For months, I have had a back-and-forth over instant messenger with a journalist friend in Asia. My argument: “Old media” or newspapers are bouncing back as they give up their reliance on the advertising-based model to focus on digital subscriptions. The advent of generative artificial intelligence, or AI, is helping put winds in their sails as large language foundational models like OpenAI’s ChatGPT, Google’s Bard or Meta Platform’s LLaMA need more factual data to train on rather than any unsubstantiated stuff a web crawler can get its hands on. For his part, my friend insists that old media can only survive on an ad-based model.
Let me lay out my case in the court of The Edge Singapore readers. In 2000, US daily newspaper industry revenue peaked at US$90 billion, adjusted for inflation in 2020 dollars. When last year’s numbers are tallied up, analysts expect they will show that ad revenues for US newspapers have fallen 70% over the past 23 years. Websites and smartphone apps enabling users to access news without a subscription have increased competition for readers and ads, hollowing out the old media. Much to the chagrin of publishers, however, revenue gains from online ads have not made up for the loss in print ads.
About a decade ago, Netflix, the video streaming pioneer quietly began building a huge subscriptions business from scratch. Its aim was to deliver the best video entertainment — movies, TV serials, documentaries — for a reasonable monthly fee. Media gurus opined that the streaming pioneer was destined to fail. You can’t build a media business without advertising, they argued. For years, Netflix lost money but doggedly stuck to its model of growing subscriber base and its library of video content. Then about four years ago, streams of profits began to gush out of the streaming spigot. The media world has never been the same again.
The Netflix model was simple: If you have compelling content, you don’t need to rely on advertising. Users will pay for quality content. Netflix raised prices every two to three years and kept growing its subscriber base. Over the past decade, many newspapers have taken a page from Netflix’s playbook and focused on digital subscriptions to make up for loss in ad revenues. Newspapers pivoting from an ad-based model to a subscription-focused one are beneficiaries of changing consumer media consumption patterns.
Young people these days no longer look at ads the way their parents and grandparents did and are happy to pay for subscriptions. Not surprising, then, that many newspapers now make more money from digital subscriptions than they ever made from the old Ad-based model.
The New York Times (NYT) is my Exhibit A. The newspaper’s revenues peaked at US$2.1 billion in 2006 with ads accounting for nearly 63% of its business. By 2016, revenues had plummeted to US$1.5 billion with ads making up just 40% of the total. Meanwhile, NYT had erected a metered paywall in 2011 to build a subscription model. Consensus estimates of NYT revenues are US$2.45 billion ($3.3 billion) in 2023 with ads falling to 20.5% of total revenues. The newspaper giant is forecast to turn a net profit of US$270 million when it reports annual earnings in two weeks. This year, consensus estimates of its net profits are US$280 million on revenues of US$2.72 billion. Ads are expected to account for just 19% of the revenues with subscriptions making up 72% of the revenues.
See also: ChatGPT’s US$8 tril birthday gift to Big Tech
As the world’s largest newspaper with 10.1 million paid subscribers (9.4 million online-only and 700,000 print subscribers), NYT dwarfs its peers The Wall Street Journal (3.5 million paid digital subscribers), The Washington Post (2.5 million) and Gannett, which owns USA Today, with just over 2 million. Paid digital subscriptions are still growing fast.
In NYT’s July-Sepember quarter, digital subscriptions grew an annualised 10% although discounting to rev up growth cut its average revenue per user, or ARPU, to just under US$10 a year. ARPU includes lower-cost subscriptions of affiliate products like its sports offering The Athletic or recipe database NYT Cooking and product review portal The Wirecutter as well as NYT Games that includes its daily crossword as well as Wordle.
You can buy the whole “All Access” bundle, or just subscribe to the newspaper or buy standalone subscriptions to sports or cooking websites. NYT added 490,000 net paid subscribers just in the last quarter of which 300,000 subscribed to the news only. Analysts expect NYT’s total digital paid subscription revenues will top US$1.3 billion this year, up from US$ 1.15 billion in 2023. That’s more than it made from ads at its peak in 2006. Full disclosure: I am a long-time NYT digital subscriber. I used to pay US$52 a year for a subscription until last year when I threatened to cancel and demanded a refund. Eventually, I negotiated the price down to US$26 a year.
See also: DBS, EnterpriseSG and IMDA launch gen AI programme to help up to 50,000 SMEs adopt tech
If you are growing your subscription base by over 10% a year, a small decline in ARPU doesn’t matter. NYT’s gross margin has been fairly stable at around 50% in recent years while operating margins are around 10%. For a sector like the news media that was left for dead a decade ago, those are incredible numbers.
Subscriptions now account for 70% of NYT’s overall revenues. Two-thirds of the subscription revenues are digital while print accounts for the rest. About 10% of NYT’s revenues come from content licensing, Wirecutter affiliate referrals, commercial printing, leasing and live events. Content licensing is seen as the next growth driver for NYT as firms like OpenAI, Anthropic as well as Microsoft, Google, Meta Platform, Amazon and Apple pay for content that they are using to train their larger language models.
