 They should have run an open offer to all shareholders with a price calculated using the conversion price of the bonds and the recent block listing. Thus allowing all shareholders the option to purchase what they needed to avoid dilution on their stake whilst raising some working capital. Yes, I know I can buy in the market at lower prices to average my price down but that does not allow me to balance my position and keep my existing stake undiluted. It's a self cantered disrespect for shareholders who have supported the company to date. It's a fairly basic equity equation required to understand how much shareholders have been diluted since IPO. There were 208,780,487 shares issued at IPO. Do the math, after the bond conversion and when these latest options work through shareholders will have a tiny fraction of their stake remaining. We can understand why the share price can't get off the blocks even when it looks like prospects may be improving. It's going to take a monumental change in fortunes to shift the price significantly now. I hope that happens but I'm getting less confident as time passes. I'm not sure I believe AWE's recent narrative shift towards AI. The Microsoft CEO has just acknowledged there is no value being realised from AI investment. Bubble bursting? |
Surely, it is irrelevant whether offered to institutional investors or private investors. It does not remove the prospect of dilution. At least with a convertible bond offering there is no upfront dilution and if the bond holders don't convert there is the opportunity for the Company to settle the bonds without any dilution at all. In all probability the bondholders will eventually convert assuming the shares perform well. |
The recent debt round was also a kick in the balls to private investors. They should have run a parallel offering and given us the opportunity to avoid dilution. We could have taken it or left it but it should have been offered. |
Better than funding themselves via costly interest rates or investor dillution...You are not understanding the point here. Share options issued via a block listing are a direct transfer of shares from investors to insiders. It is dilution however you look at it. I get the concept that the options incentivise staff but the performance here does not enamour me that I am directly paying for these options that are not deserved. The company should be generating profits and buying shares in the market to fulfil options. Not excessively fleecing shareholders. |
Block listing is very likely regarding to employee stock option plans which is locked for couple of years and will be granted in parts upon certian achievements. $10 million SBC expense was in H1 this year. So it's like on track of $20 million looking at the ytd stock price. Assuming they will be issued part by part for at least more than 3 years is reasonable imo. And as the share price increases, new issues beyond the period certainly will fall proportional to employee count. Nothing changes in my view, they motivate the employees scheduling this now to let them know they are involved a great opportunity which is one of the most important things since all company has is ip and its talent. Better than funding themselves via costly interest rates or investor dillution... |
There is no doubt the founding directors have done extremely well out of the IPO especially given that they were able to get away with valuing the company at an overpriced £3.1 billion and take a significant chunk of the proceeds for themselves. What is clear is that they underestimated the cost of what would be needed to acquire and expand the business and this has caused them some cash flow challenges. Notwithstanding that, they appear to be on the right track now and should be on the verge of profitability by close of 2026. |
The last 63 million listing got eaten up by options in three years. So in the next three years I will have been diluted 7 percent. Makes no difference it's still dilution. Some may say it's excessive but we can withhold judgement until we understand if they are making progress. I will be happy of course if the share price is materially higher in 2028. |
No. They are not on the market. They are issued to employees, new and current, from time to time. At the beginning of every month you see an RNS that shows number of outstanding shares which increases slightly each time. The dilution rate has been about 3.5% since the IPO. The nearly 7% of new shares will filter in gradually over the next few years especially as the company continues to expand and increases its employee count. By my reckoning maybe the company will have just over a billion shares in issue by 2030 but this at a time when the company expects annual revenues upwards of 2 billion or so. |
Yes it does, there are 50 million more shares in issue then there was yesterday. Our equity share is now 7% lower than it was. |
The new shares are reserved and not floated immediately. It has no immediate effect on the market. |
I guess we have the average news scenario. Diluted nearly 7 percent. That should kill the rally. |
Hi Indiestu,My interpretation is that a significant portion of the short positions are hedged positions arising from the convertible bond offering and some are hedges against longs they have on the same stock. Prior to the offering if I recall the short positions were just over 3% of total share issue. A fair portion of these shorts were placed well below the current price so hopefully they will begin to be closed out soon. I think remember seeing Credo with a similar short exposure not too long ago and look where they are now. |
It's an interesting game of cat and mouse developing here. Shorts increased through January pushing the price down on the 27th. Since then the price has recovered whilst short interest has crept up to over 6%. I am bullish of course and believe AWE are making progress so how will it play out.
Any good news could initiate a breakout and the rally could force these shorts to close out, could be bedlam and a potential short squeeze ahead. Best scenario. Hope they are hedged.
Average news and the share price could come off the boil. Shorts will keep a floor on the share price as they buy back giving another potential trade entry at a lower price. OK scenario for a short term recovery trade.
Bad news and more short sellers will join the club devasting the share price for months. Very bad scenario. |
I think we are on the run back up to IPO price. |
Bots trying to keep a lid on the sp,keep failing! |
I really appreciate your insights thank you ! |
Kahro, good points raised here. I must say first of all that I am far from having any expert knowledge in this particular subject and relying on what I have read up on.
In terms of what DeepSeek have produced in their latest model using MoE architecture, I am fairly certain others will eventually catch-up with similar or even better models in terms of efficiency. Therefore, I would fully expect these models to be able run very well using less GPUs and less of the high-end type of GPUs. Users running these models will want cost effective and efficient GPUs.
