An Editorial by Dawoud Kringle
DeepSeek is a privately owned artificial intelligence company based in Hangzhou, Zhejiang. It was co-founded by CEO Liang Wenfeng in 2023 and is owned and solely funded by Chinese hedge fund High-Flyer. During the last week of January 2025, DeepSeek released a generative artificial intelligence app called DeepSeek-R1.
Their app threw a monkey wrench into the machinery of the world’s economy.
The DeepSeek app has caused serious tech and finance ripples and threatened to render OpenAI obsolete. It accomplished this by offering its service for free (to be fair, OpenAI does have a free version), performing its most advanced tasks at the same level as OpenAI, sometimes beating OpenAI’s best models. To complicate matters for Western tech companies, it was developed for $5.6 million with GPUs that run at half the memory bandwidth. Additionally, DeepSeek’s models can distill other models to make them run better on slower hardware (“Distillation” is a means of extracting understanding from another model. One can send inputs to the teacher model and use the outputs to train the student model). This means that even a small computer like a Raspberry PI can theoretically run local Qwen models (a series of large language models developed by Alibaba Cloud. It is designed to understand and generate human-like text across various tasks). In other words, anyone can buy a GPU that fits in their pocket and fulfills at least some of their AI needs.
DeepSeek is also more energy efficient than any previously existing AI. To power it, one doesn’t need to build a nuclear reactor or a Dyson Sphere around the sun.
Compare DeepSeek’s $5.6 million start-up cost to the $100 million OpenAI spent to develop GPT-4 in 2023, the expensive (and power-inefficient) hardware it requires to operate. Its models are designed and predicted based on the average consumer not having access to the industrial-grade GPU resources and massive energy requirements that the GPT -4 needs to operate.
Investors are hesitant to invest in AI companies because of the cost of training AI models and the lack of viable short-term plans to recoup those costs. And we shouldn’t forget how tech giants like Meta routinely burn through billions of dollars of their investors’ money.
Since DeepSeek’s AI model can rival OpenAI’s best at a tiny fraction of the cost, Wall Street has to begin cutting its losses. Tech stock fell by almost $1 trillion within 24 hours. The market value of NVIDIA (an American multinational corporation and tech company that designs and supplies software, GPUs, and APIs for data science and high-performance computing, System On a Chip for mobile computing, and a major supplier of artificial intelligence hardware and software) fell by half a trillion dollars overnight. On January 30th, SoftBank thought investing $25 billion in Open AI would be a good idea. Tesla’s Q4 earnings were down by 71%.
Overnight, the Trump administration and its close ties with Elon Musk and other tech giants had their plans to embrace AI blow up in their faces. Trump had announced a massive $500 billion AI infrastructure project called Stargate. It involves OpenAI, Japanese investment group SoftBank, and the software giant Oracle. The companies were to commit $100 billion combined to raise $500 billion over the next four years to build data centers to power the AI industry under the Trump umbrella. Trump and his team have been severely humiliated. Their response will not be gentle.
The situation is so serious that Axios newsletter author Dan Primack said, “This could be an extinction-level event for venture capital firms that went all-in on foundational model companies, particularly if those companies haven’t yet productized with wide distribution.”
The situation is complicated because DeepSeek isn’t, strictly speaking, a competitor. It’s an open-source AI model trained on NVIDIA’s GPUs, running on hardware many users still buy from NVIDIA.
The market panicked, but should it have? It’s possible that High-Flyer had intended to influence the stock market to profit from the ensuing volatility and panic. Hedge funds can short-sell, borrowing shares at current prices and anticipating a decline. If the stock price drops, they can repurchase the shares at the lower price, return them to the lender, and pocket the difference as profit. Or they could have engaged in put options trading that gains value as the underlying stock’s price decreases. By strategically timing the release of DeepSeek, High-Flyer could have anticipated a market downturn in tech stocks and positioned itself to maximize profits from this decline.
Of course, investors in politically communist but economically capitalist countries are not the only ones who engage in unethical or illegal methods of lining their pockets. Neoliberal venture capitalists attempted to institute a form of techno-feudalism where people must pay subscriptions if they want to access the modern world. Venture capitalists thought this would guarantee the tech companies infinite money forever (it is not surprising why the Trump administration would find this attractive). AI was an essential and irreplaceable factor in this model. However, their schemes to make endless free money through an unworkable economic model on concepts lacking real social utility or value have been seriously threatened. DeepSeek has changed the rules of the game overnight. And the tech giants have to stand aside, powerless, and watch their entire business models collapse (powerless? I wonder; DeepSeek was recently hit by a cyber attack that forced it to limit the number of subscribers). Whether the Western tech companies and their partners in the political and financial worlds respond by developing better tech at a lower cost or by starting World War 3 remains to be seen.
It is not hard to imagine how this will affect the music business. AI has already altered the game of not only music technology and business but also music as a product that is sold and an art form that is practiced.
