Author: Matthew Lynley

Facebook has a new job posting calling for chip designers

Facebook has posted a job opening looking for an expert in ASIC and FPGA, two custom silicon designs that companies can gear toward specific use cases — particularly in machine learning and artificial intelligence.

There’s been a lot of speculation in the valley as to what Facebook’s interpretation of custom silicon might be, especially as it looks to optimize its machine learning tools — something that CEO Mark Zuckerberg referred to as a potential solution for identifying misinformation on Facebook using AI. The whispers of Facebook’s customized hardware range depending on who you talk to, but generally center around operating on the massive graph Facebook possesses around personal data. Most in the industry speculate that it’s being optimized for Caffe2, an AI infrastructure deployed at Facebook, that would help it tackle those kinds of complex problems.

FPGA is designed to be a more flexible and modular design, which is being championed by Intel as a way to offer the ability to adapt to a changing machine learning-driven landscape. The downside that’s commonly cited when referring to FPGA is that it is a niche piece of hardware that is complex to calibrate and modify, as well as expensive, making it less of a cover-all solution for machine learning projects. ASIC is similarly a customized piece of silicon that a company can gear toward something specific, like mining cryptocurrency.

Facebook’s director of AI research tweeted about the job posting this morning, noting that he previously worked in chip design:

While the whispers grow louder and louder about Facebook’s potential hardware efforts, this does seem to serve as at least another partial data point that the company is looking to dive deep into custom hardware to deal with its AI problems. That would mostly exist on the server side, though Facebook is looking into other devices like a smart speaker. Given the immense amount of data Facebook has, it would make sense that the company would look into customized hardware rather than use off-the-shelf components like those from Nvidia.

(The wildest rumor we’ve heard about Facebook’s approach is that it’s a diurnal system, flipping between machine training and inference depending on the time of day and whether people are, well, asleep in that region.)

Most of the other large players have found themselves looking into their own customized hardware. Google has its TPU for its own operations, while Amazon is also reportedly working on chips for both training and inference. Apple, too, is reportedly working on its own silicon, which could potentially rip Intel out of its line of computers. Microsoft is also diving into FPGA as a potential approach for machine learning problems.

Still, that it’s looking into ASIC and FPGA does seem to be just that — dipping toes into the water for FPGA and ASIC. Nvidia has a lot of control over the AI space with its GPU technology, which it can optimize for popular AI frameworks like TensorFlow. And there are also a large number of very well-funded startups exploring customized AI hardware, including Cerebras Systems, SambaNova Systems, Mythic, and Graphcore (and that isn’t even getting into the large amount of activity coming out of China). So there are, to be sure, a lot of different interpretations as to what this looks like.

One significant problem Facebook may face is that this job opening may just sit up in perpetuity. Another common criticism of FPGA as a solution is that it is hard to find developers that specialize in FPGA. While these kinds of problems are becoming much more interesting, it’s not clear if this is more of an experiment than Facebook’s full all-in on custom hardware for its operations.

But nonetheless, this seems like more confirmation of Facebook’s custom hardware ambitions, and another piece of validation that Facebook’s data set is becoming so increasingly large that if it hopes to tackle complex AI problems like misinformation, it’s going to have to figure out how to create some kind of specialized hardware to actually deal with it.

A representative from Facebook did not yet return a request for comment.

Netflix nears a $150B market cap as its subscribers continue to balloon

Just last quarter Netflix passed a $100 billion market cap — and we might already be talking about it as a $150 billion company before too long with yet another big financial quarter that sent its stock soaring.

Netflix, again, beat out some expectations Wall Street held for the first quarter and provided a pretty good outlook for the next quarter as well, where it said it expected to add around 6.2 million new subscribers. In the first quarter, Netflix added 7.41 million new subscribers — around 2 million of them domestic and the rest internationally. The company continued to see some pretty strong streaming revenue growth, which was up around 43% year-over-year in the first quarter this year, to around $3.6 billion.

With all this, Netflix now has nearly 119 million paid streaming memberships — and it wasn’t all that long when Netflix finally said just over two years ago that it would begin opening up in hundreds of new countries internationally. The company’s shares are up around 6% in extended trading, sending its market cap up north of $140 billion. And all this subscriber growth, too, comes before we’re seeing a new tie-up with Comcast’s cable subscriptions that may end up driving that even more. As usual, Netflix expects to lose a ton of money and says it expects between -$3 billion to -$4 billion in free cash flow, but that’s usually not what investors are looking for.

