ICM HPQC News Flash - August 2025
- ICM
- Aug 4
- 6 min read
Updated: Aug 28
ICM HPQC Fund (“ICM HPQC”) continues to see excitement in the next-generation computing space, with Big Tech pouring massive capital into AI data centre infrastructure. Meta, xAI, OpenAI and others are driving demand for supporting technologies like cooling, chips, and energy efficiency solutions.
In this newsflash, we cover:
Latest news from Meta, xAI and OpenAI on 5GW data centres
Matt’s view on data centre cooling
Portfolio updates
A quick link round-up of what we’ve been reading this month
5GW AI superclusters by 2030
In contradiction to reports of a data centre slowdown earlier this year, Meta and xAI have stolen the show with major announcements in July that they will each bring online 5GW AI superclusters by 2030. OpenAI has also announced an ambitious investment in its own data centres.
Zuckerberg expects to spend “hundreds of billions of dollars” on data centres, is paying around US$200 million for AI researchers to join Meta’s AI lab with US$1 billion dollar offers being made to the leadership teams of OpenAI and Thinking Machines (not a single billion dollar offer was accepted - a sign that this author picked the wrong career). Meta previously stated that it would spend $60-65B on data centres in 2025, up from $35-40B in 2024. Whilst the investment amount is impressive, the size of the data centres themselves is equally so - these are multi-GW superclusters that cover most of the footprint of Manhattan. By the end of 2027, Meta could have as much as 1.5GW of IT power (enough capacity to power 1 million homes). Many are calling Zuck’s strategy “dominance at all costs”. Hyperion, Meta’s 5GW data centre, is expected to start coming online in 2030. Read more.
Musk is targeting 50 million “H100 equivalent” AI GPUs in the next 5 years, which would provide around 50 exaFLOPS for AI training. xAI’s Colossus 1 supercluster uses 200,000 Nvidia GPUs based on the Hopper architecture, as well as 30,000 GB200 units based on the Blackwell architecture - 50 million is a massive step up. The estimated power consumption of the 50 ExaFLOPS cluster is 4.7 GW, assuming that the energy efficiency of AI architecture doubles by that time. This seems unrealistic but Musk has a reputation for delivering on ambitious projects. Read more.
Oracle and OpenAI have also entered into an agreement to develop 4.5 GW of additional Stargate data centre capacity in the US, which will run over 2 million chips. Read more.
Between 2025 and 2027, OpenAI has projected a ~US$35B spend on servers to power just existing products and another ~US$55B on servers to develop its technology further. Read more.
If you’re starting to feel second-hand stress for the energy grid, you’re not alone - a 5GW facility is equivalent to the energy consumption of nearly three million homes and AI data centres consume eight times the energy of non-AI data centres. In 2022, data centres accounted for about 2.5% of US electric demand but by 2030, that figure could rise to 20%, with AI data centres accounting for three-quarters of the demand. Read more.
Meanwhile, Europe made a big splash about Open AI’s “Stargate Norway”, a 100,000 NVIDIA GPU with 230MW of capacity. It’s OpenAI’s first data centre initiative on the continent and important for Europe’s sovereign AI infrastructure, but it feels rather pathetic in comparison to the investments OpenAI, Meta, xAI and others are making in the USA. Read more.
The race is well and truly on, and, hidden below the noise of the increasingly intense competition for AI compute are the real winners - those who can get a slice of Big Tech’s seemingly infinite CapEx budget - like those companies in the ICM HPQC portfolio that can deliver the tools to increase performance and energy efficiency of data centres while lowering costs.
Data Centre Cooling
Cooling is a critical part of any data centre and is becoming increasingly intensive as GPUs generate more and more heat, with heat generation being the key bottleneck to increasing the number of GPUs, and therefore performance, in a server. Alongside chip-design and photonics, there has been significant early-stage venture cash flooding into cooling startups than in any other part of the architecture across the last ten years.
Portfolio Update
ICM HPQC portfolio companies continue to ride the momentum of the sector. Diraq and Salience Labs were included in the EE Times’ Silicon 100, a recognition of their progress and continued disruption in the semiconductor space. Diraq was the only Australian company to make the list!
