INNOPOLIS Global Deep-Tech Commercialization Forum @ UKC 2026 - Panel Q&A
Aug 6, 2026 · 13:00–15:00 · Championsgate Room, Omni Orlando Resort at ChampionsGate
Questions as written in the organizer’s program (Korean). Answers drafted in English. Each answer has a 2-minute version and a 30-second fallback for when the panel runs late.
How to Use This
Speaking order. Session 1 — 2nd of 4 (after Ron Piccolo). Session 2 — 2nd in the individual-question order (after 나르마). Both times you follow someone, which means you can open by building on or contrasting with what was just said. That reads far better than a cold start.
Do not repeat the Gates Foundation credential argument twice. It anchors 1.1 in Session 1. In Session 2, the emphasis shifts deliberately to framing and the outsider story. Same audience, one hour apart — they will notice.
Number discipline. Reserve 90% for the clinical trial failure rate only. It is your strongest single number; don’t spend it on anything else.
If a prior panelist takes your point — don’t abandon it, invert it:
“Ron just made the case for X. Let me push on the other side of it —” “I want to agree with that and then complicate it slightly.”
Session 1 — Policy, Strategies and Founder Insights
13:10–14:10 · 9-min talk + panel (2 min per answer)
1.1 공통 질문
Question
한국의 많은 딥테크 기업들이 우수한 기술력을 가지고 미국 시장의 문을 두드립니다. 기술력 외에 현지 생태계에 성공적으로 안착하기 위해 가장 먼저 준비해야 할 한 가지를 꼽으신다면 무엇인가요?
Answer (~2 min)
If I have to name one thing, it’s a credibility proxy — someone other than you who will vouch for your data.
The US market validates people before it validates technology. Korean companies tend to assume that if the data is good, the argument is won. In the US, the prior question is: who stands behind this data? And that question decides whether the meeting happens at all — which means you can lose before anyone has looked at your technology.
We lived this. When a semiconductor person walks into a biotech room, the first question is not about the measurement. It’s “where is your paper, and who validated it?” So we stopped being our own validator. The Gates Foundation grant was, for us, a credential before it was capital — a review panel that owed us nothing had examined our logic in public. Same reasoning behind building the board with Stanford genomics faculty, and behind the joint research with KRIBB and KAIST.
Here’s the uncomfortable part for engineers: this is not a marketing problem, it’s an architecture problem. You have to design credibility into the company from the beginning, the same way you design the product.
So the first item on a US-entry checklist is not an office and not a lawyer. It is three people on the ground who will speak for you when you are not in the room.
30-second fallback
One thing beyond technology: a third party who will vouch for you. The US market checks the guarantor of the data before it checks the data — so you can lose before anyone has evaluated your technology. Build that into the company from day one, not into the pitch deck afterward.
1.2 개별 질문 — 윤성희 대표
Question
대기업 경험과 바이오 스타트업 창업 경험을 모두 갖고 계신데요, 한국 기술기업이 글로벌 AI/바이오 시장에서 글로벌 수요기업과 빠르게 PoC를 추진할 때 필요한 핵심 노하우는 무엇이라고 생각하시나요?
Answer (~2 min)
I’ve been lucky enough to sit on both sides of this table. At Samsung, Amazon, and SK hynix I was the one receiving PoC proposals. At Erudio Bio I’m the one making them. Let me start with what I learned on the receiving side: the PoCs that died almost never died for technical reasons.
Three things.
First — design the PoC as a rehearsal for a purchase decision, not as a technical validation. Agree on success criteria in their KPI language before you agree on anything else. Not “AUC of 0.92” — but “when this metric moves, your team cuts recalls by X%.” The person across the table is not buying technology. They are buying their own year-end performance review.
Second — put data access on the agenda of the very first meeting. The number one cause of PoC death I saw inside large companies was never underperformance. It was “we still don’t have the data, six months in.” NDA scope, data export rules, on-prem versus cloud — these are not closing items. They are opening items. If you cannot get a straight answer on data in the first meeting, that is your result: you have just learned this PoC will not happen, and you learned it for free.
Third — three six-week PoCs beat one six-month PoC. Deep tech cycles look long, but your counterpart’s budget cycle and internal reorgs move much faster. If your champion changes roles, your project evaporates. So secure a champion and a skeptic at the same time. The written record of having convinced the skeptic is what protects you inside their organization after your champion leaves.
30-second fallback
Three rules. Write success criteria in their KPI language, not yours — they’re buying their performance review, not your technology. Settle data access in meeting one — that, not performance, is what kills PoCs. And run three six-week PoCs instead of one six-month PoC, because their org will reorganize faster than your technology will mature.
1.3 플로어 Q&A — 예상 질문
Question
미국 진출 시 실리콘밸리나 보스턴 외에 플로리다와 같은 신흥 딥테크 거점 지역을 타깃으로 잡을 때의 가장 큰 실질적 장점은 무엇인가요?
Answer
In one word: competitive density.
