Posted by Dumb, Smart & Crazy – Masa, the founder of SoftBank, categorised people in these 3 categories.
Crazy people play Power Games where 1 hit can change the entire game & make you a world champion. (Read Power Law of returns). In AI this means – playing in zone where Price points aren’t established yet.
I cover Tacit Knowledge with our actual reach to 91 VCs for Deeptech AI Infra for Enterprise Decisions.
Below I cover is distilled hard-earned experience of – Apart from VCs, from 10s of Enterprise AI conversations with top MDs/Chairman/CEOs/Strategy heads of companies with $1B+ revenue – The early adopters’ market. Not traditional sales. It includes soonicorn founders & CEOs of listed companies who built companies from scratch. It’s collective hard-earned – our & their experience combined. Even below declines – some declines came after months & months of effort & are valuable.
What’s Tacit knowledge & vs conventional wisdom?
In simple words, Tacit is Unique insight & perspective built form personal observations.
If crazy people aren’t making random bets, then how do founders & VCs spot outliers?
Because crazy have a unique perspective and insight – Tacit Knowledge.
Crazy people make money & look dumb. Smart people make smart arguments and sound smart, but seldom make money
Insight->Perspective = Tacit Knowledge -> Contrarian Decisions & Actions = Possibile Venture returns
I cover that unique contrarian perspective on Deeptech AI Infra for Decisions & World Models – beyond LLMs/SLMs/Workflows in the blog.
Crazy are making calculated moves based on tacit knowledge & unique perspective. Smart people go by conventional wisdom.
Because smart play only what has conventional signals in the market, what US VCs say on Twitter, what a blog on SaaS metrics says.
How crazy people make money & why share tacit knowledge in a blog?
I share in blog because we also have built on scarce knowledge & unique perspectives of other crazy founders & VCs.
Crazy people aren’t making random bets – they see contrarian opportunities based on their tacit experience, gut & experience relevant to category momentum that smart people can’t see with Excel, knowledge or even hands-on work.
Tacit Knowledge 1 – LLMs/SLMs/Workflows automation are 10-15yr old tech, Context Engineering is 6-8yr old, Vertical solutions are built in-house, more so with AI.
Any verticalization or context engineering is already doomed. The CEOs of large companies we talked to – Google, Glean are already pitching there. Incumbents are fast-building Context Engineering. Vertical solutions are getting built in-house. Even banks globally aren’t ready to pay more than $30-60k. They trim you the moment bill goes beyond $300k annually
Tacit Knowledge 2 – Crazy VCs are small in number – estimated at 75 in India. Most are smart- data speaks for itself.
Conventional market analysis tends to overweight what is already visible, measurable and priced. Our thesis is that the largest AI opportunities may emerge precisely where those signals don’t yet exist
We decided to go to only 100-odd VCs. Below is a summary based on our research. Only 76 VCs led a round > $500k consistently at least 1 investment per year in the last 3 years in Deeptech, AI, Tech.
Ref, traxcn, cbinsight, inc42, chatGPT & our internal research
Some indication: Only 25 out of 169 VC funds analyzed by Crisil returned a distribution to paid-in capital more than 1 in around 7 years. That’s the result of moves that sound smart & safe at surface. After ~7 years, a nominal 1× return itself represents roughly a 30% real loss after Indian inflation. The real loss among those 144 could therefore be substantially larger. (Exact numbers can’t be verified)
Tacit Knowledge 2 – Small Decision & Reasoning World models don’t take much capital – $2M to begin with
I see many smart VCs have given away the fact that AI platforms are all eaten & it needs crazy money outside of their fund size for Deeptech AI. This is myth caused by US frontier models.
If Palantir is to be taken as a benchmark, There’s lots of value for Indian player to do that in India say for Finance & take it to the world.
Isn’t that crazy – Automate gut & intuition for decisions, escale the price points & competition?
Moreover – by aligning to niche topic – customer can go much beyond price point.
Tacit Knowledge 3 – Crazy or Smart? Preparing for Black Swan Events & AI Automation for Gut & Intuition: Alok Goyal of Stellaris Venture Partners, beautifully put that some of the most valuable enterprise work is that has never been outsourced and therefore has no established price. Read here
So, when we pitched to CEO of very large companies & Banks – Vaayu’s infra automating gut & intuition – Because there was no established price benchmark, our enterprise proposals were initially structured at $1–2M for the first year, with potential $2–4M lifetime value & postioned us distinctly
LLMs/SLMs make decisions faster, Context Engineering makes decisions complete Vaayu World Model makes Decisions optimal.
Tacit Knowledge 4 – Real Examples & case of World Models – to automate gut & intuition
- Reduce GNPA by 10-30 bps with world models – Indian banks per RBI have non-performing assets of ₹ 4.32 lakh crore. Reducing GNPA by 10bps means 40,000 cr benefit for banks. Just one use case. See list of Non performing assets by RBI. Read more
2. Swarm AI & Defence with World Models
God forbid, if & when a cyberdefense attack happens with Swarms anywhere in the world like what you must have heard from Anthropic or OpenAI,skyrocket the demand for such products will sky rocket. Our team created core swarm algo in lab.
Tacit Knowledge 5 – Palantir Like moat
We have figured out that World Models have way to create a Palantir-like moat. Not only technology. But also the distribution moat. Innovators & Industry Deeptech AI Alliance Labs is a registered Section 8, non-profit company – A focused research organization for same.
Summary: Craziest work in AI is in zones where price poitns & moats aren’t established. Nor these were ever automated. This goes beyond LLM/SLMs/Workflow automation & context engineering. Decision Intelligence, automating gut & intuition with world models is that perfect zone.
When crazy goes up against smart, crazy wins – Masayoshi Son, Founder & CEO, SoftBank
4800 Indians control the entire Capital & decisions in India