As hardware startup investors, we spend a good part of our lives looking at hardware projects and predicting winners and losers. Naturally, we’d like to see more entrepreneurs in the “winners” pile. So for anyone working on a hardware product, here’s our best advice on what types of projects are worth attempting in 2018 and what types you’d be better off avoiding — if you’re hoping to get funded, that is.

1. What will be hot

RaaS (Robot-as-a-service)! Some of these startups might grow as franchises, which would help them expand across multiple cities like Uber.

Robots that we never see or get scared by — in other words, robots that do dirty, dull, or dangerous jobs. Look at robots for warehousing by Fetch Robotics, for the service industry by Savioke, for inventory like Bossa Nova and Simbe Robotics, or for commercial cleaning like Avidbots. [Disclosure: We’re investors in Avidbots and Simbe.]

Sensing devices for traditional industries. Smart sensors tied to the internet of things (IoT) can help rid our world of inefficiencies and create new business models. Traditional industries from retail to agriculture and mining will benefit from new solutions.

Consumer devices that save lives thanks to body-sensing data.

Enterprise and industry IoT. Consumer products tend to be “nice to have” rather than “must have,” so the market for business-to-business IoT products is likely to grow faster.

In short, we are big on B2B.

2. What still won’t be hot

Consumer devices. The era of copycats led by Amazon, Google, and Facebook is leading to a no-man’s land for consumer-focused hardware startups.

Over-hyped consumer IoT. Companies that set ridiculously high goals and then — surprise, surprise! — don’t deliver on them will continue to bring down the market. Stories of Jawbone, Juicero, Teforia, and Pebble, for example, have raised investor wariness despite the fact that consumer device startups can be built more effectively today. (We’ve seen some companies go from prototype to market in less than two years with less than one million in venture investment, for example.) If you’re in this sector, be fast, keep a low burn, and look for alternate sources of financing to VC like grants, prize money, and product or equity crowdfunding.

This wariness on the consumer IoT front doesn’t mean VCs won’t invest, but it means they’ll be keeping the bar high when looking at return on investment, and they’ll expect startup teams to be fast and frugal.

3. What we hope to see

Prototyping tools that can double as production tools. With the right economics, even niche products in this space would be great news.

Easier regulations in key industries. Reducing bureaucracy, especially in sectors with huge potential for innovation such as transport and health care, would clear the path for all kinds of new businesses. Currently, delivery robots are facing hurdles in San Francisco, and Hong Kong’s first autonomous car is forced to test in Shenzhen.

Readiness from the health value chain. We’d like to see hospitals, insurers, and pharma companies understand the value there is in consumer health hardware and start using it. Our firm, for example, has invested heavily in this category over the past two years — from depression treatment to sleep-inducing technologies. The jury is out as to whether “digital therapies” will complete. or compete with, big pharma.

Long lasting batteries. Battery life is still one the biggest barriers for many devices.

4. What we hope we won’t see

Mega-rounds by startups with no product. No more $100 million rounds without a product, or a client, please!

Dissing Chinese innovation. Labeling Chinese hardware plays as copycats or imitations is a mistake. So much innovation is happening now in China on the hardware front, including a number of ideas we haven’t even seen in the West yet. China isn’t just innovating with drones (DJI), flying taxis (eHang), and self-balance vehicles (Ninebot), but also dock-free bike rentals (Mobike and Ofo), which are aggressively expanding overseas. And those are just some of the large, visible companies. A flurry of younger startups will make a name for themselves very soon.

Humanoid robots. ATMs and washing machines automated useful tasks with no need for a face or a pair of arms. Robots that can do backflips are cool, but do we need them? The legacy of science fiction is doing us a disservice by associating the future with androids. There are very rare cases where a humanoid shape is both suitable and cost effective.

The common thread here is that the old views are outdated. Perception always lags reality.

The takeaway

If you’re a hardware startup looking for funding in 2018, the key thing to remember is that B2B is hotter than consumer. And health is a segment that should see a lot of potential in the coming year. Still, if you have a strong enough product, breakthroughs are possible in any category.

Cyril Ebersweiler, Duncan Turner, and Benjamin Joffe are partners at HAX, the hardware branch of the SOSV fund.