One of the big questions Netflix still has right now is what kind of price tag it will carry as a tack-on to a Comcast subscription. Earlier this week, the companies announced that Comcast would bundle Netflix in to its cable subscriptions, offering yet another entry point for Netflix to ferret up potential consumers that haven’t quite cut the cord yet but still might be interested in Netflix’s content. Netflix normally carries a price tag of around $13.99, but the companies have not said what its price will be as part of a cable bundle yet.

Following Netflix’s last earnings report — which it, as you might expect, included some blowout subscriber numbers — the company rocketed past a market cap of $100 billion. Since then it’s only been an upward trend for Netflix, which prior to its first-quarter report was worth more than $130 billion. Despite increasing spend on original content, that subscriber number is still mostly where it gets its market value because it’s a forward predictor of its revenue.

Netflix late last year said it expected to spend between $7 billion and $8 billion on original content this year, a number that seems to periodically get an upward revision and is still a dramatic step up from 2017. The company in its report today said it expected to spend between $7.5 billion and $8 billion on original content, and expects that marketing and content spend to weight toward the second half of 2018.

But it has to continue to invest in original content because it is a way to attract new subscribers, and also because it’s content that it can more easily distribute across different geographies and itself has control of the rights and what happens to it. It relies on shows like Stranger Things or Altered Carbon to bring in new users, which then hopefully stick around and eventually help recoup the cost of those shows — and then the cycle starts anew.

Pinterest aims for products from the store down your street as its next big ad business

Pinterest is known for having, and promoting, a lot of business content. It’s actually a majority of the content, and it’s usually from some of the most well-known brands that feed into the kinds of sometimes dream-level wants and needs of its users.

And while Pinterest a majority of Pinterest’s potential is locked up there, the company has increasingly turned its gaze toward smaller and smaller businesses to try to entice users with local content — including that clothing store right down your street. That’s part of the reasoning behind Pinterest’s Propel program, which it started a year ago to work with small businesses that really either didn’t know what they were doing, or had just never done it before. In another step toward that goal, Pinterest has hired Matt Hogle to be the global head of small business.

Hogle spent 9 years at Facebook, working with small businesses and will be part of the effort for the company to try to find the right set of tools and strategies in place to appeal to small businesses as it starts to ramp it into a significant portion of its revenue. The number of small businesses on Pinterest has increased by around 50% year-over-year, the company said, and it looks to continue to refine a kind of hybrid strategy that mixes platforms and interactions with real people in order to entice those businesses. That’s important for, say, a local clothing store that only has one store and a limited online shop, but has products that would perform well on their own as content on Pinterest — and could quickly add revenue if they started advertising on it.

“We, as an ads business, know what customers’ business objectives are, and we capture that at the early stages,” Pinterest head of global sales Jon Kaplan said. “We know what they’re targeting, how they’re targeting, who, and we know the creative best practices. We should be able in the very near future to take all these elements and say, oh this is your objective, we’ll obfuscate all this complexity and hit a target [return on ad spend] you have. We’re not far. We’ve obfuscated a lot of the levers one can pull in Propel.”

Propel, for now, is adopting an increasingly popular human/digital approach for smaller businesses that are looking to advertise for the first time on Pinterest. If they have no experience whatsoever, or don’t even know what to do, Pinterest’s goal is to serve as a resource for best practices when it comes to creative content down to where to target it. Pinterest is increasingly rolling out more self-serve tools with more robust targeting and tracking, but the kind of small businesses — the sum of which could eventually account for a big chunk of its revenue — that Facebook has snapped up with the prospect of getting in front of the exact right people at the right time. The company says it will be rolling out the “promote” button, which allows advertisers to click a button on a pin to quickly spin up a campaign, to Australia, Canada, Ireland, New Zealand and the UK in the next coming weeks.

In the end, Pinterest still has 200 million monthly active users, which is absolutely dwarfed by Facebook’s billions of users. But at the same time, Pinterest may be able to capitalize on the good will that Facebook has torched in light of its massive privacy scandal in which information on as many as 87 million users ended up in the hands of a research firm without authorization or permission. Facebook can prove a return on spend, but it can’t at the moment prove that it’ll be able to keep doing the same things that get that return on spend now that the Internet is revolting against Facebook for its massive breach of trust. (Hogle, to be clear, joined before details on the Cambridge Analytica scandal began pouring in.)

Pinterest’s value to advertisers is that it’s able to capture a potential customer when they are just a user clicking (or tapping) around Pinterest looking for ideas. Whether planning life events or just looking for wardrobe suggestions, Pinterest is able to go to advertisers and say they can catch them in those discovery stages and then stick the right ad in front of them to get their attention. Then, the service can follow the user all the way from when they are actually interested in doing something, searching for what to do, and eventually saving that product — or buying it.

