which industry do you think is the riskiest to start a business, and which is flourishing?
With AI, it seems to be everything possible:
Everybody is a builder, maker, marketer, founder, (vibe)coder/developer.
We are experiencing the "surplus" of AI products... it is harder to differentiate (our time, resources, and attention are limited).
Is there any industry that you consider to be disadvantageous for business in the current situation, and vice versa, what would be worth getting into?
Maybe I just took it from the wrong end and see things too pessimistically š, but I would still like to know your perspective on the situation and maybe see any empty gaps that can be filled.
What I've noticed, for example, is that things related to defense are doing extremely well now, also because of the geopolitical situation in the world.
For example, an Estonian VC firm Darkstar invested in purely military defense tech, especially from Ukrainian startups. The fund recently closed its first ā¬15M round (target: ā¬25M) and backs startups like FarSight Vision (geospatial drone tools) and Deftak (drone ammunition).
Replies
I've been watching this trend play out over a long time, so I'm quite sure, but here's my 2 cents:
Centralised media, social media and other things aren't so normal for humans ā we evolved in smaller communities and tribes, and when we're all put together, it's noisy, messy, even harmful. I think we're peak (or post peak) in terms of the opportunity of new centralised social media, or other online gathering places.
Combine that now with the fact AI is producing almost all forms of media indistinguishably well compared to humans, we don't even know if we're speaking to fellow humans half the time.
So...
I think an interesting area of business is any that now connects humans directly to other humans. From city or location based groups, clubs, communities to specific niche platforms for sports, co-working and other in-person activities. I think demand for less online time will increase, as we a need to find our tribes again and stop being so lonely.
Curious to hear what others think.
@martin_rue Centralised social media ā Yes.
Decentralised ā could be a new water to tap into, but I think that things are going too slowly. Bluesky, Mastodon, Threads... they are trying to succeed,d but somehow they struggle with traction ā so far, only Threads is growing quite well, but just because of their "Meta userbase".
@busmark_w_nika For me they're all dead in the sense that true value is going to come from people eventually choosing smaller places that put them in front of other humans ā not bots, politics, AI content and constant grift. I think tools that help people create smaller, offline communities will see more and more interest as people tire of a centralised internet trying to sell or convince them of things.
@martin_rue that's the exact business I'm in. We are building a community platform (launching soon on PH) that helps businesses build organic communities for user retention.
@thebigk Awesome, good luck with the launch. Cool to see people pushing onward in this space.
@martin_rue @thebigk When is the launch?
@martin_rue @busmark_w_nika - Either end of July or in August. While we have paying users; we are still working on home and landing pages.
@martin_rue @thebigk So it will be soon! Then message me ;)
Great discussion here! I agree that many digital industries, especially those saturated by AI and general-purpose tools, are becoming riskier for new entrants. Differentiation is tough when technology levels the playing field so quickly.
From my experience building KeyHaven, a secure API key management and analytics platform, I've seen firsthand how challenging it can be to stand out...even with a solid product. One of the biggest hurdles has been marketing to developers, who are a discerning audience and often skeptical of new tools.
What stands out to me is the opportunity in overlooked sectors that are less prone to disruption: things like skilled trades, elder care, and essential services. These areas may not be as glamorous, but they offer real, ongoing demand and are less likely to be replaced by software or automation anytime soon.
On the flourishing side, defense tech and cybersecurity make a lot of sense given current global events and the increasing need for digital protection. Iād also add that niche platforms, especially those fostering local, in-person connections, could see a resurgence as people seek more meaningful interactions outside of algorithm-driven feeds.
Curious to hear if anyone else sees promise in other āofflineā or hybrid industries that havenāt been fully explored yet!
@0xtommythomas It seems that more and more people under this thread are agreeing that offline is the new gold mine :D When did you publicly release KeyHaven on PH?
@busmark_w_nika Thanks for asking! Iāll be releasing KeyHaven on Product Hunt this coming Tuesday. Excited to share it with everyone and see what the community thinks!
@0xtommythomas Feel free to dm me on LinkedIn then :)
@busmark_w_nika Just connected!
IXORD
I think that in any area where there is strong competition it is difficult to take a good position (This is in product activities). And any sales business is tied to geolocation and it is more difficult to expand than a ready-made site with SaaS service
@ixord Gauss curve.
When nobody does it = risky (because the market is not prepared and it is unknown).
When many people do it = risky (too many competitors).
IXORD
@busmark_w_nika But there are already methods and tools that help avoid such risky situations and allow the product to take a different path.
@ixord which methods and tools do you mean? :)
IXORD
@busmark_w_nika What I mean by tools are other people's methods and theories that are working at the moment. It is not a product that specifically helps your product.
TradeX
@ixord @busmark_w_nika When too many people are doing it, its a good place to be in because it clearly shows there's demand. A lot time, people build solutions and then search for a problem.
