The Man Behind DeepSeek (Liang Wenfeng)

Virgin of deep seek now putting pressure on the US tech names it's super impressive I think we should take the development out of China very very seriously deep seek just taught us that the answer is less than people thought you don't need as much cash as we once thought can you imagine how annoying it is to see you build your house for $50 million and a guy next door builds the same house for $700,000 that's got to be frustrating be a wake up call for our industries that we need to be laser focused that necessity

is the mother of invention in December 2024 a small Chinese startup shocked the AI world with a breakthrough that would challenge Tech Giants and redefine AI deep seek a relatively unknown player quietly made waves with their V3 model trained on just 2,000 low-end Nvidia h800 gpus it outperformed many top models in coding logical reasoning and Mathematics this achievement sent shockwaves through Silicon Valley prompting industry leaders to re-evaluate their approach to AI development behind this feet was leang Gwen Fung a mysterious figure whose background would soon Captivate the tech World leang Wen Fung was born in

1985 in Jan Xiang a coastal city in China's guandong Province raised in a modest household by his father a primary school teacher leang showed an early talent for mathematics while other kids played games or Sports he spent hours solving puzzles and equations finding joy in untangling their secrets this love for numbers would shape his entire career by his teenage years liang's problem solving skills stood out he had a knack for breaking complicated challenges into smaller manageable steps a skill that later helped him tackle Real World Tech and finance problems at 17 his dedication earned him

a spot at xang University one of China's top schools there he studied electronic information engineering blending his math skills with Hands-On technology applications during college leang dove into subjects like data analysis and computer systems he became fascinated by how math could explain financial markets studying tools like probability based models and algorithms to predict Trends his professors noticed his talent giving him Advanced projects and research opportunities by his final year he focused on algorithmic trading using computer programs to make fast math driven stock market decisions this work later became the foundation for his career around this

time leang faced a major Choice Wang Tao founder of the drone company DJI invited him to join as a partner though the offer promised wealth leang turned it down he believed AI would transform Industries far beyond drones instead of joining DJI he chose to start his own company aiming to Pioneer AI driven Solutions the signs were everywhere but now it's official we are in a recession the research group that makes that determination made it today and said the recession actually started a year ago when the 2008 financial crisis rocked Global markets leang Wen Fung then

a graduate student at Jang University saw a chance to put his skills to the test as Banks collapsed and economies wobbled most people panicked bang armed with his math expertise focused on solving the case OS he gathered a team of classmates to explore machine learning a type of AI that learns from data their goal to build computer programs that could analyze markets faster and smarter than humans this approach called quantitative trading relied on math models to spot patterns in stock prices economic reports and Global Trends the team started by collecting mountains of data stock prices

unemployment rates even news headlines they fed this information into experimental algorithms tweaking them to predict Market swings during the crisis it wasn't easy early models failed constantly especially as the economy kept shifting what if we adjust for investor Panic leang suggested during late night coding sessions his team persisted refining their programs to account for unpredictable human behavior slowly their algorithms began identifying hidden trends like how falling housing prices affected tech stocks that tradition IAL Traders missed months of trial and error paid off while their system wasn't perfect it started making accurate predictions in the volatile

market for example it flagged when certain stocks were about to rebound giving the team small but meaningful wins these successes Drew attention at Jang University where professors saw liang's work as proof that AI could reshape Finance by 2009 Wall Street and other Financial hubs were embracing quantitative trading just as leang had anticipated his project became a case study in Innovation during crisis a blend of math technology and sheer persistence though he' turned down a job at DJI years earlier this work cemented his belief that AI wasn't just the future of Finance but of nearly every

industry in 2013 liangwen fun took his first step into professional trading by co-founding hung XO jacobe investment management with his college friend Shu Jin here leang tested his AI driven trading strategies in real markets learning how to adapt algorithms to unpredictable conditions 2 years later the pair launched hungo high flyer technology focusing on blending advanced math and AI to create smarter trading systems their timing was perfect China's financial markets were expanding offering new opportunities for Tech driven firms like theirs high fly fire made waves in 2016 by releasing its first AI trading model unlike traditional

methods this system used deep learning a type of AI that improves by analyzing vast amounts of data to decide when to buy or sell stocks the results were impressive during a volatile Market period in early 2017 high-flyers AI trading system maintained consistent profits while competing firms experienced losses growth exploded by late 2016 high flyer managed over 1 billion yuan about $140 million outpacing older Rivals a key moment came in 2015 when the company launched 10 investment products in a single day supercharging its ability to raise funds liang's focus on constant Innovation kept their algorithm sharp

blending new AI breakthroughs with real-time Market data by 2019 high flyer ranked among China's big four quantitative trading firms this success proved that homegrown companies could compete globally using cuttingedge tech for leang it was just the beginning a stepping stone toward his larger vision of AI transforming Industries far beyond Finance as high flyer grew liangwen Fung faced a critical challenge the company needed massive computing power to keep its AI trading systems ahead of the competition in 2019 he bet big spending 200 million Yan about $28 million to build Firefly number one a supercharged AI training

system equipped with 1,100 specialized graphics cards Firefly number one could crunch financial data at lightning speed helping high-flyers AI make smarter faster trades but leang didn't stop there in 2021 he doubled down with firefly number two investing a jaw-dropping one billion yuan $140 million this upgrade packed 10,000 of nvidia's top tier a100 gpus a move that shocked the industry to put its power in perspective Firefly number two could handle as many calculations as 100,000 high-end laptops working together few companies in China let alone a trading firm had ever built something dis Advanced Firefly number two

