Steve Kraus, one of the most active health-tech investors in the industry, gave his predictions for 2018, following the J.P. Morgan Healthcare Conference this week in San Francisco.

I think the public markets will remain closed in 2018 for health-tech companies. However, I do expect 2019 and the years beyond that to start to look more fruitful for initial public offerings, as well-funded start-ups like Health Catalyst, Flatiron Health, Welltok, Grand Rounds Health and others start to scale beyond $100 million in revenue and some of them start to reach positive EBITDA levels.

Healthcare mergers and acquisitions will remain robust but will focus more around health-care service businesses — many of which may be tech-enabled — rather than pure health-care IT companies.

I’m bullish on so-called “full-stack” or vertically-integrated health-care companies continuing to thrive. And by that I mean, companies that provide all of the necessary services and support around conditions like diabetes or the palliative care experience.

These services are the ones that are getting real engagement from patients and they are able to go to risk-bearing entities with a pitch to take a population off their hands — and provide better care more affordably. These solutions will also work better for the patient as they take care of their totality of their medical issue.

Few could have missed the hype around artificial intelligence and machine learning technologies in health care in 2017, specifically related to fields like radiology and pathology.

A vast array of companies are developing algorithms to improve clinical care, but these will take time to develop commercially. In my view, 2018 will be the year that these technologies are instead applied to the “guts” of healthcare, meaning such unsexy applications as improving workflow, allowing doctors to spend more time on clinical care and less time on paperwork or operations, improving the operational efficiency of hospitals, and so on.

2017 was the year that Silicon Valley started to invest in computational biology. A lot of these venture dollars were poured into companies focused on improving the early discovery process, such as identifying targets and “hits” of those targets.

I’m betting that 2018 will be the year where these companies and investors realize that biology is not “programmable” or “codeable.” The problem isn’t that pharma and biotech are short of “hits” or drug candidates. What these companies really need is clinical-stage programs. If these start-ups really want to scale and actually get traction among the traditional industry, they will need to become real biotech and drug development companies. And this will be a different, much more capital-intensive bet than their original investors probably realized.

Apple, Alphabet and Amazon will continue to push forward in health care in 2018. If these companies make a real big splash with some kind of acquisition, it could crack open the IPO markets sooner than any of us think.

I have seen far too many companies in recent months that have raised behemoth rounds at mammoth valuations return to the market with their tail between their legs — finally open to valuation adjustments. These could still be potentially successful companies, but their investors were potentially mistaken that such companies could reach hyper-scale.

Steve Kraus is a partner at Bessemer Venture Partners.