While music industry revenues are increasing, new technologies have emerged to pose fresh challenges. Global Insight explores how companies and artists are responding.

The music industry has experienced its fair share of turbulence in recent years. It has faced significant challenges – such as from internet piracy and tech-based disruptors – but recent indicators suggest the sector is back on track. For the past decade it has seen its revenues increase year-on-year. Figures released by the International Federation of the Phonographic Industry in early 2026 show that recorded music revenues grew by 6.4 per cent in 2025, reaching $31.7bn. It was the eleventh consecutive year of growth, says the Federation.

It’s in this context that US investment company Pershing Square has offered to buy industry giant Universal Music Group in a deal reportedly worth $64.3bn. The investor already owns a stake of Universal’s business, which it purchased in 2021. Announcing the offer in April, Pershing’s Chief Executive Bill Ackman said that despite Universal’s strong artist roster and capability to realise the growth opportunities presented by AI, the company’s stock price has ‘languished due to a combination of issues that are unrelated to the performance of its music business’ and that he believes ‘can be addressed’ by the merger deal.

In April, Universal confirmed that its Board of Directors was reviewing Pershing’s proposal. The Board stated its ‘complete confidence’ in Universal Music Group’s ‘strategy and the leadership of Sir Lucian Grainge and the Company’s management team.’ Pershing and Universal did not take up Global Insight’s requests for comment.

For investors, publishing catalogues are increasingly regarded as an established and financially stable asset class. Song royalties offer highly reliable, recession-resistant and diversified income streams, which aren’t tied to traditional stock markets. And since streaming platforms have successfully turned songs into assets that generate additional income over time, music IP is seen as an attractive investment for private equity companies to buy into.

In recent years there have been numerous major deals where private equity companies have purchased music rights. In 2020, KKR bought a majority stake in songwriter and vocalist Ryan Tedder’s music catalogue, which included his compositions and their associated rights, including to reproduce, distribute and perform his songs. In 2024, Litmus Music, a rights business backed by private equity company Carlyle’s Global Credit platform, acquired Randy Newman’s share of his recorded music and publishing, including hits from the 1970s-80s and the scores to Disney film franchises such as Toy Story.

Major labels are also using private equity cash to snap up rights. In May, Sony Music Publishing reached an agreement to acquire the complete rights portfolio of Recognition Music Group using funds managed by Blackstone. Sony had already bought 4,400 songs from Recognition while the company was known as Hipgnosis. While terms haven’t been formally disclosed, reports suggest that Recognition’s catalogue of around 45,000 songs could be valued at up to $4bn.

The next set of challenges

Despite this deal activity, the outlook for the sector remains uncertain. Growth in the music streaming market has been slower than expected. The streaming giants such as Spotify, Amazon and Apple Music have rebuilt the financial infrastructure of the music industry and have been hailed as saving it from piracy and decline. They control a large amount of the royalty payments that Universal and others heavily rely on. And these giants retain a sizable chunk of the revenues themselves, with paid streaming services making up around half of the global music market.

Meanwhile, social media platforms such as TikTok and Instagram use music to boost user engagement, which in turn drives advertising revenue. There’s increasingly a user expectation that free access to music is part of the experience of using such platforms. But some music companies and artists have concerns about how much money the social media platforms are paying in return.

In response to a request for comment, Spotify directed Global Insight to its 2026 Loud & Clear report, which says that ‘for another year, Spotify was the highest-paying retailer globally – paying the music industry more than $11bn in 2025. This brings total lifetime payouts to nearly $70bn.’ The report adds that in 2025, Spotify payouts increased by more than 10 per cent year-over-year.

In the ‘Royalties’ section on Apple Music’s website, the company says it believes ‘in the value of music and paying creators fairly for their work’ and in ‘paying every creator the same rate, that a play has a value, and that creators should never have to pay for featuring’ on Apple’s curated playlists, for example. Amazon Music and Apple Music did not take up Global Insight’s requests for comment.

Announcing a change to its royalty system in 2025, Amazon Music said in a statement that it strives ‘to support the independent music sector, and is committed to contributing to the healthy and sustainable growth of the music industry.’ It further stated that ‘many’ independent artists and labels will see their royalties increase under the new payouts scheme.

From Taylor Swift to ELVIS

Currently the most significant challenge facing the music industry comes from AI. The technology is readily available, cheap and powerful, allowing industry disruptors to produce enough content on streaming platforms to drown out original songwriting and sink traditional players.

