The forces acting against online course providers since hundreds of billions began pouring into LLM infrastructure are suffocating. We set out a path to counteract them.
You are no longer competing against each other
If you've been around long enough, you'd remember the days of the online courses and MOOC platform gold rush. And as fast as it came, it faded. What remained was a hyper aggressive environment of discounting, overwhelming subscriptions offerings and a transition away from individual learners to enterprise.
The trends in online education in 2025 and shift in learner behaviours present the online courses industry with an ultimatum.
Either start worrying about the real competitor, foundational model providers, or sink.
I foresaw this 18 months ago as both an avid learner and someone who has been at the intersection of education, research and tech for a decade. Not because of any special characteristic I possess. Rather because I am always experimenting on myself with learning.
For instance, I used ChatGPT and Claude simultaneously to reteach me computer networking, which I didn't care much for in my undergrad years.
Despite the chat interface, with perseverance in few-shot prompting I managed to get it teaching me in a tone, style and method that frankly has never been possible before.
Fast forward to today, 31 July 2025, ChatGPT has formally rolled out Study Mode after a 2 week pilot and is available to every user.
They have your content, the distribution monopoly and the AI competence
Crying over spilled milk is not going to change the state of affairs. The internet was scraped and it's done.
Copyright laws are dead and will not protect you or me or anyone, especially since Trump's government has made it clear whose team they're on. Move on.
Now, here's what you actually have to be worrying about.
Between Chatgpt, Grok, Claude and Gemini alone - monthly active users total ~1.3 billion people. Suppose there's 10-20% LLM hoppers like me in these stats, that's still a round 1 billion active users.
This distribution is both unattainable and impossible to compete with, especially since it is not showing any signs of slow down.
And the LLM providers are out for blood. They want all and every market.
The juicy $2.1 trillion learning and education market is next.
Just to make matters worse, the AI FOMO engine is firing on all cylinders. You are reading this because you know you need to respond to AI, but haven't. Your enterprise clients are also panicking, and they're now reassessing everything that doesn't build on AI. So that leaves you between a rock and a hard place. Drying up demand and heating up completion.
All is not lost, and I believe there are a few strategies to counteract these forces in addition to adopting AI yourself.
Lean into the learning experience and specialise
The race for AI winners in each industry is only just starting. Nobody has taken the market, or will just yet. Google fumbled and Openai sprinted past briefly in search. Runway and Pica took off in video then Googles Veo changed the game. The Chinese models periodically come forth with engineering and efficiency marvels.
The point is that it's a chaotic scene right now. This, as an online course training or platform provider, is your advantage.
The most obvious path is to leverage your area of competence not to push out more content, rather repurpose it. Start by killing of the idea of course completion. It's fundamentally flawed and silly to track.
Instead, you ought to be thinking about how to be a companion to the user by understanding them. Do they value human interaction? Are they in need of relevance - can you enrich the material on a daily, weekly basis to provide deeper meaning to what they're learning from current events in the news?
In person lesson dates, or co-learner matching can all be thrown into the mix.
Real personalisation.
This continuous value generation and specialisation is only doable by you, but will forge a new path in learning and learner stickiness. Focussing on bringing the learner back, regardless of their career pursuits and motives is the next frontier in online courses technology.
At this point, I will shameless plug trymarcus.com - an infrastructure layer we built to allow deep personalisation and allows you to convert your content into adaptive experiences for each user.
Vertical integration, or partnerships at least
A promise is a comfort to a fool. Learners are wiser and chatgpt and perplexity can spin up data evidence in seconds to show them how much your certification or course will actually help them land a job.
That's why you need to build on deep personalisation, with the aim of landing jobs or positive outcomes for your users.
This era is the pipeline era, and you need to build one of value for the learner so that you can assist them landing an interview. It might even be an add on, assisting them securing one interview - but really trying to get them to succeed.
Recruiters and employers are always open for business for the right talent. Compiling an AI-driven report on potential candidates based on their engagement, interaction with your content and overall performance is a unique and seriously valuable proposition you can make.
If you don't have the balance sheet for M&A, get partnering. If you need some help, our clients are always ready for an intro.
Consolidation, alliances and trimming the hedges
There has to be some losers in the space.
If that's you, you won't get any notice. It'll just happen overnight.
But if you're up for a fight, you need the right alliances. As the famous Arab saying goes, better a thousand friends than one enemy. And it helps that everyone is in a similar position in the online courses and online learning space.
Before you go out shopping for targets, first the hedges need to get trimmed and all the useless weight needs to be shed. These are your lower quartile courses, the ones that have sporadic demand that's just not worth it. Either quick fire sell them to a provider in the space, or just cull them off.
You want to present yourself as the course provider worth having on the team. So your data has to be useful, your competence in a niche or two has to be high and your client book should merit some raised eyebrows.
If you are on the buying side, I just listed what you should be looking for.
Ultimately, the choice isn’t course vs. course anymore. It’s human-led ecosystems vs. AI monopolies. Either we merge, partner, and embed ourselves into the AI economy, or we’ll be erased by it.