GDPR will stale traditional (and digital?) banks for a while.
Whilst most of the banks (and other companies) have been aware of the GDPR requirements for a long time now, as with any other highly time consuming projects where technology has a potential to replace massive amounts of manual efforts (like blockchain solutions for KYC/AML or Mifid II implementation), it is unlikely that the banks will have sorted their data permissions using a digital solution this time either.
The future will show what impact this might have on account holders being caught between client facing departments and banks’ compliance divisions looking for solutions shortly before the deadline (looking forward to receiving even more paper letters from my bank in 2018…). Some of the banks might get it right and some winners might emerge turning this rather unsexy topic into a selling argument by the simplicity and user-friendliness of their solutions. It’s likely that some players will fail and we’ll see first GDPR related fines flow in 2018.
Bold prediction: 2018 will be the year of cybersecurity (and NOT the year of Bitcoin).
Manufacturing is the new breeding ground for AI companies.
Andrew Ng, one of the thought leaders in AI, just started his new AI company in manufacturing. This is little surprising if you follow his career and reasoning. Manufacturing basically affects every aspect of our physical life and the sheer amount of data is a great basis for optimization algorithms and regression. Analyzing this data can lead to lower operating costs, increased efficiency, higher machine uptime, i.e. think of predictive maintenance and faster production time for example.
Incumbents in industrial automation are also trying to move in that direction. They know the manufacturing processes very well but have little strength in Big Data and ML capabilities so far which opens up a big window of opportunity for new companies.
In 2017, we already saw an increasing number of companies that are using computer vision or machine learning to solve problems in the manufacturing space. I strongly believe that this number will keep on increasing in 2018, especially in Europe where we have a lot of highly skilled technical talent and 29M people who are working in this industry. You will see former employees from Bosch or Siemens starting their own company. This is a challenge for VCs at the same time because they might be hard to see at the very early stage.
In addition, this trend is also underlined by the amount of money VCs are putting into manufacturing startups. In 2018, we will see more top US VCs that are going to invest in European manufacturing startups, similar to the $9M Series A at Konux lead by NEA in early 2017.
Bold prediction: Munich based SaaS companies will raise more money from VCs in 2018 than their peers in Berlin.
Head of Talent
After On-premise and Cloud, the third generation of intelligent HR tools will rise.
In 2018 the transformation of traditional “HR” is finally up to full speed. We will see more products popping up that help traditional HR to productise their new won “freedom”. Stepping out of the past years’ shadow of HR products that mainly helped to digitize old school processes, the new HR products will have one major target: data.
So we will end up seeing more and more “scientific approaches” to day to day challenges like team fit, company culture and performance improvement. However hiring will continue to be the biggest challenge of them all.
Machines will become creative.
Up until today, the most common answer to the question “which jobs will never be impacted by AI?” has been “the creative ones”. As a matter of fact, most of the AI companies we’ve seen so far have been focusing on predicting a specific output such as the likelihood of a lead to convert, the optimal set of controls to optimise a greenhouse yield or the probability for a website visitor to click on an ad. All were tasks involving a very small amount of creativity.
One thing we overlooked, though, is the ability of intelligent systems to create new products such as a new product design, an image, a piece of music or even a piece of software. As MIT professor Erik Brynjolfsson explains well in this Science article, with AI comes a new computing paradigm in which developers don’t need to specify in great detail the whole design process. They just need to specify some of the characteristics of the desired output and the machine will be able to come up with a new design. This is what we call “generative design software”.
This opens a vast range of possibilities, from letting software design new music songs — check iMuze if you’re interested in this, create new images that convert better — I believe one of Merantix’s venture is also working on such a topic, design better car chassis — Autodesk did so, see image, or even mimic Obama’s lips movements.
I believe that in 2018, we’ll see a lot more business applications letting AI create or invent new designs. And I believe we can be quite excited about this!
Btw, if you start wondering what machines will never be able to do as well as humans, think about our emotional capabilities. But let’s speak about that again in 2019!
