team worked really hard for this and kudos to both the business and tech team for joining the Linux’s Agentic AI Foundation 🚀
time to kick things into overdrive, we’re looking for people who want to build with us, especially on the following use cases
> agentic payments
> agent governance and control
> agentic identity
> stablecoins and defi
> agent builders and launchers
higher 🪁 @GoKiteAI
Kite is proud to join the @linuxfoundation’s Agentic AI Foundation (AAIF), working alongside @OpenAI, @AnthropicAI, @blocks, and members including @Google, @Microsoft, @awscloud, @Cloudflare, and @business to advance open standards and interoperable infrastructure for the agentic AI era.
With AAIF’s community stewardship of foundational projects such as MCP, https://t.co/0Cmx1vBRyM, and goose, Kite will contribute our perspective on verifiable identity, programmable governance, and stablecoin native settlement so autonomous agents can execute end to end with clearer policy boundaries, stronger accountability, and auditable outcomes at production scale.
🧉 Spending the day at Binance Club House during @CoinfestAsia.
This is exactly the kind of event I enjoy. A great setting, views, and plenty of interesting insides throughout the day.
Thanks to @binance for another 10/10 event.
Henry @Henryleemr and I will be at @binance Clubhouse Bali tomorrow.
If you’re there too, come say hi. We’d love to connect with builders, users, and the wider community.🙌🏻
#bali#coinfest
not only does AI make building less of a moat now with vibe coding, but its weakening the advantage of distribution too
why? because agents dramatically reduce switching costs
with agents users can migrate from one platform to another much easier
migrating your playlists from youtube music to spotify? use an agent
migrating your journal / code / emails / calendars to a new platform? use an agent
the harder it becomes to maintain user stickiness, the weaker incumbent's networks effects become in actuality
good news for entrepreneurs perhaps
people often say stocks return 10% a year over the long run
but that's actually specific to U.S. stocks during arguably the greatest century any economy has ever had as
post-war industrial dominance + winning the Cold War + the dollar becoming the global reserve currency (and for oil)
attracting the world's best talent + creating the internet + building the biggest tech companies
theres a lot of underlying assumptions to the "buy SnP and forget" rule
history might remember the internet less for changing how we live and more as the training data dump for AI
reddit already makes so much more millions $ selling posts to AI labs than most of us realize
same for stack overflow
maybe this was the whole purpose of the internet all along, to extract our thoughts into database to invent AGI
global cybersecurity costs in 2025: $1.14T/yr
global corporate economic profit in 2025: $1.2T/yr
companies are spending almost every dollar they make to defend what they built
and every new AI model expands the attack surface
does this mean with every new model released profits actually decrease?
hackers using AI are on a rampage now
one attacker used fake AI generated hand-held ID materials that still passed exchange’s verification and drained $1.7m from a user and it was not refundable
hedge funds in tradfi are being bombarded by auto generated cyberattacks that used to be done by hand
follow up post soon
hackers using AI are on a rampage now
one attacker used fake AI generated hand-held ID materials that still passed exchange’s verification and drained $1.7m from a user and it was not refundable
hedge funds in tradfi are being bombarded by auto generated cyberattacks that used to be done by hand
follow up post soon
looking kimi3 and qwen3.8-max releases, to just load K3, you need at least 16 H200 gpus
this is wildly different from Kimi K2 which enthusiasts managed to run just on a mac
basically K3 can only run in data centers with industrial power sources aka "not meant for households"
its becoming clear that open source models are becoming not so practically open in this new era
while at the other side of the ocean Micron is projected to be the most profitable US tech company, with $41B operating income estimate for Q4 FY2026
this beats Microsoft, Alphabet, Apple, and Meta!
global demand for tokens just keeps increasing and now data centers produce roughly 35 quadrillion tokens/month
token demand is insatiable while semiconductor shares plummet 👀👀
Samsung and SK Hynix alone expected to cover roughly 50% of South Korea's entire national debt
that’s them to pay ~$435B in combined corporate taxes over FY26–28
the memory boom was a zero-sum game
memory producers (MU, Samsung, SK Hynix, etc.) gained $493B in market cap while memory consumers (Apple, Microsoft, Meta, Amazon, Google) lost $500B
meaning liquidity moved from hyperscalers to memory