Indian Electric Grid – from Passive Infra to Programmable Nodes

Why Solid-State Transformers Are the Technology India Cannot Afford to Miss

One of the most underappreciated engineering achievements in India’s modern history is the unification of its electric grid Five regional grids — Northern, Eastern, Western, North-Eastern, and Southern — historically operated in isolation, each with different frequency profiles and zero power exchange. Unifying them took decades. The final milestone came in 2013 with the 765 kV Solapur–Raichur line linking the Western and Southern regions.

One Nation — One Grid — One Frequency.

We spent years solving for access and reliability — keeping the lights on, wiring the last village, stabilizing supply. That was the right problem at that time.

The lights are on. The internet is everywhere.

And now, something far more power-hungry has arrived: Artificial Intelligence.

The data centers feeding India’s AI ambitions are industrial-scale power consumers, and they are ramping up fast. India’s data center capacity stood at 1.4 GW in 2025. By 2030, we project a fivefold increase to 8–9 GW — more capacity than India has built in its entire history, in just five years. AI data centers alone could demand 50 TWh of additional electricity annually by 2030.

We are building a $30 billion data center ecosystem — and roughly a third of that will go into power systems.

However, here’s the conundrum: AI data centers are fundamentally DC environments.

Every GPU, every server, every rack runs on direct current. The grid delivers AC. Every watt must be converted. Multiple times. In numerous lossy steps.

Modern hyperscale facilities are already moving toward 800V DC internal distribution, with rack densities surging past 130 kW. The old transformer-plus-rectifier model, designed for a world of general-purpose servers, is not built for this. And India’s reliance on diesel generators in data centers today is largely a diesel generator farm — expensive, polluting, and entirely unnecessary if the power architecture is designed right.

And now the grid must learn a new language.

The 130-year-old conventional transformer — a purely AC, passive, fixed-frequency device — cannot speak DC. Yet DC loads are multiplying on every front.

Solar panels generate DC.

EV fast chargers are DC systems.

Batteries store DC.

AI Data centers use DC.

As India races toward 500 GW of renewables by 2030, builds millions of EV chargers, and constructs AI data centers across the country, every single watt passes through a lossy AC-DC conversion that legacy infrastructure was never designed to handle efficiently.

Most of India’s renewable energy sits in Rajasthan and Gujarat. Demand doesn’t. It sits in industrial clusters and cities. We all know this mismatch. The transmission network hasn’t expanded fast enough to bridge this gap, leading to frequent congestion and curtailment. As a result, available renewable power often goes unused because it can’t be transmitted where it’s needed.

But we’ve been too comfortable working around it.

And the consequence is simple: we generate clean power, and then we don’t use it. We are celebrating capacity additions while quietly curtailing output. For instance, on August 11, 2022, about 6 GW of renewable energy had to be backed down due to grid instability.  Not because we didn’t need power. Not because we couldn’t generate it. But because the grid couldn’t handle it.

If electrons can’t move, megawatts don’t matter.

India’s underdeveloped distribution grid is not a lost opportunity. It could be a strategic advantage. The US, Germany, and Japan are facing the same DC integration problem — but they must solve it while dismantling 80 years of sunk AC infrastructure. They are rebuilding a ship at sea. India, by contrast, has a largely blank canvas. Much of its distribution grid is either new, inadequate, or still under construction.

We skipped

landlines and leaped to mobile phones

plastic credit cards to UPI

cable TV to streaming

So why not modernize the electric grid the same way?

The latecomer advantage is real — but only if the latecomer chooses the right technology.

While I want to discuss many innovations that would enable this transition, and we will do that in later blog posts, today I want to focus on one technology – Solid State Transformers. 

A Solid-State Transformer does what a conventional transformer does — steps up or down the voltage — but uses Silicon Carbide or Gallium Nitride power electronics and digital control rather than passive magnetic coupling. The difference in capability is enormous. An SST is active, intelligent, bidirectional, and natively AC/DC compatible. It can take medium-voltage AC from the grid and deliver high-voltage DC directly to a data center rack — in one step, at over 98.5% efficiency. It can enable vehicle-to-grid services, integrate solar storage without conversion losses, and deliver real-time power quality control. All in a form factor 80% smaller than conventional infrastructure.

SSTs are not vaporware. DG Matrix shipped the world’s first commercial units in December 2025. Amperesand — founded by Indian-origin veterans of ABB, GE, Siemens, and Vestas — raised $80M in late 2025 and has committed to installing 30 MW of SST systems in hyperscale AI data centers in 2026. However, the cost challenge is real — SSTs are still 3–10x more expensive than conventional transformers — but the trajectory follows that of solar and batteries: scale drives costs down fast.

India has a chance to build its next generation of grid infrastructure — data centers, EV charging, renewable integration, railways — with SST-native architecture from day one. That window is open now.

The applications are clear: SSTs in AI data centers to eliminate the AC-DC conversion cascade. SSTs in EV charging hubs to enable direct medium-voltage-to-DC fast charging with V2G capability. SSTs in renewable substations to handle the bidirectional, variable-voltage reality of solar and wind. SSTs in Indian Railways traction systems, one of the world’s largest electrification programs. And SSTs at the grid edge, enabling rural communities to be prosumers in India’s clean energy future.

However, we should not merely be a market for SSTs. It should build them. The India Semiconductor Mission has approved ₹1.6 lakh crore in chip projects. SiC and GaN — the semiconductors at the heart of every SST — are exactly the categories India is targeting. The IITs and IISc have deep expertise in power electronics. The talent, the policy, and the market are all present. What India needs now is founders willing to build.

At Speciale Invest, we are actively looking to meet founders working at the intersection of power electronics, grid intelligence, and India’s energy transition. If you are building to take the Indian grid to the next era – SST hardware, wide-bandgap semiconductors, DC-native data center power, EV charging infrastructure, or grid-edge intelligence— we want to hear from you.

The grid India builds in the next decade will power the country for the next fifty years. Let’s make sure it’s the right grid.