cross pond high tech
116.6K views | +1 today
Follow
 
Scooped by Philippe J DEWOST
onto cross pond high tech
Scoop.it!

Facebook Now Controls Nearly One-Fifth Of U.S. Mobile Display Ad Revenue

Facebook Now Controls Nearly One-Fifth Of U.S. Mobile Display Ad Revenue | cross pond high tech | Scoop.it
Facebook has vaulted past its competitors to control 18.4 percent of U.S. mobile display ad revenues.
According to eMarketer, that means Facebook will end the year as the top U.S. publisher in mobile display, with roughly $340 million in revenue for all of 2012.
The chart below presents a visualization of Facebook remarkable performance. The social network wasn't in the picture in 2011. One year later, it has seized a fatter share of U.S. mobile display revenues than heavyweights such as Google, Pandora, and Apple's iAd platform. In the process, a greater share of revenue has accrued to the top six publishers.
However, Facebook's accomplishment needs to be placed in context. First, mobile remains a blip in the total advertising landscape. And within mobile, display still accounts for less spend than search ads— 46 percent to 49 percent of U.S. mobile ad spending. (In other countries the picture is even more lopsided in favor of search.)
Finally, it's early days yet for Facebook's mobile ads, and there are no guarantees the company's mobile ad formats will continue to succeed indefinitely.
Philippe J DEWOST's insight:

Still haven't clicked on any of these ads...

more...
No comment yet.
cross pond high tech
light views on high tech in both Europe and US
Your new post is loading...
Your new post is loading...
Scooped by Philippe J DEWOST
Scoop.it!

Open Compute Keynote de Philippe Dewost a USI 2015 Unexpected Sources of Inspiration - Philippe Dewost's light sources

Open Compute Keynote de Philippe Dewost a USI 2015 Unexpected Sources of Inspiration - Philippe Dewost's light sources | cross pond high tech | Scoop.it

L'édition 2015 de la Conference USI (Unexpected Sources of Inspiration) s'est tenue début Juillet au Carrousel du Louvre et a confirmé son statut de meilleure Conference Tech sur Paris, combinant un site exceptionnel, une audience de plus de 1200 participants très diverse et de très bon niveau, et une palette de speakers dans plusieurs disciplines délivrant une quarantaine d'interventions sur les deux jours. La Caisse des Dépôts était de nouveau partenaire de cette édition avec plus d'une cinquantaine de participants dans le cadre de son programme de transformation digitale.

François Hisquin, CEO d'Octo et Curateur de la conférence, m'avait demandé d'intervenir sur la thématique d'Open Compute, dont la contribution aux enjeux de la transition énergétique n'est toujours pas correctement appréciée en Europe alors même que nous hébergeons des entreprises extrêmement talentueuses dans ce domaine.

Philippe J DEWOST's insight:

Keynote en video et slideshare pour ceux qui n'ont pas le son

more...
CHECy's curator insight, November 25, 2015 12:15 PM

Intervention de Philippe Dewost, Président d'honneur du CHECy, lors de la conférence USI 2015.

Scooped by Philippe J DEWOST
Scoop.it!

iPhone X: The Demo Gods Are Cheeky

iPhone X: The Demo Gods Are Cheeky | cross pond high tech | Scoop.it

The latest Apple product unveiling, with its overly-reported Face ID glitch, reminds us that the demo isn’t the product. That can be good or bad.

It’s Full Moon over Cupertino. iPhone X specs leaks ahead of the official presentation and kommentariat inmates howl in their cages. This isn’t new, we’ve long known how psychotoxic Apple products can be, but the phenomenon seems to be reaching a new paroxysm. A few choice examples, starting with the grand prize [no links, no feeding the master baiters]:

The iPhone X proves the Unabomber was right
Steve Jobs gave us President Trump
Apple’s Face ID Could Be A Powerful Tool For Mass Spying

But let’s not lament these and many similar howlers. For Apple, they have a positive side: They attest to the power of the brand, to the magnetism of its products. Well…most of them. I maintain a list of Apple products that remain unmagnetized and fantasize that someday I’ll have a quiet conversation with the DRIs (“Direct Responsible Individual” in Apple’s parlance) in charge of these warts.

But we’ll leave the somber jeremiad for a different day.

The sun is out here in Paris; tomorrow I head to Vézelay and the start of a two day walk along the Camino de Santiago (the pélerinage de Compostelle as it’s known here). So, let’s have fun with Craig Federighi’s iPhone X FaceID demo glitch and, more generally, with the Demo genre. (The full two-hour September 12th keynote video is here.)

