Somebody once photographed an Indian general at a briefing. Behind him, mounted on the wall, sat a CCTV camera with the Hikvision logo on it. Hikvision is Chinese. Someone circled it and posted it, and that circle did more work than a decade of policy papers.

The housing on that camera was probably Indian. The plastics, the bracket, the screws. The silicon was not. Neither was the image signal processor, or the firmware. For twenty years, this arrangement was what India called domestic manufacturing.

Manu Iyer has built a fund on the gap between assembling something and inventing it. On 13 August 2026, Bluehill.VC announced the final close of its maiden frontier tech fund at ₹400 crore, the ₹350 crore target plus the full ₹50 crore green shoe.

Check out the video of the conversation here or read on for insights.

The word stopped filtering anything

There was a phase when buying a .ai domain earned a startup a twenty to thirty percent valuation markup. Nothing about the company changed. The suffix did. Manu’s argument is that deep tech is now the same trick with a longer word.

Every startup is somehow deep tech. A software startup is deep tech. A hardware startup is deep tech. Guys copying stuff that already exists somewhere else, deep tech. Guys adding 10 or 15 percent to the status quo, deep tech.

So he stopped using it. Bluehill.VC calls itself a frontier tech fund and defines it narrowly: foundational IP in engineering or science, benchmarked globally rather than against the Indian status quo.

This sounds like semantics until you count the money. SIDBI has a second ₹10,000 crore pool for deep tech fund managers. The RDI scheme carries ₹1 lakh crore. In June 2026, the finance ministry cleared ₹1.25 lakh crore for India Semiconductor Mission 2.0. When that much capital enters a category with no working definition, the definition becomes the filter.

Which is why Bluehill.VC staffed its investment team entirely with engineers.

The entire investing team is just engineers. Generally, VC is largely commerce background people. For us it’s very clear, we want people with an engineering bent of mind.

Founded in 2017, the Chennai firm writes $1 to $2 million first cheques into 15 to 16 companies at Technology Readiness Level 3 to 4, where an idea has left the whiteboard and survived a lab. SIDBI anchors the fund alongside the Kerala and Uttar Pradesh governments. Professor Ashok Jhunjhunwala, whose IIT Madras Research Park ecosystem was valued at ₹55,000 crore, and Vinod Dham, who led Intel’s Pentium design, sit on the advisory board. All of it exists to answer one question.

The biggest challenge is to find the right people who are solving real problems, versus the people who are integrators, sourcing stuff from across the border and reselling it.

The ₹1,500 crore lesson

Under the FAME scheme, electric two-wheeler makers claimed subsidy by battery size, conditional on rising domestic value addition. Hero Electric claimed it. Okinawa claimed it. Roughly ₹1,500 crore went out over five years. Then the audits found the localisation existed on forms, not in vehicles. Clawbacks followed.

Manu thinks the clawback is the least interesting part.

The thing I always point to is the opportunity cost. The actual inventor, let’s say Ather, not having got that support. They could have been much further ahead today.

Ather Energy built its own battery management system, motor controllers and vehicle platform from zero. It got no shortcut, and it got the outcome. Ather listed in May 2025 at ₹321 a share. By August 2026 it traded near ₹1,525, a market capitalisation around ₹60,000 crore and roughly 4.7 times its issue price. Indian institutional capital does not move on thesis. It moves on comparables, and Ather became one.

What the money is buying

optoML, Bengaluru, raised $1.8 million in February 2026 led by Bluehill.VC and A99. It attacks the memory wall.

In a lot of AI chip use cases, 70 to 80 percent of the energy consumed by the chip is not going towards compute, but towards moving the data back and forth.

optoML runs multiply-accumulate operations inside the memory array in the analog domain and swaps copper interconnect for optical links. It has completed a 12nm tapeout with TSMC, signed an MoU with Kaynes Semicon for domestic packaging, and claims up to 50 times better energy efficiency for inference. What unlocked this was the Design Linked Incentive scheme, which reimburses the $200,000 to $500,000 annual bill for EDA tools from Synopsys or Cadence. India already has roughly two lakh semiconductor engineers designing chips inside global capability centres under someone else’s logo. The DLI made stepping out arithmetically possible.

EtherealX was the fund’s first investment, in 2023, at a $2.4 million post-money valuation. In January 2026 it raised a $20.5 million Series A co-led by TDK Ventures and BIG Capital, with Accel and Prosus participating, at an $80.5 million valuation, roughly 4.5 times up for Bluehill.VC. It holds about $130 million in customer contracts. Razor Crest Mk-1 targets 24.8 tonnes to low Earth orbit expendable, around 8 tonnes fully reusable, at $500 to $1,000 per kilogram. The hard part is recovering the second stage.

They’re running out of parking space for these stage ones. If I’m able to recover stage two, it’s like an aeroplane. I get to reuse that vehicle hundreds of times.

SpaceX flew 165 orbital missions in 2025, with one booster now past 36 flights. ISRO averages about six launches a year. EtherealX plans to capture re-entry frictional heat and route it into propulsion rather than carry ablative tiles as deadweight. Hot-fire tests are set for mid-2026, a demonstration flight for late 2027.

Zebu Intelligent Systems, Hyderabad, took $1 million in May 2025, before Operation Sindoor. Thousands of small drones cross the India-Pakistan border each year carrying arms and cash. Shooting them down costs more than the target and yields no intelligence. Zebu’s interceptor locks on ten metres above the rogue drone, fires a net into its propellers and parachutes it down intact, so the army can trace the flight controller, payload and communication band.

That last detail explains the whole sector.

The flight controller is the brain of the drone. So while the drone might be Indian in name, when push comes to shove, it’s no more your drone. It’s a Chinese drone.

After Operation Sindoor, an emergency order for 852 drones was issued and revoked within days once auditors found the flight controllers were not indigenous.

The machine that makes the machine

Asked whether India can copy China’s playbook, Manu disputed the premise.

They’ve been able to copy the chips. They’ve not been able to copy the machine that makes the chips. There’s a lithography machine made by a Dutch company called ASML. You take apart one screw and the entire thing falls apart.

What China did get right was state capital patient enough to underwrite losses until a moat formed. India is now attempting that through SIDBI, RDI, iDEX and ISM 2.0. The engineers were always here. The open question is whether the capital finds the inventors or the integrators, which is a diligence problem rather than a funding one.

Bluehill.VC will deploy ₹80 crore over the next six months and raise its next fund in 2027. For founders reading this in a friendly funding market, Manu offered a caution most investors would not put on a recording.

The most difficult thing a founder will do is take money from an investor. The minute I collect that money, there is a clock running. I would go so far as to say, if you don’t need to raise money, you probably shouldn’t.

Listen now!

Other ways to listen:

Apple Podcast | Amazon Music

Your Feedback matters

As always, I’d love to hear your thoughts! Whether it’s about this episode or ideas you’ve been playing around with, shoot me an email at ad@thepodium.in. Your feedback keeps these conversations going, and I’m always up for chatting about your startup ideas too.

Until next time,

Your Host,

Akshay Datt

The Founder Thesis Podcast | Learn from disruptive founders

Frontier Tech, Not Deep Tech: Manu Iyer on…