Shares in Value Logo
Product Review Img Vertical

Date : 01/09/2021

Brainchip Holdings

ASX :

BRN

Market Cap : $785.18 Million

Buy

52 Week Range : $0.28 - $0.97

Share Price : $0.49

A company with a groundbreaking product operating in a high-growth industry. We recommend a "Buy"

Company Analysis

Let’s just take a moment to think about a computer chip that can learn everything by itself in a way, just like how a human brain does. The chip processes sound, vision, and touch, not just through the usual binary data such as zero and one, but through neuromorphic processing of data – known as spikes, would that be a tech revolution? Well, that is the technology that BrainChip (ASX: BRN) is developing right now. BrainChip is pushing the boundaries of Artificial Intelligence (AI) and its key product, Akida, a Neuromorphic System on a Chip or in short NSoC – has turned a few heads in various emerging industries, including the Internet of Things (IoT), self-driving car, and data centres to name some of the sectors BrainChip’s tech can be applied to.

Brainchip has been a part of our portfolio since September last year, when we first recommended a Buy at a then share price of $0.32 a share. Today, the stock has returned 52% to members.

Graphical user interface, chart Description automatically generated
Source: Tradingview.com

Since 2011, the focus of the company is the development of software and hardware-accelerated solutions for advanced artificial intelligence and machine learning applications. However, one product has emerged as the cornerstone for BrainChip, it is its flagship technology Akida Neuromorphic Processor. Akida emulates a similar concept to the human’s neural network to provide a complete ultra-low power and fast AI Edge Network for vision, audio, and smart transducer applications. We believe that Akida has the potential to bring massive earnings growth in the very near term as the firm has just got its first batch of its Akida AI processors from its manufacturing run from SocioNext America (SNA) not so long ago, in May.

BrainChip provides key differentiation that could propel the company to be the leading Neuromorphic Chip producer: Earnings growth through three distinct revenue streams

BrainChip has been recently named among the EE Times’ Silicon 100. That is a big milestone for the company as the tech community has recognised the high potential of the AI chip producer as one of the emerging key players to watch in 2021 and onwards. The EE Times’ twenty-first revision of the Silicon 100 tracks the pulse of the industry to identify emerging technology trends and developments that hold promise for the future. This year, the publication chose to analyse the Silicon 100 in more detail with twenty-two categories that run from materials and packaging to quantum computing and security.

BrainChip has been recognised in the “Specialist GPU-through-AI processor and accelerator” category. It is a great piece of news for BrainChip, hence, the selection of the companies to the Silicon 100 is based on stringent criteria such as cutting-edge technology, market perspective, financial position, investment profile, and the quality of the executive leadership. BrainChip inclusion was based on its flagship product Akida which the community believe to be promising for the development of AI in different fields of applications.

So, what exactly consists of the Akida technology and how will it drive substantial earnings growth? BrainChip’s credit goes into developing a solid-state architecture that resembles the neuron in the human brain. It has assembled 1.2 million of these neurons accompanied by synapses on a computer chip named Akida NSoC. The technology processes events or spikes instead of the conventional data which makes it different from its competitions. Hence, this concept has some big advantages.

The technology, known as the Spiking Neural Network (SNN) is much faster and consumes only a fraction of power while processing events compared to traditional technologies processing data. The company expects to commercialize the technology at a considerably lower price point. Currently, BrainChip is involved in multiple pilot projects involving Akida and its software-based SNN which is known as the “BrainChip Studio”. Additionally, the company has many design wins and more than five hundred commercial prospects for its solutions. What we like about the Akida product is that it has the potential to generate three distinct revenue streams.

  1. Intellectual Property license: BrainChip is expected to selectively license its IP associated with Akida to other players in the semiconductor industry. These licensing revenues will be non-recurring in nature.
  2. Royalties: Once the licensing is completed and customer’s transition into commercial production, royalties will be paid to the company by chip manufacturers for each chip they sell which includes Akida IP. These royalties are a percentage of the selling prices and will be recurring in nature.
  3. Direct Sales: The company also intends to sell directly to end customers in select locations.

