Hello Reader,
It's a long May Day weekend in many parts of the world. I hope you are having a restful time with your family and loved ones. And as it happens, today is my birthday. So thank you, sincerely, for choosing to spend a few minutes of it with me.
This week, I share my thoughts about data centre water use, a new breed of AI-powered cybersecurity tools, and how workers might move ahead in the era of AI.
Water moves from footnote to headline
There was a time when 20-30MW data centres were considered massive, and city utilities had a comfortable surplus to supply them, said Mohammad Sherafatmand, CEO of Hydroleap, at his firm's networking event earlier this week.
But those days are no more, as the number of hyperscale campuses surges and utilities are confronted with water demand at a scale far beyond what they can supply. The result? A growing shift away from using potable water for data centre cooling.
But why not move away from water for cooling altogether? There are plenty of reasons, and I inked down three here: industry inertia, the very real challenges of retrofitting, and a global surge in AI workloads. The challenges of retrofitting bear mentioning, in particular, given how Singapore has indicated it will eventually apply more stringent standards on existing data centres.
If only we could learn more from one another. But alas, the data centre industry remains as secretive as ever. About a couple of years back, the engineering head of a large data centre operator contacted me. He wanted to show me the fruits of his team's labours, a complex cooling system upgrade of an operational facility in Singapore. Spoiler: the plan was eventually shot down by upper management.
Really grateful to everyone I caught up with at the Hydroleap Networking Event and the many new people I met. It is both humbling and gratifying when engineers and seasoned veterans of the industry tell me they enjoyed what I wrote, and how it has helped them.
Fight AI with AI
To fight AI, cyber defenders must use AI. That was the recurring topic I kept hearing over the last 12 months as I interviewed industry experts in cybersecurity for editorial work I do in that sector. I finally saw it with my own eyes at ElasticON Singapore.
You see, cybersecurity investigations are often complicated, laborious affairs that can take anywhere from hours to multiple days. But as AI is increasingly used to find software flaws and craft exploits, the traditional pace of investigations will no longer do.
Turns out AI can collapse lengthy investigations into minutes. The concept is simple: tap into an extensive pool of data sources across the enterprise through a lightning-fast, well-indexed search engine, with the power of generative AI on top. In fact, the demonstration of Elastic's Workflows for Security reminded me of Claude Code, albeit geared squarely towards cybersecurity investigation and remediation.
Be hungry, want to change
It is May Day, an international day for honouring workers. In Singapore, Prime Minister Lawrence Wong, in his May Day rally, said: "I cannot promise that there will be no disruption. But this I can promise you: As our economy transforms, we will create new and better jobs. We may not be able to protect every job, but we will protect every worker."
This made it into the Straits Times, a government-backed publication, with the headline: "Govt can’t protect every job in AI shift, but will protect every worker: PM Wong"
I have a simple method for reading Singapore publications. Bad news here is always sugarcoated through saccharine framing. It is only the terrible news, the kind that can no longer be glossed over, that actually shows up directly in headlines. Put simply, the Singapore government is bracing for a hard wave of AI disruption, and is preparing workers through messaging and support.
Beyond government policy, though, there's the harder question for each of us. The media prepared ahead by interviewing leaders and personalities for their views of the job landscape. And my friend, Ian Yong, quoting a podcast on CNA, observed that we need to be "more hungry" to compete against others.
I agree with the idea. However, I also feel strongly that being hungry is not enough. It must be paired with a strong desire to learn and a willingness to adapt. All three need to be present for real change. Hunger gets you started, but learning and adapting take you the rest of the way.
This is why I am writing more than ever before, without AI. I am getting on podcasts. I am thinking up fresh ideas for marketing. I am taking on consulting engagements. I am also using AI extensively, both to build apps that do what I couldn't before, and for content tasks.
Because I'm honest with myself: the writing path that I fell into for close to two decades is no more. None of which is comfortable. But discomfort beats irrelevance. I am striving to increase my value. Are you?
As usual, you can reply to this email to reach me.
PS: I fired up the Asus GX10, an ultra-compact desktop "AI supercomputer" based on the Nvidia DGX Spark yesterday. I haven't set up a Linux desktop since my teens, but this is looking good.
Regards,
Paul Mah
