Why I left
a future at EVGA.

As an autistic savant, I developed an early, consuming interest in computers and the silicon that made them work. By about fifteen, I had my first EVGA graphics card: a beautiful black 8800 GTS with 320 MB of memory and that unmistakable black shroud. It is still one of my favorite EVGA designs.

I was completely obsessed with it. It solidified my love of NVIDIA graphics hardware at the beginning of the unified-shader and CUDA era. There was already so much these chips could do, and the possibilities seemed enormous.

When EVGA accepted me years later, I felt as though I had won the lottery. I was going to work at a company I had spent years obsessing over, helping the people who made some of my favorite graphics cards in the world.

EVGA was one of North America’s leading NVIDIA partners. To me, it was much more personal than that. It was the name on the card I had loved as a teenager.

My understanding of the company’s early history was that the E originally stood for e-commerce. EVGA began as an online retailer before developing its own versions of high-end PC hardware. In later years, the E came to be associated with extreme, which suited the enthusiast company it had become.

Before I joined, I had the wonderfully naïve idea that I would spend all day building incredible gaming computers. Apparently, I thought they were just going to hand that job to some kid straight out of college.

What they actually handed me was a desk phone and a headset.

Historical context: NVIDIA’s GeForce 8800 architecture brief (November 2006). The earlier e-commerce meaning of EVGA’s name is my understanding of its history.

I eventually took around 10,000 technical-support calls and worked my way from a temp to a supervisor. I was told it was the fastest anyone had made that progression. Some supervisors had departed quickly, and I skipped an entire level.

That sounds much tidier than it felt.

When I started, I was afraid of talking on the phone. I avoided it in everyday life, too—something I now understand partly through the lens of autism. I had grown up without the support and medication I believe I needed. At that point, I could only hunt and peck at a keyboard with one finger from each hand. Keeping up was incredibly difficult.

The first few months were brutal. Training, repetition, hard work, and probably my love for the company carried me through. I slowly added more fingers to my typing until I could use all of them. I became more comfortable on the phone. Eventually, I became someone other people could depend on.

Our team mattered enormously to me. There were so many talented, nerdy guys who wanted to work with graphics cards. I felt at home among people whose intensity about computers matched my own. I loved being one of the guys.

We offered technical support by phone twenty-four hours a day, seven days a week. Agents worked through the night, and I helped manage the team to make sure at least two people were always on the phones. The aim was to keep customers from waiting for help.

As I remember it, our average customer hold time stayed well under a minute throughout my years there. I was not aware of another graphics-card board partner offering that kind of round-the-clock phone support. EVGA’s reputation for customer service meant something very practical to us: someone had to be there to answer.

On a difficult day, it could feel like us versus the customers. But we helped a lot of people, and the work taught me a lesson I still use: bring the conversation back to the problem.

A customer might arrive furious, exhausted, or already convinced nothing would work. Sometimes they had been through the wringer before reaching us. Joining their panic did not help. Neither did following every tangent.

What helped was returning calmly to the technical issue. What was the graphics card doing? When did it happen? What could we test next?

Often, the customer would begin to calm down as soon as they believed we might actually solve it. I learned a great deal about emotional resilience and where to put my energy. The problem was something we could work on together.

Before I left, EVGA had ended its twenty-four-hour phone coverage. I cannot remember exactly when, but I remember it being a significant change. It had been one of the things that set us apart.

EVGA’s graphics-card manual lists 24/7 technical phone support with a US toll-free number. The staffing, hold times, and recollection of that coverage ending are from my own time on the team and at the company.

I met Joseph on that support team. Almost at random, the two of us ended up making videos for EVGA’s YouTube channel. Joe Darwin came in and asked whether anyone had YouTube experience. We both perked up and said we did.

It became clear almost immediately that we had a tremendous amount to learn.

That first video was a horror show: stiff, strange, limited in scope, and somehow nearly six months in the making. Joseph was not yet especially comfortable on camera. Someone left a comment that, as I remember it, said, “Blink twice if you’re being held against your will.”

We still laugh about that. We just did.

Over time, Joseph became a very talented editor—better than I was—and it was clear he had a real passion for it. I found that I liked being in front of the camera. I could talk about computers all day, and I enjoyed finding a way to make other people care about them.