In Silicon Valley, a Korean deep-tech startup is one of several thousand. In an emerging hub you can be the partner, not one of many. The distance to a decision-maker at the state government, the anchor university, or the anchor corporation is dramatically shorter — and so is your cost structure.
But let me be honest about the trade-off: capital and follow-on talent still sit in the Bay Area and Boston. So the posture I’d recommend is two-track — validation and production in the emerging hub, capital and narrative in Silicon Valley. Regions with real anchors, like the Florida Semiconductor Engine or the High Tech Corridor represented here today, are exactly where that works.
Question
한국 공공 기관이나 특구재단 같은 지원 기관이 미국 현지 액셀러레이터/주정부/대기업과 협력할 때 가장 효과적인 파트너십 형태는 무엇인가요?
Answer
The most effective form is not “bring our companies over.” It’s “make the US side bring their problems to us.”
Most programs today are built so Korean companies fly to the US and pitch. But contracts get signed in the opposite direction: a US demand-side company posts a problem statement, and companies respond to it. That’s a reverse-pitch structure.
The largest piece of leverage an organization like INNOPOLIS holds is not introducing Korean companies. It is sourcing the unsolved problems of US demand-side companies and bringing those home. One well-specified problem statement from a US corporation is worth more to a Korean deep-tech company than a month of scheduled meetings.
AI 관련 질문 대비
If asked: “Is the AI boom a bubble?” / “Where does AI actually create value in deep tech?”
The honest answer has two halves.
The infrastructure buildout is running ahead of the revenue. That’s a real observation and I won’t pretend otherwise.
But the capability curve is not slowing — and for deep tech specifically, the shift that matters isn’t chatbots. It’s that AI has moved from pattern recognition to agency: systems that plan, use tools, and execute multi-step work. For a company like ours, the constraint was never the model. It was the ground truth to train it on. That’s the part nobody can download.
That’s the general lesson for Korean deep tech: in a world where frontier models are commoditized and available to everyone, your defensibility is the data only you can generate. Not the model. The measurement.
Session 2 — Breaking Into the U.S.
14:20–15:00 · 토크콘서트 (2 min per answer)
2.1 공통 질문
Question — Option A
(A) 미국 법인을 세우거나 현지 미팅을 진행하면서, ‘한국과 문화나 일하는 방식이 너무 달라서 당황했던 의외의 순간’이나 ‘웃지 못할 현실적인 에피소드’가 있다면 하나씩 공유해주세요.
Answer — Option A (shorter, lighter — good if the room needs energy)
In one of our first US partner meetings, the other side said, “This is very interesting.” I went back and told my team we had succeeded. I learned later that it is the polite American way of saying “no.”
The reverse trap is worse. The Korean habit of opening with modesty — “we still have a long way to go, but…” — does not read as modesty in the US. It reads as lack of confidence, and it gets you filed accordingly.
Americans say yes loudly and no softly. Koreans do the exact opposite. If you don’t know that, you read every signal backwards — and you will read them backwards for about two years before someone tells you.
Question — Option B
(B) 오늘 모신 기업들은 항공우주, 반도체, 바이오와 같은 딥테크 분야의 기업입니다. 사실 딥테크 분야는 기술 검증 절차가 까다롭고 진입장벽이 매우 높은 보수적인 시장인데요, 미국 딥테크 생태계에 처음 진입할 때 느꼈던 벽은 무엇이었으며, 이를 극복하기 위해 던진 결정적인 전략은 무엇이었나요?
Answer — Option B (recommended, ~2 min)
The wall was very specific: in biotech, “who produced this data” comes before “what does the data say.”
I spent two decades doing AI in semiconductors. In that world, if your data was better, you won — the argument was settled by numbers. Then I walked into biotech rooms and the first question was “which journal, which KOL validated this?” Our data was good and the meetings still did not happen. So the real wall was not the evaluation. It was access to the evaluation.
Two moves broke it open.
We stopped selling the technology and started selling the loss. We used to lead with “our platform generates 1,000× more binding data from a single sample.” Nobody moved. Then we started leading with “90% of drugs fail in clinical trials, and a large share of that traces back to bad early-stage prediction — we attack that at the front end.” Conservative markets do not buy novelty. They buy loss reduction. Same technology, same slide deck, completely different response.
We turned the outsider problem into the story. “Why is a semiconductor person doing biology?” — for a year I treated that as an attack to defend against. Then we started answering it head-on: “because the semiconductor industry faced exactly this problem thirty years ago and solved it. They called it TCAD — you calibrate your simulation against your own measurements, and then you can trust it. Biology has never had that.”
From that point on, the question that had been our single biggest liability became our sharpest asset. In a conservative market, the fastest way through the wall is usually to stop apologizing for why you don’t belong.
30-second fallback
The wall was that access to evaluation came before evaluation — good data didn’t get us meetings. Two things broke it. We stopped selling the technology and started selling the loss we remove — conservative markets buy loss reduction, not novelty. And we stopped defending our outsider status and made it the argument: semiconductors solved this exact problem thirty years ago and called it TCAD.