These generative AI music models have often been trained on millions of copyrighted sound recordings without permission being obtained, and without artist recognition or compensation. To protect themselves, record companies have reportedly pushed for broader AI-related clauses to be included within artist contracts. These allow the voices of artists and recordings to be used for AI training, in return for remuneration.

Artists are also increasingly reserving their rights to provide that their music can’t be copied for the purpose of training data sets for large language AI models, says Eileen O’Gorman, Member of the IBA Intellectual Property and Entertainment Law Committee Advisory Board. However, she says that many artists are unaware they may be waiving their rights under pre-digital record contracts. ‘What people didn’t realise is that a lot of the older contracts that were still in existence were silent with regards to digital rights,’ she says.

‘We have seen a number of these […] deals from the 1980s being revisited by the labels in circumstances where an artist remains unrecouped,’ says O’Gorman, who’s a managing partner at Gleeson McGrath Baldwin in Dublin. This is where a monetary advance has been provided by a label or publisher but the artist hasn’t generated enough royalties to pay it back yet – hence the record company’s desire to reclaim the cash by exploiting digital rights. ‘Labels can offer write-downs on unrecouped balances and an increase in the royalty rates,’ says O’Gorman. ‘In consideration, they are asking that the artists confirm that the labels have the rights to digitally exploit and waive any revisiting of past royalties or any claims.’

As a result, O’Gorman says, artists should seek legal advice when they’re approached by a label to revisit or amend an old contract. Where original deals were silent on digital rights, an artist can have some leverage in renegotiating the royalty rates being offered and receive a pay through – where the label or record company uses early earnings to pay off the investment – rather than waiting for full recoupment. But O’Gorman adds that artists must be careful not to inadvertently settle any claims from unauthorised historic exploitation, and should ensure that AI training rights are specifically reserved.

AI can also convincingly replicate an artist’s voice and style – including of deceased individuals. This goes beyond copyright into personality rights territory. There’s no copyright in a voice itself and so there’s a risk of unauthorised commercial exploitation without the approval of the relevant rights holders or their estates. AI replications also have the potential to harm both the integrity of the song and the artist’s reputation. In March, Sony Music said it had requested the removal of more than 135,000 deepfake songs by fraudsters impersonating its artists on streaming services.

Protection in this area depends upon a patchwork of right-of-publicity laws as well as court rulings. In the US in 2024, Tennessee’s ELVIS Act became the first law to explicitly treat the human voice as a protected property right in any medium, covering both living individuals and deceased artists for up to ten years post-mortem. In summer 2025 the Berlin Regional Court II ruled that using AI to clone a recognisable voice without consent violates personality rights – regardless of whether the imitation is produced by a human or by AI software.

Artists with star power and the financial means may be able to set a precedent that others can follow in the future. In April, pop star Taylor Swift filed several trademark applications to protect her voice and image from AI. Two of the filings are ‘sound marks’, a lesser-known category of trademark protection, to copyright her saying ‘hey, it’s Taylor Swift’ and ‘hey, it’s Taylor.’ Another filing protects a specific visual image of her in a costume that’s closely associated with her recent stage performances.

‘The concept of protecting sound as a trademark is not new,’ says Josh Gerben, Founding Partner of US law firm Gerben Perrott, ‘though it remains relatively rare.’ He adds that while trademark law doesn’t prevent identical use like copyright law, it does ‘stop anything that is confusingly similar,’ which is ‘a much broader right and [a] more powerful tool in an AI world.’

There’s also the issue of market saturation and rights dilution. The capacity of AI systems to generate vast quantities of musical content at minimal cost may lead to an issue of oversupply. In such a scenario, pieces of music are churned out in such great quantities that licensing markets become distorted, rights clearance is made increasingly complex and the economic value of individual works is diluted. Human creators would then find it increasingly difficult to compete within existing remuneration structures.

Answering the big questions

Martin Schirmbacher, Member of the IBA Technology Law Committee Advisory Board, says that licensing represents the greatest opportunity for traditional industry players to take advantage of AI’s growth. The major labels – Universal, Sony and Warner – have signed deals with AI startups involving cash settlements, ongoing royalties, equity stakes and commercial veto rights, thereby ‘turning potential adversaries into business partners,’ he says.

Schirmbacher, a partner at German firm HÄRTING, adds that ‘licensing only works if you can prove what was used,’ which is why the ‘mirror opportunity’ for music companies is to invest in attribution infrastructure. This includes fingerprinting technology, properly documented catalogues and the compliance architecture that AI providers will be required to maintain by EU legislation.