What we like about BrainChip’s Akida cutting edge tech is its differentiation as opposed to most chips used in AI applications today. Generally, in the industry, central processing units need frequent communications with the data centre. On the other hand, Akida functions in a similar fashion as the human brain, hence, operates as an event-based processor. This way, it consumes very little electrical power. As such, we have recognised that Akida’s largest market opportunity could be the Edge Computing sector. In our view, most Edge devices rely heavily on sensors that capture and process real-world data with limited power sources. Edge devices are pieces of equipment that provide an entry point into enterprise or service provider core networks, for instance, routers, routing switches, integrated access devices etc.

Diagram Description automatically generated

Most of the chips used in Edge applications today are power-hungry general-purpose chips, such as Graphics Processing Units (GPUs) that consume approximately 1,000 times more power than the Akida chip. Or they are low-power Digital Signal Processors (DSPs) that use a similar amount of power to Akida, but with about 1/60th of the performance of Akida. And these DSPs have very high latency, so they can only perform relatively simple tasks. Akida’s low power consumption, reduced need for internet bandwidth and improved latency provides faster response times within a system, making Akida an ideal solution for Edge devices and for the use of the Internet of Things, such as sensors in autonomous vehicles.

Through its four key differentiation: (1) Low power consumption, (2) Small and lightweight, (3) On-chip fast learning and interference, and (4) Low heat generation, we have identified three key technological problems faced today that Akida could solve, and therefore, propel the company to become the leading AI-Chip manufacturer.

  1. Internet bandwidth: According to analyst estimates, the number of IoT devices is expected to witness a three-fold rise from now to the end of FY25. Therefore, we could experience insufficient Internet bandwidth due to serious inefficiencies. With 30.9 billion IoT devices and 10.3 billion computers connected to the Internet over the next four years, there will be an increase in demands for Internet bandwidth leading to delays and capacity constraints for IoT devices. Here comes Akida, which could help to reduce internet bandwidth constraints as data stays on the device. Moreover, certain applications that Akida facilitates, such as remote sensing do not need an ongoing internet connection at all. Akida could work independently of the cloud to perform analytics on or near the device from the point of data acquisition.
  2. Power consumption: We all know that the proliferation of data centres is a leading concern for the global growth in energy consumption, which is one of the major contributors to climate change. Furthermore, energy consumption by data centres is on the rise. It is projected that in the next ten years, data centres will account for circa 20% of all electricity consumption globally compared to only 3% in 2016. This significant increase in energy use during the next ten years will lead to a jump in carbon emissions. As of today, Amazon with its cloud service AWS is one of the global leading grid computing providers. It is worth noting that one instance on AWS can consume 20Wh. Comparatively, Akida is much more efficient with its AKD1000 Chip which has an ultra-low power consumption, making embedded AI applications at the edge possible, in the range of 100 microwatts to 300 milliwatts. This is a great technological advancement, hence, Akida is 97% more energy efficient to compute complex algorithms as opposed to processing the same task in a data centre. What we like about BrainChip’s Akida solution is that it surfs on the green wave. Akida could be a game-changing technology that can be a key driver to reduce carbon emissions due to growth in energy consumption. Akida should be capable of diminishing energy demand as the technology enables edge computing, thereby drawing computing requirements away from data centres.
  3. Computing power: Whilst edge devices largely depend on data centres to operate, Akida makes it possible to decentralise it. Furthermore, there is a lesser need for data transmissions, which improves data security. BrainChip’s Akida can perform on-chip convolution, which means that the system can run complex deep learning on a neural network completely on-chip. As such, the Akida chip exhibits a game-changing technology for the “Edge AI” systems. After spending about ten years on the development of this technology, BrainChip is now on the verge of mass commercialising this chip. The Spiking Neural Network technology can learn autonomously, evolve, and relate information just like the biological brain. There is a wide range of possible applications of this technology. Akida is well-suited for remote environments, such as space and agricultural areas, where Internet access may be limited or unavailable.

In December last year, NASA placed an order under BrainChip’s “Early Access Program” or EAP to test Akida’s capacity for potential use in space flights. Especially, the EAP is only available to a select group of customers that required early access to the Akida chip, evaluation boards and dedicated support.