Bold prediction: There will be a massive correction on XRP’s price. Twitter’s stock will do very well this year because the user base will grow again, and engagement will increase.
Junior Legal Manager
Blockchain: the future of certification?
Taking into consideration the slow path of development and the conservative aspect of the legal industry, I would consider this more a long term prediction rather than an actual reality (unfortunately!).
Some of us might be familiar with uncountable certifications, last-minute notary appointments, commercial register requirements (together with its correspondent fees), countless reports and especially Anti-Money Laundering (AML) and KYC checks.
What if this could be all checked in a public database? Assuming that Blockchain is an up-to-date ledger of all kind of transactions; one could easily monitor the status of a deal on a real-time basis, verify the authenticity (and identity) of a signatory and even proof beneficial owner information.
Furthermore, this would not only apply to attesting tasks strictu sensu but; could we imagine relying on a record which gathers all intellectual property rights? These are known for entailing a long (and costly) dispute resolution. And what about an archive which contains true information regarding land ownership? This would for sure eliminate all kind of inheritance and administrative headaches.
Which would be the perks of this utopic system? Fewer costs, shorter (and even immediate) deadlines, unified reports amongst the countries, reduction of litigation cases, security regarding the transmission of big volumes of data and of course transparency towards the authorities and third parties through the elimination of the one and only intermediary that would normally surveil the legal transaction.
Of course, there are downsides here as well: the amendment of a big part of juridical constellations we know so far, the grey area regarding data privacy (who has access to what and until what extent), reduction of job positions (specially in the public administration), storage limits and primarily the challenges of integrating blockchain in traditional corporate systems and registry policies.
#SurveillanceCapitalism: services and products built around privacy protection will continue to get traction.
In 2017 I finally put some thoughts into my “dependency” on Google, Facebook, Amazon and the potential abuses that could result. I don’t think I’ll go from one extreme to the other and stop using these services entirely, but I want more control over the data I share online. This is why I started using alternatives for Google Chrome and Google Search with Brave Browser and DuckDuckGo, or that I completely stopped using Facebook (I don’t plan on abandoning Gmail, Youtube or Gmaps, I just cannot stop using Google products all together).
My prediction is that many of the services built with privacy protection “at heart” will get more market adoption / traction in 2018 (but not mass adoption). We’ll also see new interesting products emerge in more categories — like on Mobile where I couldn’t find any easy to use alternatives to Android or iOS, or to counter the increasing surveillance imposed by States.
Historically the B2B industry always gets hit by big trends with a delay (ex: social web or mobile), but some significant legal changes, like GDPR, are already starting to impact the B2B landscape. To be honest, I don’t know how and at which pace it will change the SaaS industry — many people think that the big players will benefit from GDPR — but it’s a trend I’ll be paying attention to.
Bold prediction: Amazon will launch (or buy) its BTC exchange. Like eBay bought Paypal in 2002.
Virtual worlds as the testing ground for intelligent systems.
Much of the investment and research community focuses on data as the rate limiting step for building software that learns the optimal solution to a given task. We focus our machine learning-driven companies on data strategies as a means to defensibility. Indeed, crowdsourcing the creation large scale labeled datasets such as ImageNet has historically driven the unexpected success of deep learning in recent years.
Yet there are many application domains where machine learning can bring significant value but where traditional crowdsourcing is expensive, slow and privacy infringing. For example, an autonomous vehicle needs to drive millions if not billions of miles before it can be deemed both technically safe and trustworthy by an end user.
What’s more, solving problems that reside downstream of perception, such as planning and decision making, often requires a faithful model of the problem space such that a machine learning system can discover the optimal strategy to accomplish a desired task. As a result, complex simulations of real-world environments will become commonplace in the engineering and training of intelligent systems to solve high-impact problems. Back to the autonomous vehicle example, Waymo’s vehicles drove 3 million miles in real life and 2.5 billion miles in simulation. This approach of writing a locally optimal solution using real-world data and then discovering an even more optimal solution through simulation (or vice versa) is extremely powerful. Watch this space!