Craig Federighi, Apple’s Senior VP of Software Engineering — affectionately known as Hair Force One — is an infectiously happy, eager demo-meister. But the demo gods challenged Federighi’s disposition and caused his first attempt to unlock the iPhone X with his face to fail, leaving him staring at the standard unlock code screen.

Fortunately, a back-up phone was at hand — a testament to Federighi’s lack of faith in his own work, or to years of demo hiccups. The stand-in phone worked as expected, and the demo proceeded without further divine retribution.

Philippe J DEWOST's insight:

Key Notes on Apple Keynotes

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Autonomous Cars: The Level 5 Fallacy

Autonomous Cars: The Level 5 Fallacy | cross pond high tech | Scoop.it
A counterpoint by Jean-Louis Gassee to last week’s Alphabet/Google/Waymo SD Car winner-take-all thesis. Instead of a moonshot, we’ll see messy, helpful increments. It’s natural selection at work.Last week, I “explained” how Google was set to become the Microsoft of Self Driving (SD) cars. Machine learning leadership, sophisticated simulations and a Central Valley test site, immense computational and financial resources…the Google/Waymo endeavor has secured an insuperable advantage. You’re a car manufacturer: What choice do have but to pay a premium for a Google SD license? You could go with cheaper, less advanced technology…and say goodbye to your company after the first ‘mistake’ on the road.At $1,000 a car, the SD license will be Google’s Mother Lode 2.0, the dream of any tech company looking for a second growth wave.I concluded the article by promising to temper my enthusiasm and to offer a counterpoint, starting with a Horace Dediu quote:“Those who predict the future we call futurists. Those who know when the future will happen we call billionaires.”It’s one thing to predict, as Gordon Moore did in 1965, that semiconductors would double their computing power every 18 months (Moore’s Law, seemingly slower of late). It’s something else to know to invest in Apple’s 1980 IPO (+30,797%) or, even better, Microsoft’s 1986 stock market launch (+73,293%).It feels good to predict the emergence of SD cars: It has to happen because it’d be cool if it did. But when? Betting your company on even an approximate estimate of “Level Five: Full Automation” is a risk that you don’t want to take.
Philippe J DEWOST's insight:
Jean-Louis Gassée's (@jlg) view on #Level5 autonomous driving is refreshing. @WeAreLeonard think that smart infrastructure might help both accelerate and secure deployments.
more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Open Hardware is about to radically flip both server & data center markets with huge amounts of certified, refurbished facebook servers available here and now

Open Hardware is about to radically flip both server & data center markets with huge amounts of certified, refurbished facebook servers available here and now | cross pond high tech | Scoop.it

Open Hardware is touted to propel 80% of Facebook Data Center infrastructure, saving them 2Bn$ in the process.

This market, triggered and until recently dominated by hyperscale customers, is growing 30% year on year while the OEM server market declines (-6% y/y).

As Facebook is about to upgrade thousands of OCP machines to keep up with their gigantic CPU power needs, they are expected to offload the previous generation on the market, with unbeatable price/quality ratios as these natively rugged machines will be refurbished, tested, certified before being delivered.

To keep costs at minimum, the sales process is performed 100% online, and the delivery operated by a long standing OCP partner : Horizon Computing Solutions is not only the only distributor of Open Hardware Technology ; they also completed the first Open IT Hardware Product with RuggedPOD, the most power efficent Data Center Worldwide with a PUE = 1.0

Want to test a 16 core Bi-Xeon server ? It will cost you less than $1000 and you can order here.

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Fiction: Who Killed Windows Phone?

Fiction: Who Killed Windows Phone? | cross pond high tech | Scoop.it

How could Microsoft’s Windows Phone licensing business model stand a chance against Google’s Free and Open Android? None of the Redmond giant’s complicated countermeasures worked, its smartphone platform is dead. And yet, inexplicably, Microsoft failed to use a very simple move, one we’ll explore today.

Just back from three weeks in the Country of Good Sin’s heartland, I see Microsoft’s fresh and well-received Fourth Quarter Fiscal Year 2017 Results. The numbers acknowledge what was already notorious: Windows Phone is dead.

“Phone revenue was immaterial and declined $361M.”

This doesn’t come as a surprise. Despite Microsoft’s strenuous efforts to breathe life into its smartphone platform and devices, Windows Phone had been on an inexorable downward slope for several years, confirming a Horace Dediu theorem[as always, edits and emphasis mine]:

“As far as I’ve been able to observe, any company in the mobile phone market that ended up losing money has never recovered its standing in terms of share or profit.”