In overcrowded space, if one can a small niche to enter that can act as a great catalyst to capture demand from users who are not happy with current solutions. So when market is crowded, there are always a certain chunk of users who don't like existing solutions.
One example which I can remember is, Cal AI. Calorie tracking is an over crowded space with established players. They came in the market with same product but with one extra feature that is to click photo of meal and upload it. It spread like wildfire with great influencer marketing. Captured hundred thousands of users in short time. Then every one launched same feature. But they have already acquired a great sizeable customer segment. Now they have to keep innovating but it gave them a great head start.
@ixord @divijgoyal The thing is... when the market is overcrowded, it is harder to get any attention.
Triforce Todos
I think traditional media, print, and general-purpose task management tools are becoming riskier; thereās too much saturation and little room to stand out. On the other hand, industries such as defense technology (as you mentioned), cybersecurity, and AI infrastructure are gaining significant momentum.
Also, I have noticed a quiet rise in mental health tech and tools focused on blue-collar industries, and both seem full of untapped potential.
@abod_rehman What about physical books? People reckoned that there would be fewer books (when e-books appeared), and it turned out that people still love physical books.
When it comes to mental health, it could be true, I am also in this industry, but see how more and more people join there. So it means with this traction, it will become riskier too for succeeding (overload) :D
Interesting topic.
My opinion is a bit unorthodox.
The least risky businesses are the ones nobody wants to do and where technology is still lagging behind.
Roof repairs. Home renovations. Elderly care homes. Funeral services. Daycares for children.
All of these businesses are far from AI. And you know there will be demand for them.
Yes, itās not cool to talk about this in online communities like this one, but anyone can build software. Everyone does marketing. But the things I mentioned are almost impossible to replace with AI in the next 10ā15-20 years and for some of them, probably never. And there arenāt many people rushing into them either.
Everyone wants something online.
@byalexai Physical worlds have an upper hand now :)
@busmark_w_nika It sounds illogical, absurd, and somewhat expected. But thatās how it turned out.
I was working on a medical project : a medical center in a small town. No website, no digital presence at all. And yet, 2 million euros in revenue. With just 9 staff members.
Every year, healthcare funding increases. Wars, market shifts, people always need medical services.
Of course, you canāt scale this infinitely. Only by adding more medical centers or more specialists. But itās still a timeless business.
@byalexai It is like that, I can see many people in the industry (buildings, reconstructions, etc) turning millions. Compared to that, some dev projects are irrelevant.
Riskiest IMO would be offshore dev teams. AI is making developer's lives a bit crazy with the daily release of new and improved tools and models.
Flourishing IMO would be AI for sure. Lots of opportunities to invent something new, improve on something current and even use AI to find a new opportunity.
@tinyorgtech Do you have any specific idea about that "opportunity"? What should it be?
@busmark_w_nika I read recently where a company wanted to build in a mature and saturated space. They found 2 things that they felt the leaders in the space didn't do well and built their offering around those two things. I think AI could help you determine what those 1 or 2 things are that your competition struggles with and then it could help you plan and build the solutions faster and with less overhead. So I would say the opportunity is the ability to expose the things to distinguish your offering and then the speed to build it.
Building product was never a problem. You could either do it yourself or outsource it. Building is like 'math'. You write code and get predictable results.
The challenge always has been 'distribution' or marketing it. In marketing, 1+1 can be 2 or 11; depending on how you 'position' it.
Marketing >>> Product. I know a lot of people will hate me for this. But I've been in the industry for 20 years to come to that conclusion. A bad product with great marketing will always win over a great product with bad marketing.
So, back to the question: I consider D2C industry as the hardest to build, if you are on limited budgets. It also looks like the most 'obvious' one to crack; but it's not. B2B on the other hand can produce better results if you approach strategically.
@thebigk It is sad to hear that, but it is true and you are right, "bad product with good marketing > good product with bad marketing". Do you have in your mind such an example with a horrible product but brilliant marketing? š
@busmark_w_nika haha - there are SO MANY! I'll have people pointing their swords at me if I mention products. But let's start the fight anyway :-D
So, the horrible product with great marketing is -
Starbucks Coffee.
@thebigk Have never drinked anything from Starbucks lol :D
@busmark_w_nika - they sell bad coffee for 3x the price. Thanks to marketing and brand building.
The riskiest industry to start a business in right now is the restaurant/food industryāhigh costs, tight margins, and tough competition. On the flip side, tech and digital services (like AI tools, e-commerce, and SaaS) are booming and show strong growth potential.
@emiske77 I think that COVID was more riskier time but I noticed that many people sit in restaurants during weekends (and bars as well). Literally, we have a street in the square and it is a bar next to another bar :D
Even as someone that's genuinely passionate about AI, I share a little in your pessimism. I find it hard not to with the sheer volume of emerging AI based B2B SaaS right now. I'm definitely not saying there aren't crucial gems within them but everywhere you look, the marketās flooded with the phenotype so standing out feels almost impossible, and every new tool just adds to the noise.