Upselling and Cross-selling
Instead of succumbing to click-baiting to win the pageviews arms race or sucking customer’s data which was then sold to marketing firms, NYT’s strategy has been to focus on quality news and earn the trust of its readers so that they are more than willing to pay for its news. It also stuck close to Netflix’s playbook of upselling and cross-selling a smorgasbord of content, which in turn helped improve its lifetime value per user and its ability to further invest in user acquisition. NYT also used the classic tech strategy to maximise network effects. The more subscribers it can bring into the fold, the more it can invest in content, which in turn attracts even more subscribers thereby creating virtuous cycle of growth.
Investors have been cheering. NYT’s shares bottomed around US$3.50 in February 2009 in the aftermath of the Global Financial Crisis as ads plunged and subscribers rushed to cancel their monthly US$30 print subscriptions. NYT stock is hovering around US$48 or a 14-fold increase over 15 years. Investors are rewarding the makeover of NYT from an ad-based cyclical business to a higher margin subscription-based recurring revenue platform. Last year, NYT subscription revenues grew even as the US Fed raised interest rates and growth slowed.
The future of NYT, however, is in generative AI and the content that is used for training foundational large language models. Last month, NYT sued Microsoft and OpenAI for copyright infringement. The suit alleges that millions of articles from NYT were used to train chatbots that now compete with it as a source of reliable information. Many articles were republished verbatim without crediting the paper. While the suit does not include an exact monetary demand, it says the defendants should be held responsible for “billions of dollars in statutory and actual damages” related to the “unlawful copying and use of NYT’s uniquely valuable works”. It also calls for OpenAI and others to destroy any chatbot models and training data that use copyrighted material from the newspaper.
For its part, OpenAI argues NYT’s lawsuit against it is “without merit”. Using copyrighted works to train its technologies, the firm argues, is “fair use under the law”. OpenAI also accused NYT of manipulating prompts to include regurgitated excerpts of articles. “Even when using such prompts, our models don’t typically behave the way The New York Times insinuates, which suggests they either instructed the model to regurgitate or cherry-picked their examples from many attempts,” OpenAI said in its rebuttal. The NYT articles that were regurgitated by AI “appear to be from year-old articles that have proliferated on multiple third-party websites.” The reason why OpenAI, Microsoft and others are using NYT content for their training models is because it is seen as more valuable. NYT content is considered more authoritative and trustworthy than the stuff on some Russian websites.
NYT is not alone in seeking licensing deals with news organisations to train AI systems that can produce human-like prose. Publishers including the largest US newspaper chain Gannett; News Corp, the owner of The Wall Street Journal; IAC, the digital giant behind The Daily Beast and the magazine publisher Dotdash Meredith as well as The News/Media Alliance, which represents over 2,200 news organisations in North America have been in talks to license content to the creator of ChatGPT. However, they have all failed to reach an agreement on the price and terms of licensing their content to the AI firms. News firms are reportedly asking for tens of millions, in some cases hundreds of millions. AI and tech firms are willing to pay a pittance. OpenAI is now valued at over US$80 billion after a recent funding round. Software behemoth Microsoft which has committed US$13 billion in OpenAI last week overtook iPhone maker Apple Inc to become the world’s largest listed company with a market cap just shy of US$3 trillion.
Sink your teeth into in-depth insights from our contributors, and dive into financial and economic trends
Here’s the backstory to help you put things in proper perspective. In the aftermath of the Internet boom in the late 1990s, as news publishers put up their content for free, search sites like Google and social media sites like Facebook made billions from selling ads against content that was not their own, systematically destroying the business model of newspapers. Understandably, media firms once bitten are twice shy. They don’t want to hand over the keys to the kingdom on a silver platter to rich tech bros. Two decades ago, they didn’t know Google or Facebook would one day be worth a trillion dollars. Now they know the true value of their content. Can you blame them for demanding top dollar for it?
A bigger issue for news media is credibility. NYT lost 70% of its ad revenues but is still thriving because it is seen as a trusted news source. People pay for its content because they know NYT stories are repeatedly fact-checked and vetted by editors and lawyers. The problem with the Internet, and now AI, is that there is too much unchecked, unsubstantiated garbage. The lawsuit isn’t a cash grab by a newspaper to squeeze a cash-rich tech giant. It is about the reader’s trust. You lose that trust and the billions you might get from AI licensing fees won’t save your business from going under.
Over the next year or two, I expect NYT and other newspapers to strike a deal with the tech giants. I also expect NYT to announce that it has created its own large language model and ChatGPT competitor. NYT is a tech company and really when you look at it, tech is firmly behind the venerable newspaper’s formidable comeback. Its forays in AI will entrench it as an even more serious tech player.
Assif Shameen is a technology and business writer based in North America