Let’s recall that, in September 2010, Redmond employees held what CNET called a “tacky ‘funeral’” for iPhone and Blackberry. One wonders how they’ll memorialize Windows Phone.

The gross failure of what once was the most powerful and richest tech company on the planet led to a search for a platform killer. Detectives didn’t think they had to go far to nab a suspect: Android. Microsoft’s Windows Phone was murdered by Google’s smartphone OS. How could Redmond’s money-making software licensing business model survive against a free and open source platform? Case closed.

No so fast.

Microsoft’s smartphone troubles started well before the birth of Android. In a reversal of the famous dictum Victory Has Many Fathers But Defeat Is An Orphan, Windows Phone’s collapse seems to have had many progenitors deeply embedded in the company’s decades-old culture.

But before we look at facts, let’s engage in a bit of fiction, let’s imagine Microsoft decides to fight Android on Google’s turf. In this alternate reality, Microsoft easily kills Android with one simple headline:

Windows Phone Now Free

The rest of the pitch writes itself.

Philippe J DEWOST's insight:

"it's the culture, stupid" - Even if it is easier to explain the past than act in the present to shape or avoid a future, this excellent piece by JLG reminds us all of how blinding corporate culture can be. Remembering that, in 2007, Nokia was the star according to the famous MBA mantra known as the "BCG Matrix" (#1 with twice #2's market share in a 2-digit growing market), that everybody knew that handset design should be segmented, and that mobile handset manufacturing had required decades of mastership and could not be improvised by a consumer electronics newcomer...

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

[FrenchWeb INSIDERS] Les 6 infos Tech qu'il ne fallait pas manquer

[FrenchWeb INSIDERS] Les 6 infos Tech qu'il ne fallait pas manquer | cross pond high tech | Scoop.it

Avec, entre Jeff Bezos et Jean-Marie Messier

 

Nous avons appris le départ de Philippe Dewost de la Caisse des Dépôts début septembre. Il occupait jusqu'à présent les fonctions de Directeur adjoint de la Mission «Programme d'Investissements d'Avenir» où il pilotait le secteur de l'économie numérique (enveloppe initiale de 4,25 milliards d’euros).

Philippe Dewost a commencé sa carrière professionnelle chez France Telecom comme co-fondateur de Wanadoo en 1995. Il a occupé des rôles de dirigeant dans plusieurs startups (Ukibi, Realeyes3D) avant de rejoindre en 2009 ImSense, une start-up de Cambridge spécialisée dans le traitement d’images, acquise par Apple en 2010 et dont la technologie eye-Fidelity est désormais utilisée 1/2 milliard de fois chaque jour.

L’ex pilote du rapport French Tech en 2013 rejoindra début Septembre un Groupe du CAC40 dont le nom n’a pas encore été révélé.

Philippe J DEWOST's insight:

Escape Velocity = vitesse de libération

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Plébiscite boursier pour Balyo, champion de la robotique industrielle !

Plébiscite boursier pour Balyo, champion de la robotique industrielle ! | cross pond high tech | Scoop.it

Le spécialiste de la robotisation des chariots de manutention a réussi son introduction en Bourse. Il a levé 39,8 millions d’euros et le titre cote déjà 28% au-dessus du prix d’introduction.

Les robots ont le vent en poupe. Le succès de l’introduction en Bourse de Balyo vient encore le confirmer. La petite société française a en effet développé une solution de robotisation visant à rendre autonome les chariots de manutention.

Alors que le groupe a enregistré un chiffre d’affaires de «seulement» 5,2 millions d’euros l’an dernier et que le seuil de rentabilité n’est pas attendu avant 2019, il vient de lever près de 40 millions d’euros en Bourse. Un montant qui pourrait même être porté à 45,8 millions d’euros en cas d’exercice intégral de l’option de surallocation.

L’offre a été sursouscrite près de sept fois, autorisant la société à fixer un prix d’introduction à 4,11 euros par action, tout en haut de la fourchette 3,05 - 4,11 euros proposée en amont. La «clause d’extension» a, évidemment, été exercée en totalité.

Près de dix fois le chiffre d'affaires

Et le plébiscite des investisseurs ne s’arrête pas là. Le titre a débuté sa cotation le 9 juin, uniquement sous la forme de «promesses d’actions», et, à 13h30, il cotait déjà 28% au-dessus du prix d’introduction. Le tout dans un volume de marché très important, dépassant le million de titres échangés.