Once we got through the initial friction, it became a marriage made in heaven. We developed a smooth production routine and a permanent set that Andrew Han liked. We could produce content at the drop of a hat, in between the other jobs we were already doing.

That partnership continued when I moved from support supervisor to product marketer and went upstairs, closer to the media room, sales, and Andrew’s office. Video production had become a real part of what I could contribute.

EVGA’s media room was tucked away at the back of the upstairs floor, overlooking the warehouse. It was far enough from the daily traffic of the office that we could build elaborate sets and trust they would still be there, untouched, when we came back. That quiet corner gave us room to do some fairly ambitious things.

Two livestreams in particular still stand out.

Twelve hours for St. Jude.

One was a twelve-hour charity livestream for St. Jude Children’s Research Hospital. I was responsible for the recording and production setup. I also took turns in front of the camera and joined in the activities throughout the day. We had several filming stations to switch between, so there was a lot to keep running while the show was live.

Viewers could donate money to put the people on camera through increasingly ridiculous forfeits. As the donations came in, things got a little out of hand. By the end of it, we were eating full spoonfuls of some of the hottest chili sauces I had ever encountered.

I already knew my former customer-service supervisor, Jaeson Wilda, was a champion hot-sauce eater. His cubicle was decorated with ribbons and some impressively large trophies from the hot-sauce contests he had competed in. He was also a wise coworker and an excellent boss, and he had kindly brought in some of his hottest chilies for the occasion.

We raised money for charity, the recording setup held together through all the station changes, and I finished the day completely exhausted. I was also very proud. I had helped represent the company, put on a complicated live production, and contributed to something worthwhile.

Learning from K|NGP|N.

Then there was the stream with Steve Burke from GamersNexus and Jay from JayzTwoCents. EVGA’s own Vince Lucido—K|NGP|N—was showing them how to push RTX graphics cards under liquid nitrogen. Liquid-nitrogen overclocking is a specialty of its own, and what I remember most is Jay and Steve watching Vince work and trying the techniques for themselves. This time, I was responsible for the filming and stayed behind the camera.

I also remember Vince’s collaborator, Illya “TiN” Tsemenko, being around that day. TiN helped design the specialized hardware that made their record attempts possible, including the circuitry of the RTX 2080 Ti K|NGP|N. To me, he and Vince represented the art of extreme overclocking at its highest level.

Vince was part of a worldwide community of overclockers who traded places at the top of the 3DMark leaderboards. I remember him holding the top spot in Port Royal at the time. He and TiN understood how to coax a little more out of silicon that everyone else might have considered finished. Getting to see that expertise up close was a privilege.

We ordered a very large tank of liquid nitrogen from a local supplier and timed its delivery close to the start of filming. It arrived just before we were due to go live.

As the tank was wheeled in, it rolled over some of my video cables. Its weight internally severed the wires.

We had minutes before the livestream was scheduled to start automatically. Suddenly, I was scrambling to run replacement cables and get the feeds back. It was an extremely hectic start.

Once we were running, I could appreciate what was happening in front of the cameras. Extreme overclocking involves finding a chip that responds well to the combination of voltage, temperature, and tuning. Liquid nitrogen let them explore limits far beyond everyday cooling.

The cold brought its own problems. As the cooling pot and nearby hardware fell below the dew point of the room air, moisture could condense or freeze on their surfaces. The circuit boards could look as though they were sweating. Frost might look spectacular on camera, but as it thawed, the water could short exposed circuitry.

That was where the putty, Vaseline, and shop towels came in. Moldable eraser putty and petroleum jelly helped protect vulnerable areas, while towels caught moisture. Preparing the boards was only the beginning. Throughout the session, temperatures, frost, and wet spots needed constant attention. Keeping everything running smoothly was a skill in itself.

A small gas torch was on the bench, too. Some chips will stop running or refuse to restart when they get too cold—problems known as a cold bug and a cold-boot bug. Overclockers use the torch to warm the metal cooling pot back up to a workable temperature. Keeping a system running under liquid nitrogen can mean carefully adding heat as well as taking it away.