Bold prediction: With more ML services being deployed, we’ll see cyber risk of adversarial attacks go mainstream, leading to major hacks and new security software catering to ML companies.
Every investor will be a “frontier tech investor”.
I think 2018 will be the year when every investor will be a “frontier tech investor”. We saw that coming, but today in most of the companies we meet there’s some form of machine learning and/or crypto and/or hardware.
I also think that new platforms will emerge in different verticals — voice for consumers at home and for mobile; VR for games; wearables for sports (and more) which will result in an ever increasing fragmentation of the development stack. I guess that the funding environment for such businesses in 2018 will be more welcoming, but also competitive.
For “traditional” SaaS and eCommerce businesses, it seems like 2018’s financing environment might be tough. If you’re a founder in that space, the bright side is that you will only need to win the FAANG oligarchy, because there will be limited competition from well financed startups.
In terms of the next areas of “geekiness development”, more recently, we have seen some renewed interest in software eating healthcare (including pharma and genetics) and the food supply chain (agriculture et al). My guess is that customer behaviour is changing and there are new techniques and access to vast amounts of data that are creating great opportunities.
Bold prediction: we will see clear signs of the decline of Silicon Valley as the “de facto” monopoly hub for tech innovation.
It will be a very volatile year for crypto markets.
Having attracted a lot of new participants and funds and having shown an amazing growth in value in 2017, I expect that crypto markets will be having a very volatile ride in 2018. Numerous major events are looming on the horizon that can have a big impact:
- A lot of technological progress is expected and things such as growth of decentralised exchanges, performance improvements, proof of stake adoption or lightning network of Bitcoin can shift market perceptions of where value will be generated long-term by a lot.
- Regulatory oversight is intensifying and it can play out in many ways — both very positive as well as negative scenarios are thinkable.
- Many tokens trading at very high valuations offer very little transparency and do not have much of a product to show — the markets may start paying more attention to quality soon.
- It seems that a lot of institutional capital is still on its way to enter crypto markets.
- Applications illustrating the potential of crypto may start getting more “mainstream” adoption (e.g. Ripple and Western Union).
All this volatility will be the financial background to the progress that entrepreneurs and technologists building products on this scene will make. And it is this progress that matters most for the long term, not the short term swings which will inevitably be grabbing headlines and emotions of many.
Bold prediction: Berlin will establish itself as the capital of crypto in Europe. Total crypto market cap will hit $2 trillion, at least temporarily.
More and more SaaS startups will turn to alternative sources of funding.
Over the course of the last couple of years, a mismatch has emerged between the expectations of VCs and SaaS startups looking for funding. Because of the fund sizes of Series-A-and-up VC funds, return expectations, and the power law distribution of exit proceeds, most VCs are only interested in companies with strong potential to become unicorns, i.e. to become worth $1B or more. As a result, most VCs are looking for SaaS companies with an annual ARR growth rate of 2.5–3x or more. On top of that, they want to see an addressable market worth several billions and a strong potential to create some kind of “moat”. All of this makes sense, but the truth is that the vast majority of SaaS companies doesn’t fit this profile.
At the same time — and this is a difference to many consumer Internet categories with stronger winner-takes-all characteristics — a lot of these “non VC compatible SaaS startups” can become profitable companies that will be able to grow revenue and profits 20–30% y/y for many years. My guess is that there are thousands of SaaS companies out there that would make for a relatively safe investment and a 10–15% IRR. In a world of enormous liquidity and historically low interest rate, this should draw the attention of investors whose return expectation and risk tolerance is different from venture capital. The obvious ones include PE funds and providers of debt finance. Another possibility would be small IPOs, which were popular back in the days but have become rare because of the increased costs of regulatory compliance. And then there is — of course — the elephant in the room, ICOs.
Bold prediction: In 2018 I will manage to convince at least one fellow P9er to become a vegetarian.