Before I examine how it would affect the music business, I strongly urge the reader to consult an article that music activist and professional musician Ken Hatfield contributed to Allegro Magazine (https://www.local802afm.org/
Theoretically, the DeepSeek model could lead to free or low-cost generative AI music apps. This would make music production tools accessible to a broader market. This would include independent artists, amateur musicians, and hobbyists who may not have the resources to invest in expensive software or equipment or the skills to use them. This democratization of music creation may level the playing field, but it could also devalue music of superior or unique quality. One theoretical possibility is that it may eliminate (or severely diminish) the existence of professional musicians, as AI would cultivate a surplus of talented amateurs. While some may argue that generative AI’s allowing more people to experiment with music creation would surge new genres, styles, and innovative sounds, it is equally possible (and quite probable) to lead to an absolute homogenization of musical styles and genres.
It also caused a shift in skill sets. Musicians and producers had to adapt by learning to use AI tools and AI-driven workflows with traditional musical skills.
One idealistic possibility is that AI apps could serve as educational tools. By providing real-time feedback and suggestions, music theory, composition, and production techniques could be taught. Some argue that if AI handles more of the technical aspects of music production, music education might shift toward fostering creativity, storytelling, and emotional expression. There is evidence that this is already happening.
There are many problems with this. Along with the need for a shift in skill sets, AI can generate high-quality instrumentals, vocals, and even full compositions. This is attractive to smaller projects or independent artists with limited resources – not to mention greedy corporations that have no wish to pay for something they can get for free. The need for hiring session musicians or professional producers has decreased and could theoretically be eliminated.
With more people able to create and release music (even without knowing anything about music), the market could become oversaturated, making it harder for individual artists to stand out. It would also necessitate the development of new revenue models. Artists might rely more on live performances, merchandise, or fan subscriptions (e.g., Patreon) as streaming revenues become even more diluted (as we all know, the only ones who make any money from streaming platforms are streaming platforms).
I must pause with a brief digression before continuing:
Patreon is a monetization platform that provides business tools for content creators. It allows them to run a subscription service and sell digital products, and it helps artists and other creators earn recurring income by providing rewards and perks to their subscribers. Many professional musicians use Patreon to supplement their income. Until 2024, Patreon could go around Apple’s 30% commission using alternative payment processors. But now, Apple has begun forcing Patreon to use its in-app payments system. This means Apple will get a 30% commission on all new subscribers made with their app. In other words, Apple is trying to take almost 1/3rd of the income from independent creators by quadrupling their transaction fees. Refusal to comply will result in Patreon being kicked out of the App Store.
Apple’s mafia-like genocidal tactics against independent artists, professional musicians, and content creators do not stop there. When an independent artist releases a single through iTunes, Apple has to pay him / her one dollar for every download. Apple is now phasing that out for independent artists. The major labels will still have access to this, but the independent artists and labels are being kicked to the curb.
Recording studios and professional music producers have been facing reduced demand. With the obvious exception of studios and producers who work with genres that require highly specialized in-person acoustic environments (such as classical and jazz), AI is forcing them to adapt. One way would be to offer specialized services or integrate AI into their workflows, which would require a great deal of ingenuity.
Anyone in the music business knows that generative AI has raised questions about whether it’s the user, the AI developer, or nobody who owns the rights to AI-generated music. The US Copyright Office has recently issued decisions regarding the legality of AI-generated works. AI-generated music could inadvertently replicate existing works, leading to plagiarism or copyright infringement accusations.
It is not inconceivable that labels and streaming services would have access to AI tools vastly superior to anything consumers and small business professionals have access to. If this is the case, it is not difficult for them to be able to predict or calculate when something is going to go viral. There is a pattern that suggests this. When something is predicted to be about to go viral, Apple Music or other large corporations buy the independent labels and platforms, promotional vehicles, or Digital Service Providers that control it. This way, they control the viral. Their obvious objective is to eliminate artists and replace them with AI-generated content. This would be the cornerstone of a system where they no longer need to pay mechanicals or publishing to the artists and publishing owners.
Streaming platforms like Spotify or Apple Music have become inundated with AI-generated tracks. This makes it harder for untrained and unsophisticated listeners to differentiate between music created by humans and by AI, or at least to discover high-quality, human-created music (and let’s face it; many people simply don’t understand enough about music to even care. All they require is background music for their parties). Platforms could solve this by creating a new content category and developing special AI-curated playlists highlighting the best AI-generated music, but it’s doubtful they will.
Inevitably, AI would enable the creation of virtual artists. Of course, there is nothing new about this. The Japanese have created virtual bands since the 80s. These “Idoru” have roots in Anime and Japanese Idol Culture (such as the Macross Mecha franchise and fictional singer Lynn Minmay). Generative AI will doubtless take this to unprecedented levels. It would be simple to have them perform in virtual reality (VR) or augmented reality (AR) environments. As AI becomes more sophisticated, the line between human-created and AI-generated music will blur and challenge traditional notions of artistry and creativity.
Live music experiences may undergo a fundamental transformation beyond the aforementioned VR & AR Idoru performances. Artists could incorporate AI-generated visuals, backing tracks, or even real-time musical improvisation into their live shows (see the story Chatbot from my book Bedtime Stories for Musicians and Other People: https://a.co/d/gnQk338). This can potentially create new music and art forms that couldn’t otherwise exist.
AI-generated music tools could inspire entirely new artistic movements and cultural trends. With the introduction of DeepSeek into the tech ecosystem, these anomalies, both good and bad, have become amplified and democratized even more.
Time will tell how all this, combined with presently unknown factors, will play out. Ultimately, musicians must take control of how this technology affects our lives.
