The increasing power consumption and heat generation of processors and other datacenter equipment have brought liquid cooling into the spotlight. The growing interest in this technology is further evidenced by recent investments made in the field.

One notable development is the acquisition of CoolIT Systems, a long-standing player in the liquid cooling market, by global investment company KKR. The deal, reportedly valued at $270 million, is aimed at enabling CoolIT to expand its operations and serve a wider range of global customers in the datacenter market. This market encompasses enterprise, high-performance computing (HPC), and cloud service provider segments.

KKR’s investment in CoolIT indicates its anticipation of a profitable return. However, their statements regarding the acquisition also reflect a recognition of the challenges facing the datacenter industry in terms of sustainability. Kyle Matter, Managing Director at KKR, emphasized the increasing data and computing needs and their potential environmental impact. He expressed a belief that liquid cooling plays a crucial role in reducing the emissions footprint of the digital economy.

Projections suggest that liquid cooling will witness significant growth, potentially capturing up to 26% of the datacenter thermal management market by 2026. This is driven by the deployment of more high-performance infrastructure. CoolIT, which is soon to be acquired, has already demonstrated its growth potential by securing a spot on the Financial Times’ list of fastest-growing US companies this year, ranking at number 218.

Alan Priestley, a former technical marketing manager at Intel and currently a VP analyst at Gartner, highlighted the necessity for many companies to invest in liquid cooling to address the challenges associated with managing high-performance servers. As processors become more powerful, liquid cooling offers an effective solution to address heat dissipation concerns and optimize server performance in datacenters.

According to Priestley, CPUs currently consume around 250W to 300W of power, while GPUs range from 300W to 500W. For servers handling demanding workloads such as AI training, those equipped with up to eight GPUs can draw as much as 7-10kW per node.

Priestley further explained that datacenters are striving to increase rack densities by incorporating more memory per node and higher-performance networking. Accommodating such heightened performance requirements necessitates increased power consumption.

Andrew Buss, a senior research director at IDC, concurred with this assessment. He emphasized that as chip or package power densities continue to rise, liquid cooling becomes a more efficient and preferred option.

Buss highlighted that support for direct liquid cooling loops is now being integrated into many modern datacenter facilities and colocation providers. He pointed out that companies like Atos/Bull have embraced direct contact liquid cooling loops for their power-dense high-performance computing (HPC) servers. This allows them to fit six AMD Epyc sockets with maximum memory, NVMe storage, and 100Gbps networking into a compact 1U chassis, all cooled by a custom cooling manifold.

The growing demand for higher performance and power-intensive applications is driving the need for efficient cooling solutions like liquid cooling in datacenters. By adopting liquid cooling technologies, datacenters can effectively manage the increasing power requirements of advanced processors and GPUs while maintaining optimal performance and mitigating potential heat-related issues.

According to Moises Levy, an expert in datacenter power and cooling research at Omdia, the global adoption of liquid cooling is expected to continue increasing.

Levy suggests that while liquid cooling has reached or is nearing a tipping point for specific applications with compute-intensive workloads, its widespread adoption in the broader datacenter market is still on the horizon. He highlights that direct-to-chip and immersion cooling technologies are likely to be the primary disruptors, projected to have the highest compound annual growth rate (CAGR) in the coming years.

Direct liquid cooling, supported by CoolIT, involves circulating a coolant, typically water, through cold plates directly attached to components like processors. This type of system is relatively easier to implement within existing rack infrastructure.

On the other hand, immersion cooling submerges the entire server node in a non-conductive dielectric fluid coolant. Specialized racks, some of which position the nodes vertically instead of horizontally, are typically required for this type of system. Immersion cooling tends to be favored for new-build server rooms.

As liquid cooling technologies continue to advance, their increasing adoption is expected to bring significant benefits to datacenters in terms of improved efficiency and enhanced cooling capabilities.

European cloud operator OVHcloud has developed a unique system that combines two cooling approaches for optimal performance. Their innovative solution involves utilizing water blocks attached to the CPU and GPU while employing immersion cooling with a dielectric fluid for the remaining components.

While OVHcloud currently reserves this system for their cloud infrastructure handling intensive workloads like AI, gaming, and high-performance compute (HPC) applications, they have indicated potential future expansion.

In a similar vein, GlobalConnect, a leading data center colocation provider, plans to offer immersion-based cooling as an option to all their customers. Teaming up with immersion cooling specialist GRC, GlobalConnect announced their system deployment in February. They aim to gradually introduce this advanced cooling technology across all 16 of their data centers located in Denmark, Norway, Sweden, Germany, and Finland, based on customer demand.