Balyo est actuellement valorisé à 148 millions d’euros, soit 9,9 fois le chiffre d’affaires attendu cette année par les analystes d’ID Midcap. Ils attendent pourtant une véritable envolée des ventes en 2017, à 15 millions d’euros (5,15 millions en 2016).

Philippe J DEWOST's insight:

Une très belle sortie pour le Fonds Ambition Numérique, souscrit par la Caisse Des Dépôts et géré par sa principale filiale Bpifrance dans le cadre du Programme des Investissements d'Avenir.

Avec ce démarrage en trombe de la cotation, tout commence (bien) désormais ... Bravo à Fabien Bardinet et à toute son équipe !

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Algolia raises $53 million for its search engine API

Algolia raises $53 million for its search engine API | cross pond high tech | Scoop.it

French startup Algolia just raised a $53 million Series B round led by Accel, a couple of years after raising $18.3 million with… Accel also leading the round. So it looks like it’s a love story between the VC firm and the software-as-a-service startup.

The reason is quite simple. Algolia is still growing like crazy, with its annual recurring revenue doubling every year. Algolia’s goal hasn’t changed — the startup wants to provide the best search experience to everyone building websites and apps out there.

Jason Lemkin’s SaaStr fund, Jyoti Bansal, Clark Valberg and Des Traynor are also participating in today’s round, as well as existing investors Alven Capital, Point Nine Capital and Storm Ventures.

Just like Twilio is handling calls and text messages for you so you don’t have to, Algolia can also take care of the search box on your website. In just a few lines of code, you can replace your search feature with Algolia’s search.

You’ll need to feed Algolia with your search data and customize it to your needs, as well. After that, every time a user starts typing a letter, it’ll query one of Algolia’s servers and return results in no time. It’s quite remarkable to see how quickly and accurately it works.

And chances are that you’ve been using Algolia without realizing it. The company is managing the search feature on Medium, DigitalOcean, Twitch, Stripe, Periscope, Crunchbase and many others.

The company now has 3,000 customers and handles 25 billion searches every month. Recently, the startup launched a fallback version of Algolia called Algolia Offline that you can use if you’re in airplane mode or don’t have a strong signal. Algolia also optimized its search feature for places.

With today’s funding round, Algolia is going to open another office in London after Paris, San Francisco, New York City and Atlanta. And of course, the company wants to get new customers, build new features and maybe one day become the Stripe of search.

Philippe J DEWOST's insight:

La French Tech at its best !

Kudos to the Algolia team and Accel for this nice Series B

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Inside Facebook's plan to eat another $350 billion IT market

Inside Facebook's plan to eat another $350 billion IT market | cross pond high tech | Scoop.it
On an ordinary work day in mid-2016, a handful of Facebook engineers were sitting on the couches in a corner of the company's Menlo Park, California, headquarters when one of them tossed out a wacky idea.He suggested doing something that had never been done before and could potentially upend the $350 billion telecom market."It can't be so difficult to build our own system," the engineer said, referring to the telecom equipment that sends data across cables and wireless networks, and which the engineer suspected could be made to operate faster and cheaper than the pricey equipment sold by big vendors like Nortel, Huawei, Ericsson, Cisco or Juniper Networks.The engineer was suggesting building the telecom industry's first "white box" transponder, made with off-the-shelf parts such as chips from Broadcom and Acacia Communications, optical equipment from Lumentum, and software from one of the many new networking startups cropping up these days.Facebook's director of engineering Hans-Juergen Schmidtke, who was among those on the couch that day, was at first a naysayer."I was a little bit skeptical about it at the time," he recounted to Business Insider. As a former engineer at Juniper Networks, Schmidtke knew from experience that building telecom equipment systems was an expensive undertaking that involved hiring teams of specially trained engineers and sizeable R&D budgets. "Building a system ten years ago was like building a new company," Schmidtke said. Still, Schmidtke agreed to help this tiny group hack together a white box system at one of Facebook's famous hackathons. Three months later they had a working prototype. Six months later, on November 1, they announced it to the world as a real product called Voyager. The product’s unveiling sent shockwaves through the telecom industry, putting gear makers on notice that the lucrative market they controlled for decades was about to get turned upside down — and not necessarily to their advantage.While the effort is essentially a side-project for Facebook, a social networking company whose bread-and-butter business is online advertising, the stakes could not be higher for the telecom equipment companies which risk seeing their specialized products reduced to interchangeable commodities and their influence diminished. For the industry’s established companies, there’s unease about Facebook’s growing clout and its ultimate intentions. But in a sign of how serious Facebook’s foray is being taken, there’s already a recognition by some that the repercussions could be even more painful it they don’t adapt.Voyager has already been tested by Facebook and European telecom company Telia over Telia’s thousand-kilometer-telecom network. Plus, German telecom equipment maker ADVA Optical Networking is manufacturing the device and, as of a few weeks ago had nine customers trying it out for their telecom needs, a mix of big telecom companies and enterprises, it said. And Paris based telecom provider, Orange is also testing the device, working with Equinix and African telecom company MTN."We pulled it off essentially showing that when a few engineers can build a system within six months, the world has changed," Schmidtke said.One person told us that Schmidtke, who is insanely proud of Voyager, has become a star in his own corner of the tech world. When he and his team "go to conferences, they treat him like a tech celebrity, like a rock band," that person said.
Philippe J DEWOST's insight:
Fantastic read for any telecom engineer who wants to understand how a handful of determined engineers, when properly managed, can turn tables in a few months and displace a complete industry.Open Hardware is no longer a weak signal and has morphed into a powerful trend about to become a shockwave.To achieve such irony, very few people have realized that this game changer may be a fantastic opportunity for Europe, and that some core TIP players are totally unknown and unrespected French companies...
more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Intelligence artificielle : un logiciel abat en 1s le travail que des avocats font en 360.000 heures