But the point is that such an AI model is now more accessible to businesses with lower budgets. Therefore, more customers are likely to want to adopt AI solutions. This in turn could increase the demand for the supporting AI infrastructure. Some of these businesses may want to have their own datacentres rather than relying entirely on hyperscaler datacentres so this may open up a new market area and it is one that Alphawave should be able to play well in with their highly efficient connectivity solutions.
Other architecture type models such as transformers, and convolutional neural networks are not going to go away and will still be required as they will perform better at specific tasks that MoE based architectures cannot achieve. So there should still be an abundance of AI applications that need a large number of GPUs.
The rate at which AI infrastructure is growing should at some point slow down and tail off as you have suggested but I suspect that in the short term the growth rates will remain at least as already predicted by analysts or due to the possibility of the arrival of a new market of intermediate type datacentres such growth rates could actually increase in the next 5 to 15 year period.
In terms of valuation, if I compare the EV/EBITDA multiples on a selection of competing or at least similar NASDAQ players (Marvell, Broadcom, Synopsys, Cadence, Rambus, Arm, Astera Labs and Credo) based on latest 2024 figures and current prices, I get a median value of 47 compared with 28.7 for Alphawave (based on $50 million EBITDA). On the forward 2025 figures the median value is 41.8 compared with 14 for Alphawave (assuming the 2025 guidance EBITDA of $100 million). So by that comparison Alphawave does look pretty cheap although one should take into account that the American market, particularly for AI related stock, is probably somewhat overheated at the moment. |
I really want to know if hardware can be used so efficiently that, if hyperscalers in the West figure out how to do so, it will lead to less CapEx and infrastructure needs than analysts previously projected. Or will they continue to crave the highest performance possible, maintaining the projected demand despite efficiency gains?To put it simply: will they choose to have 1 GPU that delivers the same performance as 5 GPUs, or will they still prefer having 5 GPUs because they offer even better performance using Deep Seek's method?Technological development will undoubtedly be unlimited, but there must be a limit to infrastructure growth-say, 10 years from now. Beyond that, revenue for hardware companies will mostly come from replacing existing GPUs with more advanced ones, with only modest growth, perhaps around 3%. If this replacement cycle results in fewer GPUs overall, the intrinsic value of these companies should be questioned.I believe Alphawave is undervalued and has potential beyond the data center business. However, this scenario might make it less undervalued than I currently think. I'm really not sure. |
DeepSeek has cited a cost of less than $6 million to train its model, but bear in mind that this cost only relates to the pre-training run. It does not include the costs of R&D, nor does it include the cost of owning the infrastructure on which the model was trained. High-Flyer is the hedge fund that has funded this start-up and they have invested more than half a billion dollars in the last 20 months that DeepSeek has been in existence.
However, $6million in pre-training is very competitive when you consider that other competing models have typically spent multiple $10s of millions. The DeepSeek V3 model requires less than 70% of parameters compared with GPT 4o and runs almost as fast. The model is claims to be more versatile and has improved reasoning capabilities compared with its previous version.
Also, with the model being open-source this allows developers to download, modify and integrate the model into their applications which encourages innovation and collaboration within the AI community. Take note however that the benchmark claims made by DeepSeek only references the ones where it is showing some leading gains over its competitors and that there are in fact plenty of other benchmarks where it is not as good.
The key achievement in performance efficiency (i.e. requiring less parameters) by DeepSeek is in the use of what is called Mixture-of-Experts (MoE) architecture, MoE architecture is essentially one large model comprised of many other smaller experts that specialize in different things, an emergent behaviour. This is an architecture that was first developed in the 1990s long before AI was even a thing and which the likes of Google, OpenAI and Microsoft have all been experimenting with for their AI models for a while. DeepSeek has managed to take a lead in that development by establishing a gating network algorithm that efficiently routes the tokens to the right expert in a balanced way. It is this breakthrough that led to fears of this meaning there would be less investment needed in AI infrastructure.
However, there is a bull case to say that in fact what this really means is that the opposite will happen and that investment in AI infrastructure will start to increase significantly because with the arrival of more cost efficient open-source AI models like DeepSeek V3, more businesses can now afford to invest in running their own models and to meet such rising demand will require more datacentres to be built. The phenomenon where a more efficient technological breakthrough increases the demand and therefore the resources required for the new technology is called Jevons Paradox.
This breakthrough will probably be a positive for Alphawave Semi because they are in the business of developing and supplying some of the fastest and most efficient connectivity hardware and I would imagine that if more datacentres are having to be built to meet the increasing demands then this expands the market into which Alphawave sell into. Perhaps not all datacentres will need the state-of-the-art highly efficient and high speed connectivity that Alphawave offer but many will and some will look to the Arm based System-on-Chip architecture based on chiplets, and some will want to use the more cost effective accelerator chips than what Nvidia offer such as the Rebel chip by Rebellions, both of which are supported by the connectivity built by Alphawave. |
Kahro, I am currently inclined to believe the DeepSeek revelation is likely to ultimately be a positive for AWE in the long run. I will try to provide you with a more detailed analysis with the pros and cons this weekend when I get more time. |
Can you share your thoughts on the deepseek matter ? |
https://www.forbes.com/sites/petercohan/2025/01/26/nvidia-stock-may-fall-as-deepseeks-amazing-ai-model-disrupts-openai/An interesting read ref Deepseek. It seems the US has shot itself in the foot with its protectionist policies. I guess the AI bubble was waiting for the pin to pop it. |
NVDA down 10% in pre-market. |
Valhamos - Thanks. |
Register for Free to get streaming real-time quotes, interactive charts, live options flow, and more.
Hot Features
Premium
