Hi, I'm Nick. I work on AI infrastructure strategy across APAC.
Everyone talks about models. I'm more interested in what's underneath: chips, power, land and the capital it takes to bring them together.
I'll be posting about compute economics, data centres and energy in the region. Views my own.
opus 5.5 is still my favorite model. but anthropic has no always-on agent, so more and more of my personal life now runs on grok bot. i suspect this is the game changer.
Claude Duet is an early research preview, available through a waitlist on Pro and Max plans in the US. It can still misjudge what’s urgent, and we’ll keep tuning it based on what early users tell us.
Join the waitlist: https://t.co/kNwzHHeYrO
Read more: https://t.co/GT1xV1RJ7f
@zephyr_z9 genuine q: how much of the q2 to q3 deceleration is demand vs capacity constrained? if it's capacity, the q4 reaccel case depends on how much new compute lands this quarter
@BoringBiz_ Not only that, backing up open models are also a genius move of building a moat against future trillion dollar labs building their own chips.
ANTHROPIC'S FIRST JAPAN DATA CENTER COULD COST $6.3 BILLION
TIL: Japan's NTT is reportedly in final talks to build a massive 500MW AI data center for Anthropic in Osaka.
The project could begin operations as early as 2028. But securing that much electricity may be the real challenge...
[Get more breaking AI infrastructure news straight from Korea and around the world, on Antfeed]
#Anthropic #Claude #NTT #Japan #AIInfrastructure #DataCenters #ArtificialIntelligence $NVDA $AMD $AVGO $MSFT $GOOGL
Southeast Asia (Malaysia, Thailand, Indonesia) used to be where power was abundant and cheap. Not anymore. After massive data centre builds over the last 3 years, power there is now almost as constrained as in East Asia. Secure your power now or get in line.
As a normie who's late to the party, I've been playing with agents for a few weeks now and it's life changing. I recruited 10-ish agents to support the team and easily tripled my team's productivity.
My favorite is Chief, my own personal chief of staff. I've played the chief of staff role for years, and honestly it has done a significantly better job than I did. Give it your email, calendar, Slack and Notion access and you're set.
Cost of deploying new compute has been trending up YTD. Lead times are getting longer too with memory, storage, and networking equipment shortages. Everyone is compute constrained and is grabbing what they can.
Allegedly “the most important chart” for understanding the current state of the AI economy. The model layer seems to be going through a period of commodification as non-frontier models saturate everyday use while frontier models compete fiercely. Compute layer sees massive increase in total demand for compute regardless of the source or use case.