EVGA designed hardware for exactly this kind of work. Its DARK motherboards paired complex multilayer circuit boards with substantial power delivery. The X299 DARK, for example, used a twelve-layer PCB and two eight-pin CPU power inputs alongside the main twenty-four-pin connector. Those features supported electrical stability under demanding overclocks; the insulation and moisture control were still essential.

I remember multiple 1,600-watt, Titanium-rated power supplies around the setups, with separate power for the graphics hardware and the CPU and motherboard. The scale of it was something to see. This was brilliant, wonderfully excessive engineering. I worry we may never see another era of EVGA quite like it, and I don’t want it to be forgotten.

I loved getting to spend time with Vince, Jay, and Steve. I respect all three immensely, and it was a pleasure to film people who loved hardware as much as I did.

I loved that day in the office. Even with the cables.

Watch the recordings: EVGA’s St. Jude fundraiser highlights, including Jaeson and the ghost peppers, GamersNexus’s archived overclocking livestream with K|NGP|N, Jay, and Steve and its recap of the overclocking match and charity fundraiser. The production details and behind-the-scenes memories are my own.

Technical context: EVGA’s RTX 2080 Ti K|NGP|N design credits (archived April 2019), EVGA engineer TiN’s X299 DARK overclocking guide and Burning OC’s guide to insulation, temperature control, and extreme-overclocking equipment.

I joined in February 2016, a few months before the GTX 1080 launch. That era still feels special to me.

I could help with the launch, go home and fire up the new Doom in those stunning id Tech 6 visuals, then come back the next day and talk about it with coworkers who had been doing exactly the same thing. It made me feel like a kid again.

It gave me belonging, and some of what I had missed in a misdiagnosed youth.

I also got early, extended hands-on time with Oculus during that first big consumer-VR push. Remember when HTC had a headset? The original Vive looked like a diver’s helmet covered in tiny portholes. For someone like me, getting to spend time with all of this was extraordinary.

As my role grew, I worked closely with Joe Darwin and Andrew Han. I represented EVGA at distribution events on their behalf. I think they trusted my ability to explain the products and put on a good show.

At some of those summits, we received wonderful feedback about our presentations and our presence compared with other brands. I was proud of how we showed up, and of what I contributed.

At an ASI summit at Pacific Palms Resort in City of Industry, a woman came up after my presentation and said, “I could listen to you talk about computers all day!”

It made my day. A few years earlier, I had been frightened of answering the phone. Now I was traveling, speaking for a company I loved, and finding out that I could do it well.

Launch timing: NVIDIA’s May 6, 2016 announcement gives May 27, 2016 as the GTX 1080’s retail launch date. My February 2016 start at EVGA is from my own account.

Some of my concerns began while I was still on the support team. When NVIDIA introduced the Pascal Founders Edition cards in 2016, I remember people at EVGA worrying about what that meant for us. We were used to selling NVIDIA’s reference designs with our name on them. NVIDIA had sold its own cards before, but this was when the tension became clear to me: the company supplying our GPUs was also making a prominent pitch to sell finished cards directly to our customers.

At that launch, the GTX 1080 Founders Edition carried a premium over the announced starting price for partner cards. The concern was the relationship itself. We depended on NVIDIA while competing with it for the same buyers.

By the time I was giving speeches for EVGA, the first RTX generation was arriving. I helped with marketing for the RTX 20-series launch; some of my work from those years is preserved in my EVGA 3D and motion-graphics archive.

Preparing for those launches could be difficult in ways customers never saw. We would receive the GTX or RTX specifications at the same moment the public did. NVIDIA was fiercely protective of its announcements, and I remember how angry it could become with partners over information that leaked before an official reveal.

Jacob Freeman came up with the idea of sending each online retailer a set of product images with a different colored dot in a corner. If an image leaked and NVIDIA came back to us, we could identify which retailer had originally received that image set.

I believed in the technology. But believing in a direction and persuading someone to pay for it today are different challenges.

RTX asked gamers to pay more for hardware whose defining new feature was still waiting for widespread game support. The price increase varied by model: NVIDIA’s announced starting price moved from $599 for the GTX 1080 to $699 for the RTX 2080, while the corresponding Founders Editions moved from $699 to $799.

