Latest post
TechnologyHyperscalers finance AI chips over six years but the hardware is obsolete in two or three. That gap quietly sets the compute price every AI product is planning around.
AI Advisor · Founder, 256 Technologies
Hitesh
Sarda
I help leadership teams cut through AI hype and make evidence-based decisions: what is feasible, what the ROI really is, and the smallest pilot that proves it. 25+ years building production systems, including work at Citi, Virtusa, Yash, and ValueLabs. Today I run 256 Technologies, an applied AI lab in Hyderabad, and I have been writing here since 2009.
181
posts
7
topics
2009
since
New book · Free PDF
Implementing AI Agents in Your Company
A Blueprint for Innovation
A practical, hype-free playbook for the managers and operators who have to turn AI agents into real business outcomes. Read it free, no signup.
What I do
Book an advisory hour →
AI Advisory
Acting as your fractional Head of AI: strategy, feasibility, ROI modeling, vendor assessment, and roadmaps grounded in real engineering constraints. No hype, no vendor lock-in.
AI Executive Training
A six-week program that builds an AI-native leadership team: hands-on executive workflows, governance and risk, and a 90-day adoption roadmap that moves leaders from watching to leading.
MVP Development
Tangible code beats theoretical roadmaps. I build the smallest prototype that answers the key technical and business questions, across agents, computer vision, and ML, before you commit to a full build.
Recent Writing
Full archive →
The AI boom is funding a two-year asset with six-year money
Hyperscalers finance AI chips over six years but the hardware is obsolete in two or three. That gap quietly sets the compute price every AI product is planning around.
Writing code got free. Reviewing it is the job now.
Microsoft's engineers merged 24% more pull requests with AI. The constraint didn't vanish; it moved to review, and most teams are still counting the wrong thing.
Automation is cheap. Understanding is the bill you don't see.
AI makes producing work nearly free while draining the human judgment that catches it when it's wrong. Deciding which understanding to keep in-house is now strategic.
Someone is subsidizing your AI compute
The AI infrastructure boom is partly lending to itself. That changes the cost curve you're planning around, and most companies building on it haven't noticed.
AI does your hardest work, not your messiest
A frontier model just produced a novel math proof while similar tools still fumble routine operations. The dividing line is verifiability, not difficulty.
Why the smartest model can make your product worse
Frontier models keep setting records, yet what decides whether a real-time AI product works is the second your user waits, not the model's score.
Your backlog is the only benchmark that matters
Public coding benchmarks have decoupled from real work. The teams getting value from AI build a small evaluation from their own merged pull requests instead.
When the regulator becomes your product manager
Every new car in Europe now ships an eye-tracking AI model no buyer chose. How that feature is failing is a preview of what happens when AI gets mandated onto a product.
