Burkay Gür
Related people
Episodes
Notes
Sourcery · Inside fal’s $1.65B Series E & F Led by Sequoia & Bessemer
-
Has AI Taken Hollywood Jobs Yet?
fal co-founders Burkay Gür and Gorkem Yurtseven say AI has had zero impact on Hollywood labor so far. Hollywood podcasts and the news keep arguing that AI is going to kill Hollywood. The work means talking with people in Hollywood all the time, and the damage people expect is still a projection. Business from Hollywood has been increasing over the last six months and has reached a meaningful share of fal's revenue. Almost no worthwhile Hollywood content has been made with it. The change is coming and there are signs of it, but the warning has circulated for four years and has been fear with nothing real behind it.
Adoption in Los Angeles has also been slow for a regulatory reason. Lawyers a
-
How Did Fal Get Its Name?
fal co-founders Burkay Gür and Gorkem Yurtseven say fal is the initials of features and labels, two old machine-learning terms from before the current AI wave. The short name is pronounced fal.
The name is also a Turkish coffee-house word. After the coffee is finished, someone reads a fortune from the grounds left in the cup. In Turkey, everyone has done that reading. Molly O'Shea had tried Turkish coffee and had not had a fortune read from the grounds. The founders offered to do a reading. The spelling holds both ideas at once: the machine-learning initials, and the coffee-ground prediction.
-
Why Can't a New AI Model Buy Its Own Inference Compute?
fal co-founders Burkay Gür and Gorkem Yurtseven say a lab with a new model has to plan inference compute before the launch. fal sits between model providers and the studios and application teams that consume the models, so it sees supply and demand on both sides. Compute is scarce enough that a successful model often cannot be served on hardware the lab buys alone. Partnering with an inference provider for the launch is the practical path. fal works closely with model providers. Customers usually run several models at once and time their own product releases to model releases.
Someone has to reserve capacity ahead of time. Otherwise the compute goes to coding agents or to whoever is spendin
-
Why Does Fal Send Engineers Home on Weekends?
fal co-founders Burkay Gür and Gorkem Yurtseven say fal looks for agency, hunger, and humility in engineers. Agency means a person can act independently, take responsibility, and own decisions. Once that is there, the other traits follow. Customers and models change so fast that fal often wants engineers to make business decisions or work with customers and decide around them. At a large company such as Amazon or Coinbase, software engineers stick to software engineering. Agency is hard to test, and the best signal is having already worked with the person.
Hunger is the close cousin: people have to want to win and build a large company. A conversation plus a reverse sell surfaces it. fal te
-
Why Did Coinbase Make Candidates Pass a Crypto Quiz?
Burkay Gür says Coinbase made candidates pass a crypto quiz posted on Reddit before anyone on the team would take a phone call. He joined in 2016, when Coinbase was almost 100 people. The quiz tested how much a person knew about crypto, including technical terms, and how deep that interest went. It ran before AI, so the answers were hard to find. The quiz was a key factor in Coinbase's long-term success, because Coinbase wanted fanatics who would keep the work going.
fal is about 150 people today, close to Coinbase's size in 2016, and the same bar still decides early hires. Most of the job is staying current. A team can hire strong people and hope they feel motivated, or it can hire people
-
How Did Fal Move From Data Infrastructure to Image Models?
fal co-founders Burkay Gür and Gorkem Yurtseven say fal started in 2021 as a compute company, before the current AI wave. The thesis was that the next data-infrastructure company had to own compute, the way Snowflake and Databricks let SQL scale in the cloud. fal bet that Python would be the next layer and worked with teams to move machine-learning jobs into the cloud.
ChatGPT, Stable Diffusion, and the early image models then forced a choice between staying in data infrastructure and entering AI. Open-source language models drew a price race toward the bottom. With five or six people and little money in the bank, selling tokens at a negative margin was off the table. Image models were wher
-
Who Pays $5 for a Five-second AI Video?
fal co-founders Burkay Gür and Gorkem Yurtseven say a five-second video priced at five dollars still found essentially infinite demand. Cheaper five-second clips, including from Kling and Minimax, had started around 40 cents and settled near 30 cents. The five-dollar price was a jump of more than ten times, and the first reaction was to ask who would pay that for another five-second clip.
People paid it, and later models that cost even more also got used. Professional work showed up at the same time, in ads and in movie production. Marketers started using the higher-quality models because cheaper ones could not capture the demand. Usage rose, capabilities rose, and prices rose together.
La
-
Why Did Fal's Image Business Come Back?
fal co-founders Burkay Gür and Gorkem Yurtseven say about half of fal's revenue now comes from video models and half from image models. The move from images to video was fast. Video revenue started at zero and reached about 70 percent, and video looked ready to take over completely. Image models looked finished, and everything ahead looked like video.
Image editing models brought image work back. Edited frames also help video models, which interpolate between frames more effectively. Image revenue that looked like it was dying made a comeback, and the split settled at half and half.
Video is still early. The best model makes about 10 to 12 seconds, and newer ones may reach about 30 seconds
-
Why Is Generative Media One of Only Three Product-market Fits in AI?
fal co-founders Burkay Gür and Gorkem Yurtseven say AI has three product-market fits: search, coding agents, and generative media. Search is already so large that it gets a pass. Coding agents are the other clear case, and they are growing fast. Generative media is the third, and the only other place where one person can spend $10,000 or $20,000 a month.
An individual on a coding product can spend that much on a cloud or OpenAI subscription. The same bill shows up when generating video is the job, at an AI-native studio or a traditional studio. Someone sits at a computer and generates video all day. No other part of the industry has that individual-spend pattern. Demand turned out to be ess
-
Why Does Fal Have No Pure Managers?
fal co-founders Burkay Gür and Gorkem Yurtseven say fal has no pure managers and does not plan to add them. fal is flatter than a typical startup, and the deepest reporting line is two levels, by design. The management layer is what they call super individual contributors: people strong enough at the job to do hands-on work and also lead teams. fal also has four forward deployed engineers.
Teams form around people in the group who are unusually good. Strong hires come in, leaders emerge, and teams gather around them. Leadership has been built that way from the start, even before the super individual contributor name was in use.
The shape has a limit. Communication bottlenecks show up while