It’s that potential value for advertisers that’s taken it to a $12.5 billion valuation in its most recent financing round. But while Pinterest is probably still a sort of curiosity buy for larger advertisers and brands, the challenge has been to chase down the businesses that don’t have huge marketing budgets and might not even be considering Pinterest as a potential advertising platform. After all, the playbook for Facebook is robust and there are plenty of success stories, and the same is true for Google. But at the same time, that bespoke coffee shop down on Valencia Street in San Francisco may have products that line up perfectly for a Pinterest user that’s looking for a holiday gift down the line.

“Over the coming nine months, into 2019, I think we’re gonna continue to invest in ways that make it easy for our small businesses to grow with us,” Hodle said. “Small businesses are unique in the sense that they don’t have the time or energy or resources — the CEO might also be the HR person and the marketing person, and so on. We need to make it extremely lightweight and seamless. At the end of the day, if we can’t provide value and deliver on the money they spend with us and demonstrate the value that’s created there — sales being that number one performance indicator for these companies — then we’re failing. The things we’re gonna continue to build are ways to make that process as easy and seamless.”

“Small businesses add disproportionate value to Pinners and our ad ecosystem,” Hogle said. “I truly believe that small businesses are part of the fabric of every community, they’re the engine that drives every economy around the world. Their relevancy, category or vertical, or proximity to consumers creates disproportionate organic value to Pinners and also creates disproportionate commercial value. Our goal is to show the people the most relevant ad they possibly can based on what they’re trying to accomplish. We can’t do that if we do not meet the needs of the vast majority of businesses that exist. It is a strategic decision, but it’s not just one that goes into paid ads, it creates disproportionate value for the entire platform and ecosystem.”

Hogle’s conversations started mid last year with Tim Kendall, Pinterest’s former president who left in November last year. What started as an exchange of ideas turned, as these conversations often do, it discussions about a potential job at Pinterest, and finally months later he ended up joining to start helping with the company’s small business efforts. Pinterest is increasingly looking to staff up with a suite of executives that will help it get its business in order ahead of a potential IPO. That includes a new COO, Francoise Brougher, who joined in February from Square.

But if Pinterest is going to eventually go public, and get its employees and investors paid out for their efforts, it needs to show that it can be a business that’s beyond just a curiosity budget for a big brand. Getting small businesses on board, like the ones Hogle looks to capture that are right down the street, are a big part of what the company hopes will eventually show that it has a diversified revenue stream and not beholden to just the big and potentially fickle budgets of larger brands.

“Our ad platform is not very old by industry standards, but the rate of development on our self serve tools and the rate of development on interfaces for our small customers is moving at a place where I’m really pleased,” Hogle said. “The ability to carve out opportunity for advertisers is something that’s moving really quickly. Is it where it needs to be long term, of course not, but we’re gonna continue to invest. One of the reasons I joined was it was very clear to me that small businesses were a priority. We are going to invest in products and we’re also going to invest in educational programs. We’re gonna invest and provide in the necessary resources.”

Google is acquiring GIF platform Tenor

Google will be acquiring Tenor, which powers a variety of GIF keyboards on phones and messengers like Facebook Messenger, the companies announced today.

Tenor will continue to operate as a separate brand within Google, the company said in a blog post. Tenor has increasingly positioned itself as a search company, using that as a metric for engagement and success as users tap into a massive database of GIFs. The company said it has more than 12 billion searches every month, and is one of the first major exits for a small but relatively hot space around tools that allow users to easily share GIFs. The company works with advertisers to create sponsored GIFs that slot into its searches, which are usually pretty compact and offer an opportunity to generate a lot of engagement.

GIFs have increasingly been pretty interesting because they offer an opportunity to compress a lot of information into something that’s easily shareable. Tenor CEO David McIntosh will often say that the company is about conveying emotion — and really, that isn’t something that often goes very well over text. If you’re watching the NCAA Men’s Basketball tournament, you’re probably better off searching for a GIF of your team rather than just blasting a text message to your group of friends.