Diraq secured $330K from the Australian government to explore how quantum computers can enhance the performance, sustainability and security of energy networks. Diraq will build software that allows non-experts to perform experiments on Diraq’s systems. Read more.
Q-CTRL announced that its new maritime quantum navigation solution was successful in field trials at sea. The result reinforces Q-CTRL’s leadership in the development and
deployment of quantum-assured navigation solutions which allow navigation without relying on conventional positioning, navigation and timing signals.
This follows on from Q-CTRL’s successful field trials earlier this year which verified commercial quantum advantage in both ground and airborne GPS-free navigation. Read more.
In the quantum computing arena, Q-CTRL’s Fire Opal product is now available on Rigetti’s Ankaa-3 quantum computer, an 84-qubit quantum computer accessible through Rigetti’s cloud platform. Read more.
Flash Snap: Quick Link Roundup of the Latest Plays (In Case You Missed Them)
80,000 Hours breaks down AI 2027, a report of a possible future where artificial intelligence radically transforms the world in just a few intense years, in an easy to understand video [Youtube]
Elon Musk describes an ambitious five-year goal to scale up to the equivalent of 50M Nvidia GPUs [PC Mag]
D-matrix explain why data centres struggle to keep up with the demands of generative AI [D-Matrix]
Meta’s lagging AI lab pushing billions of dollars of investment in AI data centres and talent [Semianalysis]
OpenAI and Softbank's ambitious US$500B Stargate AI project has faced significant scaling back amid leadership disputes between Sam Altman and Masayoshi Son. [Yahoo Finance]
Google has increased its capital expenditure budget to US$85B, primarily to expand AI infrastructure through data centres and custom chips amid surging computational demands. [Gizmodo]
Google has announced a US$25B investment in data centres and AI infrastructure, plus US$3B for hydropower plant modernisation. [CNBC]
xAI has entered negotiations with Saudi Arabia's Humain and another firm to lease data centre capacity, seeking cheaper energy and political advantages. [Bloomberg]
Hailanyun has launched a US$223m underwater AI data centre near Shanghai that uses seawater cooling and offshore wind power to reduce environmental impact. [TechRepublic]
The United Nations has called on tech companies to power AI data centres with 100% renewable energy by 2030 while managing water usage responsibly. [Bloomberg]
Nvidia and AMD have received US approval to resume selling specific AI chips to China, reversing previous Trump-era restrictions on such exports. [Yahoo Finance]
Alphabet's Tensor Processing Units have emerged as a potential challenger to Nvidia's GPU dominance in AI infrastructure, particularly for inference workloads. [The Motley Fool]
Amazon has built Project Rainier, a massive AI supercomputing cluster using hundreds of thousands of Trainium2 accelerators, for Anthropic across multiple US data centres. [The Register]
Important Note:
The information in this article should not be considered an offer or solicitation to deal in ICM HPQC Fund (Registration number T22VC0112B-SF003) (the “Sub-fund”). The information is provided on a general basis for informational purposes only and is not to be relied upon as investment, legal, tax, or other advice. It does not take into account the investment objectives, financial situation, or particular needs of any specific investor. The information presented has been obtained from sources believed to be reliable, but no representation or warranty is given or may be implied that it is accurate or complete. The Investment Manager reserves the right to amend the information contained herein at any time, without notice. Investments in the Sub-fund are subject to investment risks, including the possible loss of the principal amount invested. The value of investments and the income derived therefrom may fall or rise. Past performance is not indicative of future performance. Investors should seek relevant professional advice before making any investment decision. This document is intended solely for institutional investors and accredited investors as defined under the Securities and Futures Act (Cap. 289) of Singapore. This document has not been reviewed by the Monetary Authority of Singapore.
ICM HPQC Fund is a registered Sub-fund of the ICMGF VCC (the VCC), a variable capital company incorporated in the Republic of Singapore. The assets and liabilities of ICM HPQC Fund are segregated from other Sub-funds of the VCC, in accordance with Section 29 of the VCC Act.
Get to know us better:
Comments