There was another part of that price that made the situation harder for EVGA. As I understood the arrangement, we were required to offer a model at NVIDIA’s advertised starting price. If we did not, NVIDIA could reduce our allocation of GPUs, and it could be difficult to tell whether a reduction was a response to our pricing. We needed to keep a loss leader—a card we sold at a loss—to keep NVIDIA happy and protect our access to the chips.

The models on which we did not lose money had better circuit boards and cooling of our own, but they also carried a substantial premium. To a customer comparing them with the advertised starting price, it could look as though EVGA was charging an enormous amount just for those additions. The loss leader set the expectation; the cards with room for us to make money had to justify the gap.

That was becoming more difficult. Over successive generations, I had watched NVIDIA refine its reference board designs and the way its GPUs managed their clock speeds. Technologies such as GPU Boost automatically adjusted clocks within power and temperature limits. From where I sat, extra spending on custom circuitry and cooling was no longer buying the kind of everyday performance advantage that overclocking had once made possible.

The better hardware still took work to design and build. But we were asking customers to pay a significant premium for our choice of components when the improvement they could expect in their games was often marginal. It was an increasingly difficult proposition to sell.

I remember the loss leaders being sold out almost continuously. The cards closest to the price customers had been promised were the hardest to find, leaving the more expensive models to carry the sale. That did not help us move those cards through to customers, either.

Many customers could skip a generation and buy another NVIDIA card later, when the value made sense to them. For a manufacturer like EVGA, those decisions carried very different consequences. We still had cards to sell and a business to sustain; enough customers sitting out a generation could be financially devastating.

For many of the customers we were trying to reach, the immediate benefit was a much harder story to tell than it had been with Pascal. The hardware arrived ahead of much of the software. Even Battlefield V’s ray-tracing update did not arrive until November 2018, after the September RTX launch.

Then, when ray tracing did arrive, turning it on could impose a substantial performance cost. To me, the first generation struggled to deliver the experience the promise had suggested. The RTX 30 series would later bring second-generation RT cores and a substantial improvement, but that did not make the original sales pitch easier in 2018.

NVIDIA could afford a long technological bet. My concern was whether a company as lean as EVGA could comfortably absorb the difficult part of it.

My memory of the GTX 10-series launch is of a frenzy: supply constraints, allocation headaches, and customers desperate to get cards. My memory of the RTX 20-series launch is very different. Away from those frequently sold-out loss leaders, cards were easier to obtain, and overall demand felt much softer than expected.

The fall in mining demand left another problem behind. I remember the 3 GB GTX 1060 in particular becoming a major inventory problem. The stock was aging and very difficult to move. The challenge included both selling the new generation and clearing the old one.

At the same time, I remember our Intel X299 enthusiast motherboards becoming difficult to move as AMD became more competitive. From where I sat, we were feeling pressure in both graphics cards and motherboards, and power supplies could not simply make up for everything.

Launch dates and prices: NVIDIA’s GTX 1080 announcement (May 6, 2016); RTX 20-series announcement (August 20, 2018), with availability beginning September 20, 2018; Battlefield V ray-tracing update (November 14, 2018); and RTX 30-series announcement.

Technical context: NVIDIA’s explanations of the Pascal Founders Edition, sold directly and through partners, GPU Boost and Pascal’s clock controls, and the RTX Founders Edition board, power delivery, and cooling. NVIDIA’s own retail cards predated Pascal; its GTX 460 retail manual dates to 2010. The launch concerns, specification timing, marked retailer images, allocation pressure, loss leaders, stock availability, aging 3 GB GTX 1060 inventory after mining demand fell, and difficulty justifying EVGA’s premiums are my recollections and assessment from inside the company.

A major layoff followed. I hope you never have to experience a really bad one from the inside.

I talk on this website about the importance of automating work. I also know what it looks like when people lose their jobs, and how far that reaches into the lives of their families and loved ones.

From my perspective, the company had not prepared adequately for the downturn. That failure was unacceptable to me when so many people depended on the decisions being made at the top. It permanently changed how I think about leadership and work culture.