2.2 개별 질문 — 에루디오바이오
Question
대표님께서는 미국 아마존, SK하이닉스 등에서 경력을 쌓으시며 미국 시장 안에서 수요 기업이 원하는 바를 몸소 느끼셨을 텐데요, 글로벌 진출을 준비하는 스타트업들에게 ‘미국 시장 진출을 위해 이것만큼은 꼭 해라!’ 하고 추천해주실 포인트는 무엇인가요?
Answer (~2 min)
At Amazon and SK hynix I sat on the side that evaluated vendors. Here is the one thing I took away from that seat.
Imagine the performance review of the person who would buy your product.
Large US companies do not buy “good technology.” They buy something that moves my KPI. And that KPI is different for every team — manufacturing and R&D inside the same company want opposite things, and will tell you opposite things in the same week. The most common mistake I see is Korean companies selling to a company. Companies do not buy. People buy. You need to know what that person wants to be able to say to their boss in December.
One more thing from that same seat, and this one is harder to hear. Alongside the technical score, we weighted a question almost as heavily: “will this company still exist in three years?” I watched startups with clearly better technology lose on that question, repeatedly. So a US entity, US customer references, a US investor — these are not vanity. They are survival signals. In deep tech, where the adoption cycle outlives the average startup, they can matter more than the demo.
To compress it: you are not selling to a company, you are selling to a person — and what that person is buying is not your technology. It’s their own results, and their own safety.
30-second fallback
Imagine the performance review of the person who would buy your product. Companies don’t buy — people do, and they buy whatever they can report to their boss in December. And know that alongside the technical score, we always asked “will this company exist in three years?” — I saw better technology lose on that question repeatedly. A US entity and US references aren’t vanity. They’re survival signals.
2.3 플로어 Q&A — 예상 질문
Question
만약 시간을 되돌린다면 또다시 창업이라는 길을 선택하시겠습니까?
Answer
Yes — but earlier.
I spent about fifteen years inside large companies before I left. What I learned in the last few of those was how to move an organization, not how to create a market. Those are different skills, and you can only learn the second one outside.
And one thing I would change: customer validation before technical validation. We spent a year on “does this technology work?” when the question that actually mattered was “who buys this, and at what price?” The engineer in me is still uncomfortable saying that out loud.
Question
험난한 미국 진출 과정에서 ‘아, 이 사업/이 기술은 미국에서도 반드시 되겠다!’라고 확신했던 결정적 순간은 언제였나요?
Answer
There were two, and the smaller one mattered more.
The visible one was the Gates Foundation grant — a review panel that owed us nothing examined our logic and validated it in public.
But the decisive moment came earlier and quieter. A pharma contact looked at our data and asked, “how much would you charge us for this?” They did not compliment the technology. They asked the price.
That is the moment I’d tell any founder to wait for. Praise is free — everyone will tell you your technology is impressive. When someone asks what it costs, they have already started imagining owning it.
Question
우리 딥테크 기업들이 미국 시장에서 마음껏 성장할 수 있도록 혁신 생태계 조성을 위해 정부나 특구재단에 바라는 점이 있다면 한 말씀 부탁드립니다.
Answer
Three things.
Move the success metric from “events attended” to “US customers acquired.” One paid PoC with a US demand-side company is worth more than ten IR demo days. What gets measured gets optimized, and right now we are optimizing for attendance.
Fund presence, not visits. The US market does not open to visitors. One six-month residency beats ten three-day trips. Entity setup, accounting, and legal are small amounts of money that function as decisive barriers for an early company.
Stop treating failure as grounds for clawback. In deep tech, failure is data. If missing a milestone costs you your next round of support, nobody tests a large hypothesis — everyone tests a safe one, and safe hypotheses do not win international markets. I think this is the most structural reason Korean deep tech underperforms abroad relative to how good the underlying technology is.
예상 밖 질문 대비
If asked: “Why should a US company work with a Korean startup instead of an American one?”
Two honest answers. Speed and depth of engineering — Korean deep-tech teams will do a level of hands-on iteration that is genuinely hard to buy in the US at the same cost and cadence. And manufacturing adjacency — Korea sits next to the fabs, the materials supply chain, and the clinical infrastructure.
The thing we have to earn is the third leg: being easy to work with across a twelve-hour time difference. That is a solvable, operational problem, and it is usually the one that actually loses the deal.
If asked: “You founded in the US and in Korea. Which entity should a Korean company start with?”
We run both, so let me be practical rather than ideological. Start with the customer, not the entity. If your first serious buyer is American, a US entity is not optional — it changes their procurement risk assessment, and I saw that scoring from the inside. If your first buyer is Korean or your R&D depends on Korean institutions, incorporating in the US early just adds cost and complexity.
What I would say is: decide it deliberately, not by default. Most companies drift into the answer, and the drift is expensive.