“With their deep library of content, Tenor surfaces the right GIFs in the moment so you can find the one that matches your mood,” Google Images director of engineering Cathy Edwards said. “Tenor will help us do this more effectively in Google Images as well as other products that use GIFs, like Gboard. Tenor will continue to operate as a separate brand, and we’re looking forward to investing in their technology and relationships with content and API partners. So whether you’re using the Tenor keyboard or one of our other products, you can expect to see much more of this in your future:”

When you open Tenor, you’ll only find a small slice of GIFs that are available as the company is looking to compress the amount of time you actually spending digging around for a GIF you want to share. The theory is that if it’s easier to find and share one, you’ll do it again and again. This isn’t dissimilar from Google’s approach either, offering itself as a utility that’s a quick get-in, get-out experience that builds a level of stickiness that’s hard to unseat. Google is, of course, worth hundreds of billions of dollars off the back of a massive advertising business that basically prints money.

Tenor isn’t the only one in the space. Giphy, for example, also has a GIF keyboard and has a pretty large database of GIFs. Giphy says it has 300 million daily active users, though depending on who you talk to in the Valley that can mean a couple different things. Nevertheless, all of these companies have been able to attract venture financing. There’s also Gfycat, which positions itself as a tool for creators, that says it has 130 million monthly active users.

The terms of the deal weren’t disclosed. But by positioning itself as a search company that slots into a messaging ecosystem, Tenor seems like a natural piece of the puzzle for Google. It also gives the company a small wedge into the messenger space as it’ll have an opportunity to touch all the platforms that are connected to Tenor like even Facebook messenger, though that one tends to flip between GIF platforms indiscriminately.

Facebook CEO Mark Zuckerberg will reportedly testify before Congress

After declining a summons from a UK parliamentary committee that’s investigating how social media data is being used this morning, it appears that Facebook CEO Mark Zuckerberg may end up finally going before Congress to testify amid a wave of privacy debacles, according to a report by CNN.

The initial wave of this scandal began last week when it was revealed that Cambridge Analytica, a political consultancy, had acquired information on more than 50 million Facebook users via an app that accessed that information through the Facebook platform several years ago. While Facebook is not unfamiliar with privacy snafus, this appeared to be one of the final straws, forcing Facebook CEO to issue a sort of non-apology-ish apology — as well as take out full page (in wildly not-very-engaging full text form, weirdly enough) ads in several major newspapers.

CNN reports that Zuckerberg has come to terms with the fact that he will have to testify, and that the testimony may come within a matter of weeks amid pressure from lawmakers, the media, and the public.

Zuckerberg in particular has been a face absent in front of Congress, but it appears the pressure from this incident may be too much for the company to handle without getting his face out there in order to neutralize the fallout. Facebook offered CTO Mike Schroepfer and chief product officer Chris Cox to meet with the committee, but Zuckerberg again was absent here. Indeed, it was several days following the reports that Zuckerberg and COO Sheryl Sandberg made a formal appearance, as well as a number of interviews with various media outlets to try to explain themselves out of the situation

Last week, Zuckerberg said in one of those interviews that, “if it ever the case that I am the most informed person at Facebook in the best position to testify, I will happily do that.” Clearly, it appears that Congress wants the chief executive officer of the formerly-$500 billion advertising-driven business to be the one to do the explaining amid the increased fallout over the Cambridge Analytica debacle.

We reached out to Facebook for a comment and will update the post when we hear back.

Pinterest is slowly rolling out its automated shopping ads to more marketers

Pinterest is looking to continue to increase its portfolio of ads, though sometimes that can take a little while to see the light of day — and that includes a new-ish tool called Shopping Ads that’s slowly getting opened to more developers.

Getting new ad formats is important for a smaller company looking to build out an advertising business, as it has to show potential advertisers it can offer an array of tools to play with as they experiment with that service. The company said today that it’s expanding those shopping ad tools to hundreds of additional advertisers after launching a pilot program last year as it looks to continue to ramp up that tool. Pinterest has to be able to convince marketers that it should be a mainstay advertising purchase alongside Facebook and Google, which are able to routinely show returns in value for their advertising spend.

Shopping ads automatically create promoted pins from an existing product feed for a retailer. That means it’s basically one less thing for retailers to worry about as they add more and more content to the service. Most of Pinterest’s content online is business content as users share products they might be interested in one day buying or already own. As Pinterest gets more and more data on this, they’ll have a better handle on what ads work best, and hope that businesses will hand off the process in full to something more automated.

Pinterest hopes to capture that routine user behavior of planning what they want to do next, whether that’s an outfit to wear that day or some kind of major event or purchase down the line. Getting a hold of those users in the moment they might be interested in a new product is key to the company’s pitch to advertisers. You can more or less consider this a continued test as the company starts to slowly give the tool to the advertisers it works with before it becomes generally available. If it works, it could probably end up down the line in the hands of all advertisers, which could help for small- to medium-sized businesses without a lot of experience build out their early marketing campaigns.