Those of us who remained had to wear more hats and fill roles left by people who were gone. I became one of the company’s sales managers, largely because I was still there—not because I had arrived with a deep background in sales.

There was no longer time for the video work Joseph and I had built together. He had rightly moved on to start a family. A part of the job I loved had disappeared.

Even then, I did not immediately decide to leave. I thought we might get through it. The company had taken a hit, but we had survived, and I still had my position. Perhaps there was something we could continue to build?

There were parts of sales I was beginning to enjoy. I liked being an evangelist for EVGA, and I thought I represented it well. Traveling with senior sales colleagues to Chicago and other distribution shows gave me a broader view of the PC industry—and a welcome chance to get out of the office.

But the numbers kept needling at the back of my mind.

The business increasingly stopped making sense to me. Graphics cards defined EVGA’s identity and drove an enormous share of its revenue, yet I understood power supplies to be doing much of the work of sustaining profitability.

Years later, reporting on the NVIDIA split put graphics cards at roughly 78% of EVGA’s revenue and described much healthier margins on power supplies. Those were figures reported in 2022, after I left. They echoed the imbalance that had worried me during my time there.

JayzTwoCents’s September 2022 account made the later pressure unusually concrete. He said EVGA had told him it was losing money on a $1,399 RTX 3090 Ti FTW3 while NVIDIA’s Founders Edition was selling for $1,099. Jay described losses across the board, but Steve Burke’s reporting was more specific: lower-tier cards such as the RTX 3060 were still profitable, while EVGA was losing hundreds of dollars per card from the RTX 3080 through the RTX 3090 Ti. Both were discussing the market nearly three years after I left. Those reported losses on premium cards went beyond the loss leaders I remember from my own time there.

As I understood it, we depended on NVIDIA for the core components of our cards while operating within pricing constraints that could make a profitable product very difficult to deliver. Later public reporting described shrinking partner margins and NVIDIA’s own cards undercutting its board partners.

I cannot tell you whether NVIDIA intended to squeeze partners out. I can tell you that, from my position, the room available to EVGA looked increasingly narrow.

Andrew Han struck me as brilliant and short-tempered. I was short-tempered at that time, too. We were also both serious audiophiles.

He had an extraordinary collection of vintage audio equipment in Taiwan, and I knew of him showing it to Steve and other visitors at his office and home. I never saw the Taiwan equipment myself. What I did get to hear was some of the beautiful audio equipment he would occasionally bring into our office.

That passion also found its way into the EVGA NU Audio sound card, developed with Audio Note (UK) and released while I was on the marketing team. I worked on its product page, too. To my ears, it was one of the best-sounding PC sound cards ever released. We would do A/B listening comparisons against competitors to demonstrate the difference. I thought it was an extraordinary product, and helping people hear what made it special was work I could believe in.

I considered Andrew a true golden ear. And I think it takes one to know one.

One idea he articulated has stayed with me: he wanted our pricing sheet and the logic behind it to come down to two buttons. One would run promotions aimed at maximizing sell-through and revenue; the other would prioritize profitability.

We were adjusting graphics-card promotions for numerous retailers every month, trying to respond to a market that could change quickly. At the time, I did not think software could make those decisions well enough to reduce the process to something so simple. What I have learned about AI since has changed my view. I can now see how machine learning could help predict demand and evaluate promotions against those different goals. Looking back, I think Andrew was ahead of the curve in imagining what that system could do.

I could still see the relationship with NVIDIA becoming a larger and larger point of conflict. I also believed Andrew saw a bigger role for me in the company.

His manner when I left felt characteristically rude to me. My interpretation was that he was frustrated I was jumping ship when he wanted me to stay and take on more. I cannot know exactly what he was thinking.

When I walked out, I remember predicting that they had four or five years before this became a major confrontation—another generation or two of graphics cards. I did not see how Andrew could win that confrontation on the terms available to him.

Less than three years after my departure, EVGA’s break with NVIDIA became public on September 16, 2022—just four days before NVIDIA announced the GeForce RTX 40 series in a September 20 broadcast at its GPU Technology Conference.

EVGA presented the split as a decision made on principle, rooted in what Andrew saw as a lack of respect from NVIDIA. That was very much the Andrew I knew. Respect mattered enormously to him: earning it, keeping it, and expecting it in return.