‘The players who control both sides of that equation – the licence and the means to monitor it – will define the economics for decades, much as they did when streaming replaced piracy,’ says Schirmbacher. He adds that licensed AI platforms can enhance the fan experience through personalised remixes or interactive listening, creating additional subscription revenue without displacing human creativity. Here, then, AI is leveraged to create rather than simply threaten.

However, rights protection remains a serious concern. ‘Most copyright frameworks were designed around the concept of human authorship, meaning that a creative work must originate from a natural person,’ says Gloria Niembro, Chair of the IBA Trademark Law Subcommittee. ‘AI-generated content does not fit neatly within that model.’

Niembro, an IP partner at Chevez Ruiz Zamarripa in Mexico, says that in many jurisdictions, there’s still no clear answer as to whether AI-generated works can be protected by copyright, and if so, who would be considered the author or rights holder. In Mexico, for instance, the prevailing view is that works generated autonomously by AI aren’t eligible for copyright protection due to the absence of human authorship.

The question of training data is equally unresolved, says Niembro. The use of existing works to train AI models may or may not constitute copyright infringement depending on the jurisdiction. AI companies have argued that text and data mining exceptions in the EU or the ‘fair use’ principle in the US, for example, cover this activity, but music companies disagree.

In the GEMA v OpenAI case in late 2025, the Munich Regional Court found that AI training constitutes ‘reproduction’ under German copyright law. The Court also found that OpenAI’s chatbot ChatGPT reproduces complete training data (‘memorisation’), which means it falls outside the EU’s text and data mining exceptions. These exceptions permit the use of copyrighted works to train AI models only in certain circumstances and provided rightsholders have the option to opt out.

Earlier in 2026, OpenAI submitted an appeal to the Higher Regional Court of Munich. OpenAI says the case ‘raises key questions about how copyright law should apply to advanced AI systems in Germany and potentially across Europe, and its implications go far beyond the nine song lyrics in dispute.’

Some lawyers believe that today’s copyright law can work in the age of AI, with existing principles simply applied in a different manner. ‘I would push back against the common narrative that the law can’t keep up. It can – and recent case law proves it,’ says Schirmbacher. He adds that copyright law was deliberately written in technology-neutral terms and believes there’s no need for new legislation. ‘We need consistent application and better enforcement of the [laws] we have,’ he says.

Schirmbacher adds that in the EU, the music industry has a significant advantage precisely because there’s no ‘fair use’ defence. In the US, for example, AI companies can argue that ingesting millions of copyrighted tracks is transformative – that it adds something new – and therefore is fair use.

However, such an argument doesn’t exist in EU copyright law. ‘In Europe, the default position is clear: if you use protected music to train an AI model without a licence, you infringe, as [the] text and data mining exception mostly does not apply,’ says Schirmbacher. ‘The burden is on the AI company to justify its use – not on the rights holder to prove harm. That is a powerful starting position and it makes calls for entirely new legislation largely unnecessary.’

‘Questions of whether one work substantially reproduces another work, for example, are equally applicable when determining whether a human-created work amounts to copyright infringement and whether a work generated with AI assistance amounts to copyright infringement,’ says Timothy Creek, Vice-Chair of the IBA IP Transactions and Technology Transfer Subcommittee.

But Creek, Principal Lawyer at Davies Collison Cave in Melbourne, adds that legislators globally may still need to consider the potential reform of copyright law in an AI world. ‘Laws more explicitly dealing with the use of copyright materials from human creators being used to train generative AI is one potential avenue for reform,’ he says. ‘Requiring AI companies to gain legal copies of the creative works and regulating both licensing and usage practices could be one form of legislation to provide better copyright protection of works and prevent the under-licensing of works for use in training and developing generative AI databases.’

AI developers are increasingly required to provide greater transparency regarding training data, says Niembro. Meanwhile, licensing-based solutions aimed at ensuring remuneration for rights holders whose works are used in model training are becoming more widespread. ‘The EU’s emphasis on transparency and human oversight may influence other jurisdictions,’ says Niembro, though she believes regulatory convergence will probably ‘be gradual rather than uniform.’

The music industry’s continued financial growth, coupled with the substantial level of investor interest, are strong indicators that the sector remains buoyant. And, while fast-evolving technology continues to present significant challenges, it’s also creating new opportunities for traditional players.

Neil Hodge is a freelance journalist and can be contacted at neil@neilhodge.co.uk