Daily Musings

Evening all

I always thought a career in finance would bury my dreams of becoming a wine farmer. Turns out there is hope after all as the latest provider of leveraged loans are a syndicate of agricultural firms.

Seems the street wisdom that investing is an art, not a science, has never been truer, in light of this ode to creative capital. While planning extensive hands-on research into the wines I will produce, I share today’s musings.

Cheers
Philip

Today’s Stories

DataCrunch gets fresh USD 13m in a chunky seed round. Total funding now @ USD 18m. Lead: byFounders VC. Contributions from J12 Ventures and a roster of prominent angel investors. Continuous backing comes from Finnish insurance company Local Tapiola and local bank Nordea. The startup provides AI computing infrastructure on-demand enabling cost-effective access to compute clusters — specifically for AI workloads. A clear value proposition as the global AI momentum is starting to hit significant bottlenecks — foremost the immense cost and increasingly limited supply of GPUs. The company operates data centers on its home turf in Finland and Iceland - here renewable energy prevails and the cooler temperatures lower the necessary cooling costs, which are a sizeable share of the total operating expenses. The start-up launched in 2020 and increased annual top line by 250% since. Not the venture growth seen in other cases but presumably the long ramp up phase to kick off operations and get the necessary infrastructure in place, I am keen to see how top line progresses from here.

While Datacrunch makes AI operations more accessible, Oriole Networks is looking to solve the looming energy constraints:

Oriole Networks—the venture leveraging light to train LLMs more efficiently—has secured an additional USD 22M in funding. Led by Plural, with participation from existing investors: UCL Technology Fund, XTX Ventures, Clean Growth Fund, and Dorilton Ventures. Founded in 2023 as a UCL spin-out, the company’s innovative photonics technology promises to train LLMs up to 100x faster. This clear value proposition has crucial implications for the energy consumption of data centers, especially as demand is expected to triple by 2035, challenging the currently planned power supply. With total funding now @ USD 35M, Oriole aims to scale its technology by 2025.

Chart Art

And in case Oriole does not succeed, perhaps nuclear reactors can keep our incessant ChatGPT queries alive.

AI's push is driving up data center rack densities and energy demand is surging. MSFT's electricity consumption is set to skyrocket—5-6x higher by 2030 than the in 2020 expected levels. I’m not a formidable gambler as recent Roulette attempts have proven, but I'd wager these forecasts could climb even more as we get closer, especially given the tech sector's need for a reliable power supply. With that in mind, nuclear is catching the eye of global big tech. Accordingly SMRs are seeing a spike in demand, as the recent EUR 452m funding for X Energy highlights. Add to that the US DOE plans to channel USD 900m for SMR deployment at 'fleet level.' The proposed funding splits into two tiers: Tier 1: USD 800m earmarked for up to two pioneering project teams rolling out first-of-its-kind reactors. Tier 2: The remaining USD 100m reserved for follow-on 'nth-of-a-kind' projects, aiming to clear early hurdles, bolster the supply chain, and enhance confidence in costs and timelines.