The question arises: Can liquid cooling help achieve sustainability objectives? OVH shared that its combined system is significantly more efficient than traditional air cooling methods. They claim that in tests, their cooling system achieved a favorable partial power usage effectiveness rating (PUE) of 1.004, which specifically measures the energy consumed by the cooling system.

However, Buss, an industry expert, urged caution in adopting liquid cooling and emphasized the need for careful consideration, particularly in waste heat management. He highlighted that implementing “liquid cooling done right” can certainly contribute to enhanced efficiency and environmental sustainability by reducing reliance on compressor-based cooling and leveraging heat-exchanger technology to maintain optimal cooling loop temperatures.

Nevertheless, Buss emphasized the importance of proper implementation, as simply discharging the heat into the environment, such as a lake or river, can have detrimental effects. Therefore, the design of the ultimate heat path should be carefully planned to maximize reuse opportunities whenever feasible.

The European Union (EU) has recently expressed its desire to see more cities utilizing waste heat from data centers to heat residential homes. However, challenges arise because the heat generated is often not at a sufficiently high temperature, necessitating additional energy consumption to address this limitation. Despite these obstacles, some data center operators, like QTS in the Groningen region of the Netherlands, have ventured into exploring such initiatives.

In the previous year, the United States Department of Energy made investments in projects aimed at reducing energy consumption for cooling purposes in data centers, albeit with a relatively modest funding amount of $42 million. Additionally, we highlighted the swift adoption of liquid cooling by Chinese data centers as a response to new government regulations.

Among the liquid cooling vendors that secured investments was Iceotope, a UK-based company that received £30 million ($35.7 million at the time) in a funding round led by a Singapore-based private equity provider, with a focus on penetrating the Asian market.

Intel also forged a partnership with Green Revolution Cooling to explore liquid immersion technology. However, the chip giant recently decided to halt its plans for a $700 million research and development lab dedicated to cooling technology in Oregon, as part of its cost-cutting measures.

A faded felt bicorne hat worn by Napoleon Bonaparte sold for $2.1 million at an auction on of the French emperor’s belongings.

Yes, that’s $2.1 million!!

The signature broad, black hat, one of a handful still in existence that Napoleon wore when he ruled 19th-century France and waged war in Europe, was initially valued at 600,000 to 800,000 euros ($650,000-870,000). It was the centerpiece of Sunday’s auction collected by a French industrialist who died last year.

But the bidding quickly jumped higher and higher until Jean Pierre Osenat, president of the Osenat auction house, designated the winner.

‘’We are at 1.5 million (Euros) for Napoleon’s hat … for this major symbol of the Napoleonic epoch,” he said, as applause rang out in the auction hall. The buyer, whose identity was not released, must pay 28.8% in commissions according to Osenat, bringing the overall cost to 1.9 million euros ($2.1 million).

While other officers customarily wore their bicorne hats with the wings facing front to back, Napoleon wore his with the ends pointing toward his shoulders. The style, known as “en bataille,” or in battle, made it easier for his troops to spot their leader in combat.

The hat on sale was first recovered by Col. Pierre Baillon, a quartermaster under Napoleon, according to the auctioneers. The hat then passed through many hands before industrialist Jean-Louis Noisiez acquired it.

The entrepreneur spent more than a half-century assembling his collection of Napoleonic memorabilia, firearms, swords and coins before his death in 2022.

The sale came days before the release of Ridley Scott’s film Napoleon with Joaquin Phoenix, which is rekindling interest in the controversial French ruler.

Rebecca Barnatt-Smith explains how marketers and content creators can use AI-powered predicative analytics, content personalisation and scheduling tools to create successful social media campaigns.

Is artificial intelligence (AI) the next big thing for social media marketers?

With over 4.26bn social media users to serve, AI is set to transform targeting and improve content personalisation for a more focused marketing future.

AI is not a new phenomenon in the marketing world. When surveyed, over 56pc of chief marketing officers (CMOs) said they use automated assistants for content personalisation and tracking consumer insights. AI-driven social strategies are just the next step in a fast-approaching digital future of campaigning.

However, could a push for AI-infused social campaigns pose ethical concerns for future marketers? From breaching consumer privacy to decision system bias, with great technology comes great responsibility.

Here we look at AI’s impact on social media marketing and discuss some of the best AI-infused platforms that are tipped to lead social strategies in 2023.

How can AI improve your social media?

Using AI, you can quickly segment large demographics into targeted groups, track viral trends and schedule personalised content responses in seconds.

If you want to compete against commerce giants and industry leaders, your social content should be consistent, compelling and customised to each and every consumer. Here are some insights into how AI can help.

Content personalisation

In 2023, 73pc of shoppers expect brands to offer them a personalised experience and content that speaks directly to their values. AI can enhance a brand’s personalisation potential in a number of ways.