Source: BrainChip

FY21, FY22 onward outlook: First production chips received: commercialisation expected in October

BrainChip collaborated with SocioNext America (SNA), which is a global developer of advanced Application Specific Integrated Circuits solutions for the production, development and manufacturing of the Akida chip. SNA and BrainChip have secured commercial production at Taiwan Semiconductor Manufacturing Company in Taiwan with the first batch of production chips received during this month of August. These chips are being tested at the moment and once the quality control process is completed, Akida should be commercially available early in October. Although BrainChip has just received its first chips production, we would like to emphasise the other four early wins as well in the run-up to the commercialisation phase that has now started:

  1. First license agreement: BrainChip announced the first license agreement with a major ASIC manufacturer in December last year. This license was paid at the beginning of this year and royalties are projected for FY23.
  2. Fifteen EAP customers: About fifteen “Early Access Programs” have been received and paid for Akida engineering samples during FY20. These comprise major car manufacturers and NASA. We expect some of these clients to convert to IP licenses and furthermore chip sales.
  3. Proof-of-concept agreement: BrainChip has also signed a proof-of-concept agreement with a supplier of electronic components to the automotive industry. This proof-of-concept is a great opportunity for BrainChip as it will open the door to the possible deployment of the Akida chip in LiDAR 3D object recognition, highly in demand for the development of self-driving cars.
  4. Technology to identify COVID-19: BrainChip has signed for a research project with Biotome to work on the possible use of Akida chip in the identification process of COVID-19 antibodies in blood samples. Biotome has paid for this research project to develop a commercial medical device for antibody classification for a range of diseases.

BrainChip closes the first half of FY21 with revenue growth

Despite having not yet started its flagship product commercialisation, BrainChip reported half-year revenues of US$ 767,545. That is a serious increase compared to the same period last year, which was effectively pre-revenue booking of a tad more than US$ 13 thousand in what was described as a “development service revenue”. On a comparative basis, that is a stunning gain of 5,629% year-on-year. Due largely to research & development costs, the company booked an operating loss of more than US$ 8.4 million. BrainChip finished the half-year with around US$ 17.6 million in cash which is quite comfortable.

Of the US$767,545 total revenues, Brainchip ascribed US$ 412,500 of it to license revenue, with US$ 303,040 from development service revenue and US$ 52,000 to product revenue.

During the last five years, BrainChip expenses were mainly the results of intensive research and development costs. However, since FY20, the company has significantly improved its cash/expense ratio, and we have seen expenses limited to slightly above US$ 11 million while having a cash balance of US$ 19 million. As of H1FY21, BrainChip currently has a comfortable level of cash which is more than sufficient in our opinion to support the ongoing operations for the next two-year period. We have also seen good management in terms of working capital, which is expected to grow by a CAGR of 4.81% from FY21 to FY23.

Chart, line chart Description automatically generated

Revenue forecast

BrainChip is in its final stages of commercialising its breakthrough Akida chip. The architecture and behaviour of this disruptive, AI-based computer chip are similar to that of a biological neuron. This neuromorphic processor is expected to unlock the full potential of edge computing through its unrivalled four key differentiators which we have mentioned in this report. The number of edge-based devices is expected to increase substantially, hence, connections are projected to rise significantly by a CAGR of 13% for the next four-year period. As BrainChip has progressed rapidly beyond the development phase and is moving closer to its commercialisation expected for October this year, we expect earnings growth to catch up rather swiftly for FY22.

Recommendation

As BrainChip moves closer towards the commercialisation of its Akida chip, the company appears well-poised to target high growth end markets, such as industrial, consumer, mobile and automotive, with multi-billion-dollar business opportunities. The first revenue opportunity has already opened as BrainChip signed an IP licence agreement during December last year. If this agreement turns out to be successful, we believe BrainChip can potentially attract dozens of IP licensing customers going forward. Given an asset-light licensing model with recurring royalty revenues, we expect high gross margins once Akida sales ramp up. The first batch of Akida chips has just started this month through its wafer fabrication partner, Taiwan Semiconductor Manufacturing Company. We expect BrainChip to convert current prospects via its “Early Access Program” pilot projects. With all these positive catalysts ahead, we definitely think that BrainChip is a rock-solid FY22 earnings winner, therefore a “Buy” recommendation from us.

 

Scroll to Top

Login

By submitting this form, I agree to the TERMS AND CONDITIONS and PRIVATE POLICY