Intelligence artificielle : un logiciel abat en 1s le travail que des avocats font en 360.000 heuresLa toute nouvelle intelligence artificielle de JPMorgan est l’employé du mois : elle peut résoudre en une seconde des arbitrages financiers qui prennent normalement quelques 360.000 heures de travail (soit une année) aux avocats de la firme. Le travail de cette machine artificielle, baptisée COIN, est d’interpréter les accords de prêts commerciaux. Et fait en prime nettement moins d’erreurs que les humains.
more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Renault reprend les activités de R&D françaises d'Intel dans le logiciel embarqué

Renault reprend les activités de R&D françaises d'Intel dans le logiciel embarqué | cross pond high tech | Scoop.it
En s'emparant des activités de R&D dans le logiciel embarqué basées en France d'Intel, leader mondial des semi-conducteurs, la marque au Losange réalise le doublé. Le constructeur accélère clairement dans la voiture connectée, et redore son blason dans son pays d'origine où il vient sauver des emplois.
Philippe J DEWOST's insight:
Le retour de la Deep Tech dans le viseur des Corporates !
more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Ector lève 2 millions d’euros pour déployer son service de voituriers à l’international

Ector lève 2 millions d’euros pour déployer son service de voituriers à l’international | cross pond high tech | Scoop.it
La start-up Ector, qui propose un service de voiturier à la demande dans les gares et les aéroports, a bouclé un tour de table de 2 millions d’euros. Le fonds Ville de Demain, géré par Bpifrance dans le cadre du Programme d’investissements d’avenir (PIA), 123 Investment Managers, ainsi que Via-ID, un fonds de développement de Mobivia Groupe (Norauto, Midas…), et Day One Entrepreneurs & Partners ont participé à l’opération. En décembre 2015, la jeune pousse avait bouclé un premier tour de table de 500 000 euros.Le marchéFondée en 2013 par Manoël Roy, Jonathan Grandperrin, Charles Raymond et Thomas de Longvilliers sous la marque Smart Park, Ector a développé un service de prise en charge des véhicules dans les aéroports et les gares. Les utilisateurs réservent la prestation en ligne sur la plateforme Web ou l’application mobile. A leur arrivée au dépose-minute de la gare ou l’aéroport, un voiturier récupère la voiture pour la stationner dans un parking de proximité, où elle peut bénéficier de services d’entretien (nettoyage, plein d’essence, contrôle technique…). Au retour du voyageur, la voiture est restituée au dépose-minute pour gagner du temps. Lancé fin 2015 à Orly et Roissy Charles-de-Gaulle, le service dessert désormais 13 hubs (8 aéroports et 5 gares) en France et en Europe. Sur son segment d’activité, Ector doit faire face à des acteurs comme Drivy, TravelerCar et OuiCar, qui proposent la location de voitures entre particuliers, ou encore Stanley Robotics, qui a conçu un robot-voiturier capable de garer des voitures dans les parkings.Les objectifsCe nouvel investissement doit permettre à la start-up parisienne de consolider sa position sur le marché français et de mettre les voiles à l’international. Dans ce sens, elle prévoit de renforcer ses effectifs avec des profils commerciaux, techniques et opérationnels. Prochainement, la société lancera son service dans de nouvelles villes, à commencer par Marseille et Bordeaux.
Philippe J DEWOST's insight:
Un nouvel investissement du Fonds "Ville de Demain" souscrit par la Caisse Des Dépôts et géré par sa filiale Bpifrance dans le cadre du Programme d'Investissements d'Avenir.Ravi pour eux à la fois comme client et comme souscripteur du Fonds !
more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Salesforce created an algorithm that automatically summarizes text using machine learning