Automatically harvesting behavioural and historical consumer data, AI-generated platforms can quickly learn about a user’s interests and predict what products or services they’d be most likely to interact with, resulting in a hyper-individualised experience that can boost engagement and increase the chances of conversion.

However, with 69pc of consumers now concerned about how their data is collected and used on mobile apps, it’s important to use content personalisation tools with caution.

“As consumers continue to learn and become more informed about their data rights and how their data is currently used, I expect we’ll see more and more calls from consumers to have their data protected,” claims Swish Goswami, CEO of browser extension platform Surf.

The key here is to keep your consumers in the loop. Give your followers a chance to choose what they share, and make sure the data you collect is transparent. Personalised ads, posts and targeting is a business game changer, as long as you have consent.

Automated content posting

Creating content for your brand is the driving force behind audience engagement.

While experts recommend that brands upload social media content daily, this process can be time-consuming. Using AI-driven social media tools, marketers can feel the pressure drain away, as automated assistants not only create original content formats but automatically schedule them too.

For example, AI-infused content planner Sprout Social can generate personalised tweets that reply to fans and followers in seconds. Instead of physically manning social channels and checking for replies, Sprout Social monitors a brand’s comment section before analysing the tone and sentiment of a reply. Sprout can then suggest an auto-response that aims to carry on the conversation between the brand and the consumer.

While automatic replies can pose ethical questions about a brand’s true identity, Sprout Social ensures that before an automatic reply is posted, the social media manager is able to review and edit the content. This guarantees that the brand’s voice still has a human tone when connecting with its audience.

Hubspot is also a nifty tool to have under your belt, especially if you’re struggling to develop new content ideas. By simply pasting a content link into Hubspot’s content generation feature, it uses AI to quickly analyse the metadata and create an original social post.

Social media advertising

Social platforms are the perfect vessels for advertising success. Whether you choose TikTok or Instagram, with the ability to post a pop-up on a user’s scroll-down feed, or a sponsored TikTok that blends seamlessly into a For You Page, social channels allow for a more organic future of ad placement.

However, with so many brands utilising social media, it can be hard to make your ad stand out from the crowd. Your ads must be full of compelling captions, quick links to your online store and contain a personalised hook for your target consumer.

Using AI, brands can optimise their ad performance on social channels. With the ability to analyse historic campaigns and current trends among industry leaders, AI-driven ad tools such as Sprinklr can make recommendations for smarter campaigns that drive better results.

Also, AI-infused ad strategies are more likely to be personalised to each user’s feed. AI tools like Phrase can generate customisable ad phrasing that adapts to target individual customers. This is a great way to ensure your ad captions remain fluid and speak directly to a diverse set of leads.

Predictive analytics

While it’s easier than ever to track social media performance, acting on your results can be tricky. AI-generated monitoring tools utilise the data harvested on content engagement, clicks and consumers, and turn these insights into predictions for new campaigns, content formats and new target groups to work on.

The key here is to take these predictions and turn them into content campaigns that frame the values of your brand. It’s also important to do your own research before jumping into an AI-generated content campaign, as just like humans, AI can have a decision system bias.

“AI is fallible and in a perfect world should be used critically, responsibly and democratically,” says Annie Brown, founder of the creative sharing platform Lips. “AI is only as fair and accurate as the algorithm, and the algorithm is only as fair or accurate as the human-generated information it gathers.”

For example, if the only data your AI tool collects is from a specific consumer group, it’s likely to inherit the same biases. Therefore, it’s important to perform your own content research if you want your brand voice to remain objective on social media.

However, with more data to inform their strategy, brands that use AI to influence their social campaigns are more likely to see higher conversion payoffs.

As social platforms continue to become more visual, AI can also enhance video and image analysis. For example, AI algorithms can now identify certain aspects of Instagram images and TikTok videos, making it easier to gather more data on a user’s interests and behaviours.

Visual analytics could help a brand improve its content styles as AI tools learn more about audience preferences and the formats going viral.

Could AI take social media marketing to the next level?

AI can enhance the experience a consumer has with a brand on social media. With predictive analytics at play, the content targeted users receive is more likely to speak directly to their values.

While there are still ethical concerns surrounding an AI-infused future of campaigning, there’s hope on the horizon for data-sharing transparency and the impact of algorithmic biases as both consumers and marketers take control of how data is gathered and shared.

As machine learning gets even smarter, the possibilities are endless for brands that want to get close to their leads. From automated responses to automatic content creation, the future of social media marketing is AI-driven.

By Rebecca Barnatt-Smith

Rebecca Barnatt-Smith is a freelance content writer and multi-media marketing executive at Solvid Digital, specialising in social media trends and widespread digitalisation in the marketing sector.