sen per share.

“There were basically three periods of selling. One of them was in December 2019 to February.  And then, with the Covid-19 outbreak and MCO, the share price tanked from 20 sen to 8 sen in March. In April, Khazanah started selling again. We were so disappointed with them for divesting, especially during this period when the government was trying to help every company,” he says.

“I actually had my chief operating officer and CFO phone and ask them why they were doing this, trying to cash out at such a time. So, they stopped for a while, but started selling again in July. The last block of shares they sold would have been at between 30 sen and 40 sen, the day we announced the turnaround story and CIMB produced its analyst report on us.

“Our transactions for the day shot up to 160 million shares, which was a record high, and the share price went from 30 sen to 40 sen. CIMB had set a fair value of 39 sen based on the expected 2021 profit of about RM7 million, but the market beat the target price within a day.”

A few weeks ago, the company did a private placement of 54,940,000 shares at 57 sen each — a discount of six sen or 9.52% to the five-day volume weighted average market price (VWAMP) of 63 sen as at Aug 12 — raising RM31 million. “We did a private placement for a couple of reasons. First, we are looking at buying a stake in a test handler company to get a say in telling them our requirements for the CMOS image sensor testing, but there is nothing concrete yet,” says Ng.

Testers need to work with specific types of test handlers that transport the chips and move them into position so they can work. Ng points out that while its competitors have both testers and handlers and are able to optimise how the two work together, Aemulus has to deal with third-party handlers who refuse to optimise their handlers to its testers.

“These third-party companies want to sell standard, off-the-shelf handlers. So, the performance advantage of our CMOS testers are cut down from 3x to 1.5x because handling is slow. We are thinking of taking a minority stake in one of those handler companies so we get a say in telling them our requirements, so we can get the full 3x,” he says.


An IP powerhouse

The company will also be using the private placement funds for R&D. “Aemulus is slowly moving from a typical equipment seller to becoming an intellectual property (IP) powerhouse. That is because we realise our IPs have become very valuable in China. We are going to use the money to produce more IPs, be it trade secrets or patents. And we will use the sale of our IPs to build up our war chest,” says Ng.

Aemulus is in the process of selling its first IP for RM3.4 million. “That is pure profit — it looks like other income, but with 100% margin. And we have done a back-of-the-envelope valuation for the rest of the IPs we own and reckon that they are worth, conservatively, tens of millions of ringgit. We are working on new IP, including for 5G, RF and AI-driven image processing,” he says.

Ng says the IPs are what makes the company especially attractive to Tangren. “The objective of the joint venture is to continue absorbing IPs from Aemulus and localise these for the Chinese market. At our future board meetings, we will discuss considering IP sales as one of our sources of revenue.”

The company registered a profit of RM87,000 in the last quarter and part of the sale of its IP will be registered in this quarter. “Of course, we have some expenses involved, but the profit will still be sizeable,” he says.

He adds that its main markets are in Asia — China, Taiwan, South Korea, Thailand, the Philippines, Singapore and Malaysia — as well as the US. And it has diversified its customer base. “Broadcom used to be our largest customer, but that is no longer the case. Now, we have a more diversified pool of customers such as Skyworks, Carsem, Infineon, Unisem and Tesna as well as some China-based clients.”

      Text Size