Salesforce created an algorithm that automatically summarizes text using machine learning | cross pond high tech | Scoop.it

This year, people are expected to spend more than half their day reading e-mail, articles, or posts on social media, and it’s only going to get worse. To help solve this problem, researchers at Salesforce have developed an algorithm that uses machine learning to produce “surprisingly coherent and accurate” summaries according to MIT Technology Review.

Automatic summarization would be a particularly useful technology for Salesforce, which produces a variety of customer-service focused products. The company notes that the resulting summaries could be used by sales or customer service representatives to quickly digest e-mails and information, which would allow them to spend more time focused on their customers.

To that end, Salesforce is turning to machine learning to find ways to summarize longer blocks of texts, which it could eventually incorporate into its products. The company announced that it made two breakthroughs in natural language processing, introducing a new, “contextual word generation model,” and a “new way of training summarization models.” Together, the two advances allow researchers to automatically create summaries of longer texts that are accurate and readable. The company acquired a deep learning outfit MetaMind last year, which was behind the research.

Automatic text summarization works in two ways: extraction or abstraction

The researchers explain that automatic text summarization works in two ways: extraction or abstraction. With extraction, computer can draw from pre-existing wording in a text, but it’s not very flexible. Abstraction allows the computer to introduce new words, but the system has to understand the original article enough to be able to introduce the right words.

This is where deep learning neural networks come into play. They processing numerous examples of sentences and words to spit out new representations of each phrase, which allow the system to interpret texts and introduce its own words. The researchers let their model to look back at the text it’s working off of for additional context. It also looks back at earlier generated examples, to ensure that it’s not repeating itself.

Salesforce’s approach uses two teaching methods: teacher forced and reinforcement learning

The other breakthrough concerns how the researchers train the system to learn and improve itself. They used two approaches: teacher forcing and reinforcement learning. Reinforcement learning is a method that draws inspiration from how animals learn, and been used to teach Google’s DeepMind how to play video games. In this instance, the model is allowed to generate a sequence of words, and the result is then scored with an automated evaluation metric known as ROUGE (Recall-Oriented Understudy for Gisting Evaluation). The algorithm updates itself with higher scores, leading to better outcomes with future summaries. Teacher forcing is when the results are scored word by word off of an established reference, which provide “very decent results,” but which doesn’t allow for much flexibility.

Researchers found that “ROUGE-optimized RL helps improve recall...and word level learning supervision ensures good language flow, making the summary more coherent and readable.” Scored against this system, they found that their joint model scored higher than other approaches, and Richard Socher, Salesforce’s chief scientist, noted that he didn’t think that he’d ever seen “such a large improvement in any [natural-language-processing] task.”

The results are pretty astonishing: the researchers provided several examples, showing the original article, a human-generated summary, and a summary generated by their own model, and in each case, the summaries are considerably shorter than the original text, but contain the essentials in a readable form. Despite their advances, there’s still considerable work to be done in this field: MIT Technology Review spoke with Kristian Hammond, a professor at Northwestern University, who noted that the advance “shows the limits of relying purely on statistical machine learning,” but that it’s a step in the right direction.

Philippe J DEWOST's insight:

In 2 words : really impressive

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

What does Google want with HTC's smartphone business?