At the same time, I don't hear enough about the accessibility gap: yes, low-code platforms sound amazing, but unless youāve got the data sets, the technical chops, and the engineering bandwidth, youāre stuck watching better-funded teams blow right past you. It's not a new issue but I find it exacerbated.
To me, that makes design one of the most unfortunately disadvantageous industries today: generative models are already churning out logos, layouts, even album art and beats, and while Iām genuinely ambivalent about AIās creative takeover, itās clear that traditional designers and artists are suffering the crossfire. It's filling a lot of shoes.
@dheerajdotexe I can definitely agree that AI will twist with many industries (because everything is more accessible ā even I can create something with the code) ā but quality + some differentiator matter.
I think that physical businesses will now have an advantage because they are the most difficult to replace.
Riskiest? If we're talking in general, cab or taxi companies (unless it's autonomous), graphic design (at a basic, beginner level).
Flourishing? No one comes to mind, but if there's a tech company that can target tradesmen/blue-collar work somehow, that seems like it could work really well. Mental health/digital wellness and cybersecurity are obvious choices. An underrated hot take: I believe fashion tech is another interesting niche that I see growing (beauty tech too maybe? not sure if that's a thing but came to mind).
@jordansvision Do you mean cabs because of attacks? :D Or? I have heard or watched somewhere that the riskiest is lumberjack :D
Re fashiontech, this one is interesting. I remember one project where clothes changed their colour according to haptic.
Services, definitely.
Especially in smaller IT service companies that are often founder-dependent. A lot of friends have mentioned that sales are drying up, partly because of the AI hype. Thereās this growing (and false) perception that low-code or āvibe codingā tools can replace full-fledged enterprise solutions. Thatās making it riskier to sell actual software services right now.
@chaosandcoffee Do you mean that the service is the riskiest? I would say there will be more products, but I also have in my mind that the majority of people are "lazy" to do something, so they will delegate it to people who are willing to do it.
Business in non-specialized SaaS is risky due to market saturation and security concerns, making it difficult to stand out or scale safely.
On the flip side, Generative AI is flourishing, driven by enterprise adoption and growing demand for real-time, secure solutions. Entrepreneurs and consultants tapping into these spaces will find opportunities for growth and innovation.
@vivek_sharma_25 But isn't there too much AI? I think once someone comes up with gen AI, it will be easier for everyone to generate anything.
āIf AI can build it in a weekend, why should I care?ā
If somethingās that easy to spin up, itās probably not defensible. Novelty isnāt value.
Example: yet another AI summariser? Dead on arrival.
But a tool that plugs into a hospitalās EHR system and translates medical notes into structured data ā thatās messy, valuable, and hard to replicate.
The real plays now arenāt cheap clones, theyāre bets on infrastructure, integration, or identity. Products that are hard to copy because theyāre tied to real distribution, real trust, or real-world complexity.
@adam_martelletti Things that cannot be replicated so fast and easily will dominate the market if they are useful.
For sure brick and mortar businesses - they're usually behind the times on tech / automation, and could see a significant benefit. They're also not very tech savvy, so it could be a harder sell - having to explain how it works - but also potentially a higher willingness to pay instead of DIYing it.
@wesley_liaw yeah, many people do not want to work physically. But China is quite good at creating robots that can be pretty skillful (and sometimes dangerous) ā remember the video when one decided to attack the audience :D
@busmark_w_nika Terminators are on the way haha! Sooner or later..
Great question. Iād say influencer marketing is risky yet flourishing at the same time. So many pain points still existāghosting, hidden pricing, and slow negotiations. Thatās why weāre working on an auction-based model for sponsorships, hoping to reduce some of those risks for both brands and creators.
@uxyemi Do you mean BidVid?
@busmark_w_nika Yes, BidVid. We've just released beta to the public ahead of launch https://bidvid.io.
@uxyemi I hit the "Notify me" button :)
@busmark_w_nika Oh cool, thanks. See you at the launch š
I think LLMs have done a good job in solving multilingual translation tasks, but they have not yet permeated more actual cross-language and cross-cultural communication scenarios. Human translation companies (language service providers) are dangerous, but the opposite side is worth getting into.
I totally get what youāre saying. It feels like every day thereās a new AI tool launching, and itās getting harder to tell whatās actually useful. Everyoneās building something, so standing out is tough.
Right now, I think starting something in the general productivity or AI-helper space is pretty risky. There are just too many similar tools, and unless you have something super unique or a very specific audience, itās easy to get lost.
On the other side, I agree that defense tech seems to be growing fast, especially with everything happening around the world. Also, industries like healthcare, legal, and security are still figuring out how to use AI in a safe and trustworthy way. That feels like a good space to explore.
Lately Iāve also noticed more tools that mix automation with human input. Not fully AI, but AI that helps you work faster without taking over everything. Those seem promising too.
Youāre not being too negative at all. These are real challenges. Itās just about finding where there is still real need, not just hype. Curious what others think too.