Steve’s report came from a private meeting with Andrew and Joe. I recognized the approach. I had seen them disclose information directly to media they trusted when they wanted their account heard accurately, without it being distorted along the way. In my experience, they trusted Steve and Jay enormously. Steve still questioned the company’s reasoning.

Jensen Huang is now one of the world’s wealthiest people, and NVIDIA has achieved extraordinary success. It became the world’s most valuable public company in June 2024. But Jensen and Andrew on the way up often reminded me of each other.

I remember Jensen getting a Ferrari 458, then Andrew getting an Audi R8. From where I stood, it felt like one-upmanship. I found that side of the relationship childish.

After the split, Jensen publicly called Andrew “a great friend.” I don’t know what happened behind closed doors, or where their relationship stands today. I’m sad the partnership ended the way it did. I wish both men had handled it with more maturity, and I expected more from Jensen in particular, given NVIDIA’s power in the relationship.

Jensen, with all the success NVIDIA has had since then, do you see what happened any differently? Is there anything you wish you had handled with more generosity, humility, or maturity? I would like to hear your answer.

I do not hear people talk about Andrew Han very much anymore.

After Jacob Freeman joined NVIDIA, I reconnected with him and tried several times to see whether he could help me find a role there. It never came together. I still wanted to work there, but getting in was extraordinarily competitive.

I was 26 years old when I left EVGA. I could no longer believe in the future I was being asked to help build.

That does not erase what EVGA gave me. It gave me friends, confidence, a creative partnership, and a place to belong. It took someone afraid of answering the phone and helped turn him into someone who could stand on a stage and speak for the company. I worked alongside some of the most talented people I have ever met.

Jay also spoke warmly about EVGA’s customer service and its willingness to stand behind its products. I was glad to hear that part remembered. I knew some of the people who had earned that reputation, one phone call at a time.

It is hard to overstate what Andrew accomplished. He built a brand around commodity PC hardware and gave it a distinct place in the enthusiast world. EVGA brought together exceptional hardware, an extraordinary online community, and customer service that made people loyal to the name. His standards reached from the products to the people and the way the office was run. He expected quality throughout the company, and that expectation became part of what EVGA meant.

To me, he was competing in a market full of equals, and the company he built still stood above.

In many ways, it remains the most rewarding work experience of my life. That is why I wanted the whole story here: the love, the work, the people, and the reason I left.

— Brendon Ray Hedrick

Follow me on LinkedIn ↗ for future writing and new additions to the archive.

My employment dates (February 2016–December 2019), my age at departure, and my expectations for the company are from my own account. Public dates: GamersNexus’s report of the EVGA–NVIDIA split (September 16, 2022); NVIDIA’s RTX 40-series announcement at GTC (September 20, 2022); and Reuters on NVIDIA becoming the world’s most valuable company (June 18, 2024).

The 2022 business context: PCWorld’s September 16, 2022 account of EVGA’s revenue mix and margins; Jon Peddie’s reporting on EVGA’s decision; and JayzTwoCents’s September 16, 2022 account, including his pricing and loss discussion and his assessment of EVGA’s service.

Steve’s September 16, 2022 report records EVGA’s explanation of the decision as a matter of principle and respect and distinguishes profitable lower-tier cards from losses on higher-end models. My account of Andrew’s emphasis on respect and his and Joe’s relationships with the media comes from my own time at EVGA.

Jensen’s public remarks: PC Gamer’s September 22, 2022 account of its question to Jensen Huang about EVGA, including his description of Andrew as “a great friend.” Wealth and company context: Forbes’s profile of Jensen Huang. The Ferrari and Audi sequence is my recollection; the impression of one-upmanship is my own, and I do not know their private motives or the present state of their relationship.

Later career context: NVIDIA identifies Jacob Freeman as its GeForce Evangelist in July 2023.

Audio context: Jon Peddie’s firsthand account of Andrew’s vintage audio collection and EVGA’s NU Audio manual, including its replaceable op-amps. The office listening sessions, A/B demonstrations, and judgments about sound quality are my own recollections and listening impressions.