What does Google want with HTC's smartphone business? | cross pond high tech | Scoop.it
Google has announced it’s acquiring a $1.1bn chunk of HTC’s smartphone business, and with it providing the once leading Taiwanese phone brand a much needed lifeline. But what does Google want with part of a smartphone business?Google isn’t buying the whole of HTC, just a relatively large part of the Taipei-based company’s smartphone business and not its Vive virtual reality headset business. Google gains half of HTC’s research and development team – about 2,000 people – and a non-exclusive license for HTC’s intellectual property, allowing it to take advantage of some of HTC’s advances in smartphone technology.HTC gets a cash injection, which will help it survive in some very competitive markets, and Google gets to continue its “big bet on hardware” according to Rick Osterloh, the company’s senior vice president for hardware.It’s “a business decision to have access to one of the best R&D teams”, said Neil Shah, research director at Counterpoint Technology Market Research. But it’s also “a sort of emotional decision to save its close partners”.Little history of hardwareWhile Google is the creator of the Android operating system, which is now used on more than 2bn devices a month, or 89% of mobile devices according to IDC, it has only dabbled with making its own smartphones and tablets. It routinely partnered with firms such as HTC, LG and Huawei to make the Nexus series of a devices, which sold in low volumes and acted as showcases for each new version of Android.Google bought Motorola in 2011 for $12.5bn (£9.24bn), and while it ran it as a separate company selling smartphones aimed at the low end, the acquisition was really about a large stock of important patents.“Its main reasoning was to acquire Moto’s patent portfolio so as to protect against Apple (and Microsoft) while also providing stiffer competition to Samsung (although Google would never admit this),” said David McQueen, research director at ABI Research.Google sold Motorola to China’s Lenovo in 2014 for $2.9bn without the collection of patents.
Philippe J DEWOST's insight:
Hardware Is Not Dead. Talent is what matters.
more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Tesla's Hurricane Irma Update Taps Into Our Deepest Fears Of 21st Century Driving

Tesla's Hurricane Irma Update Taps Into Our Deepest Fears Of 21st Century Driving | cross pond high tech | Scoop.it

Earlier this week, Tesla remotely upgraded select Florida Tesla owners’ cars to expand their mileage capacity in an effort to ease and assist with Hurricane Irmaevacuation efforts. The move was praiseworthy and appropriate, but at the root of the gesture lies a terrifying prospect of our automotive future.

 

Tesla briefly sold a 60 and 60D trim level of its Model S and Model X vehicles. These models had 75 kWh battery packs installed, but were software limited to have less range to artificially create a more affordable entry-level tier for buyers. Buyers still had the option to upgrade to the full capacity for a charge if they changed their minds, and Tesla would “unlock” the capability via an over-the-air software update.

 

With category four Hurricane Irma headed straight for Florida, Tesla unlocked the full capacity of 60 and 60D model owners in Florida to give them about a 30 mile range boost while evacuating. It was genuinely helpful and an extremely savvy public relations move for the company.

 

But what it also previewed is our imminent future of unprecedented corporate control over how we drive and what we drive. I briefly mentioned it in the article yesterday, but it’s not hard to imagine a worst case scenario where a company or corporation becomes a critical decision maker in disaster scenarios, like with Hurricane Irma, out of consumer and government control in a critical moment.

 

Now, I’ve never been one to play into the fears of autonomous driving or ridiculous theories of car hacking, though I recognize vehicle computer security as one of the most important developments in the automotive industry going forward. But this issue is concerned with the relationship between the company and the company’s product.

 

What would happen if Tesla didn’t unlock the range of those cars? It’s not likely that any of the owners would become stranded, as Electrek reported most of Tesla’s charging network was still functional at the time. But that could have easily been the scenario, and then we face a situation where people were physically capable of evacuating sooner but limited by an option box they didn’t check. We now face a reality where we know our vehicles may hold more potential than we have access to, and that gets complicated in life or death scenarios

 

.../...

 
Philippe J DEWOST's insight:

Interesting story reminding us that, like a Patek Philippe, you don't fully own a Tesla ; unlike a Patek Philippe however, there are software upgrades in a Tesla and they are not only triggered with an extra payment...

 
more...
Emmanuel HAVET's curator insight, September 13, 8:16 AM
Cette mise à jour automatique de la part de Tesla met en lumière ce qu'il se passe déjà avec d'autres acteurs qui eux n'ont d'accès qu'à nos données. Et là, c'est la douche froide...
Scooped by Philippe J DEWOST
Scoop.it!

VINCI : Philippe Dewost appointed director of Leonard, the VINCI innovation and foresight platform - NASDAQ.com

VINCI : Philippe Dewost appointed director of Leonard, the VINCI innovation and foresight platform - NASDAQ.com | cross pond high tech | Scoop.it

Philippe Dewost has been appointed director of Leonard, VINCI's new Group-wide innovation and foresight platform. He took up his duties on 1 September and reports to Pierre Coppey, Executive Vice-President of VINCI.

Philippe Dewost, a graduate of the École Normale Supérieure (Ulm), is a Corps des Mines engineer and holds an MBA from the Collège des Ingénieurs. 

He is a co-founder of Internet service provider Wanadoo and notably re-launched Orange France Telecom's home devices business.