wasn't just powerful it was aici it slashed energy use by 40% and costs by half compared to older systems how smarter cooling methods energy saving designs and custom parts that sped up data flow between gpus these tweaks let High Flyer train bigger AI models without burning through cash or electricity though built for stock market predictions the Firefly system soon became key to High Flyers bigger Ambitions leang saw their potential to tackle AI challenges far beyond Finance from Health Care to climate modeling in May 2023 liangan Fung took his biggest risk yet pivoting from Finance to

pursue General artificial intelligence AGI AI that can outperform humans at most tasks from writing code to diagnosing diseases While most AI tools focus on narrow jobs like chat Bots or image generators AI aims to think and adapt like a Human by July 2023 leang launched deep seek deep seek deep deep seek deep seek deep seek deep a startup with a bold Mission create human level AI this put him in direct competition with China's Tech Giants all racing to dominant AI but leang had a plan instead of chasing quick profits he bet on young Talent hiring

fresh graduates from top universities raw smarts speat experience here deep seek focused on their work instead of seeking media attention the team avoided publicity to focus on long-term research but being small had perks we're like a speedboat one engineer said big companies are oil tankers powerful but slow to turn deep seek had two advantages Firefly supercomputers from liang's Finance days providing massive computing power for training AI open-source ideals sharing tools to collaborate with researchers worldwide in May 2024 deep seek dropped a bombshell deep seek V2 an AI model that matched giants like GPT 4 Turbo

but cost 170th the price just one Yuan per million words processed this wasn't just cheaper it reshaped the rules of AI here's how it worked deep seek V2 combined two breakthroughs the new multi-head latent attention helped to process information much faster while using less computing power this was an important achievement since it let the model perform well without needing as many resources something AI researchers had been trying to accomplish for a long time deep seek saves money by using a method called mixture of experts when someone asks a question the system figures out which expert

model is best suited to answer it and only turns on that specific part for example if you ask about Finance only the finance expert is activated while other parts stay off this smart approach helps deep seek run much more cheaply than if it had to use the whole system for every question companies quickly lowered their prices making small businesses and startups very happy finally they could afford AI tools once reserved for Tech Giants analysts called it democratization of AI breaking the myth that advanced Tech needed billionaire budgets deep seeks achievement was particularly notable given the

company's relatively small size compared to Tech Giants the success of V2 demonstrated that Innovation and clever engineering could level the playing field allowing smaller teams to compete effectively with well-funded competitors the model's low energy use addressed a growing concern ai's environmental cost by needing fewer computers to run deep seek V2 showed how to make AI more environmentally friendly which matters because data centers around the world use more electricity than entire countries this breakthrough had potential applications in ed Computing mobile devices and other scenarios where processing power and energy consumption were limiting factors deep seek prepared

to unveil a groundbreaking AI project on December 26 2024 they launched deep seek V3 a model that marked a major step forward in AI technology what made this achievement special was that it was done using Basic Hardware deep seek V3 was built using just 248 Nvidia h800 GPU news which many consider basic equipment in AI development this was very different from Big Silicon Valley companies which usually use hundreds of thousands of more powerful gpus despite using simpler Equipment Deep seek V3 performed better than models trained on much stronger Hardware showing excellent skills in coding logical

thinking and math the model worked as well as open AI gp4 which was seen as the best AI system available Andre carpy who helped star open AI praised how well deep seek V3 worked with limited resources deep seeks method was also much cheaper training deeps V3 cost about 558 million Yuan while GPT 4's training cost between 63 and $100 million this showed that you don't always need more computing power and money to make better AI deep seeks V3 success came from smart new approaches like FPA mixed Precision training and predicting multiple words at once these

methods helped deep seek use less computing power while maintaining quality the training took less than 2.8 million GPU hours while llama 3 needed 30.8 million GPU hours to understand how efficient deep seek v3's training was think of it like a Formula 1 race car that beats other cars while using a smaller engine and less fuel this success got many AI experts talking they realized that clever methods and efficient programs could help smaller companies compete with big tech companies by showing that top quality AI could be made with limited resources deep seek changed how people think

about AI development this breakthrough created new opportunities for researchers and organizations working with smaller budgets or limited access to Advanced Computing equipment at Deep seek success came not just from advanced technology but from their unique approach to building their team while their AI breakthroughs amazed the industry the way they ran their company was just just as Innovative their achievements weren't only about complex programs and Powerful computers they were about the people making these ideas real deep seek stood out for its small young team they had just 139 engineers and researchers much smaller than their competitor

open AI which had about 1,200 researchers this small size surprised many people in an industry that usually believed bigger teams were better leang Wen Fung had an unusual way of building his team he looked for Bright Young Talent especially recent graduates or people with just a year or two of work experience he often hired from top schools like chinqua University and pay King University choosing young potential overe experience was risky but it led to Great Innovation the company was set up to encourage new ideas deep seek had very few management levels which helped make decisions

quickly and let team members take charge of their work leang said the company worked from the bottom up letting people naturally find their roles and grow in their own way without too much control from above this simple structure made a big difference in how people worked young researchers felt free to suggest and try new ideas without many layers of management New Concepts could quickly go from idea to reality without getting stuck in paperwork and procedures this change in competition made many big companies rethink their plans and how they use their money scale ai's founder Alexander

Wang shared his honest thoughts about it he said deep seek succcess was a tough wakeup call for American tech companies while the US had become too comfortable China had been making progress with cheaper and faster methods Wang's word showed how the global AI field was changing and reminded big companies they needed to stay alert and keep improving Mark Anderson a prominent investor called Deep seek R1 one of the most amazing breakthroughs he had ever witnessed he was especially impressed that it was open source and could transform the AI industry Anderson's comments showed how deep seeks

new approach could change not just the technology but also how AI companies do business the AI Community began wondering how deep seeks achievement might shake up the market and challenge big companies like open Ai and [Music] meta [Music] fore for [Music]