He also served as CEO of imSense ltd, a British startup based in Cambridge. The company, which was acquired by Apple in 2010, developed and patented one-of-a-kind imaging technologies that are now used more than 500 million times a day across most Apple products.

Starting in 2011, Philippe Dewost headed the Digital Economy section of the Investments of the Future Programme at Groupe Caisse des Dépôts (CDC). At the end of 2015, Philippe Dewost also launched LaBChain, a CDC Blockchain initiative.

Philippe Dewost was the prime mover behind La French Tech, the government's French startup ecosystem initiative, and is also Honorary President of the Centre des Hautes Etudes du Cyberespace (CHECy).

"Philippe has more than 20 years of experience in Internet and mobile ecosystems within large tech and telecoms groups and startups. This expertise will be of invaluable help in developing Leonard, the new VINCI open laboratory focused on the future of cities and infrastructure," said Pierre Coppey, Executive Vice-President of VINCI.

"With Leonard, VINCI is offering me a chance to contribute to the digital transformation of durable structures and move from bytes to concrete. I am honoured and proud to have this opportunity to work with them over the long term to write the digital record of the builders," said Philippe Dewost in taking up his new position.

Philippe J DEWOST's insight:

With Leonard, VINCI is offering me a chance to contribute to the digital transformation of durable structures and move from bytes to concrete. I am honoured and proud to have this opportunity to work with them over the long term in order to pave the digital trace of builders #BuildTech #ConcreteTech

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

A computer was asked to predict which start-ups would be successful. The results were astonishing

A computer was asked to predict which start-ups would be successful. The results were astonishing | cross pond high tech | Scoop.it

When a magazine challenged a technology company to use AI to pick 50 unheard of companies that were set to flourish, the experiment yielded dramatic results.

In 2009, Ira Sager of Businessweek magazine set a challenge for Quid AI's CEO Bob Goodson: programme a computer to pick 50 unheard of companies that are set to rock the world.

The domain of picking “start-up winners” was - and largely still is - dominated by a belief held by the venture capital (VC) industry that machines do not play a role in the identification of winners. Ironically, the VC world, having fuelled the creation of computing, is one of the last areas of business to introduce computing to decision-making.

Nearly eight years later, the magazine revisited the list to see how “Goodson plus the machine” had performed. The results surprised even Goodson: Evernote, Spotify, Etsy, Zynga, Palantir, Cloudera, OPOWER – the list goes on. The list featured not only names widely known to the public and leaders of industries, but also high performers such as Ibibo, which had eight employees in 2009 when selected and now has $2 billion annual sales as the top hotel booking site in India. Twenty percent of the companies chosen had reached billion-dollar valuations.

To contextualize these results, Bloomberg Businessweek turned to one of the leading “fund of funds” in the US, which has been investing in VC funds since the 1980s and has one of the richest data sets available on actual company performance and for benchmarking VC portfolio performance.

The fund of funds was not named for compliance reasons, but its research showed that, had the 50 companies been a VC portfolio, it would have been the second-best-performing fund of all time. Only one fund has ever chosen better, which did most of its investments in the late 1990s and rode the dotcom bubble successfully. Of course, in this hypothetical portfolio, one could choose any company, whereas VCs often need to compete to invest.

Recently, Bloomberg asked Goodson to repeat the feat. Here, we’ll take an in-depth look at the methodology behind the new list, and also broader trends set to flourish in the market.

Philippe J DEWOST's insight:

Rage against the machine. I suggest a close look to the list closing this post as one never knows...

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Following Italy and Spain, N26 will bring Apple Pay to France later this year

Following Italy and Spain, N26 will bring Apple Pay to France later this year | cross pond high tech | Scoop.it

We’ve been busy the past few weeks. First, we announced we would be bringing Apple Pay to Italy, and shortly after we announced we would bring it to Spain. Our customers in France started to raise some eyebrows. Our comments section was flooded with questions and requests, as it always is, which is why we’re excited to break the next piece of news.

Later this year, N26 will bring Apple Pay, which is transforming mobile payments with an easy, secure and private way to pay that’s fast and convenient, to customers in France. N26 is proud to bring Apple Pay to France. Supporting Apple Pay is another example of the innovation for which N26 is so well known.

Philippe J DEWOST's insight:

Another reason to consider N26 as one of the best #neobanks out there...

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Le plan radical de Nexedi pour couvrir les zones blanches en Très Haut Débit pour 200 M€ et en 12 mois seulement

Le plan radical de Nexedi pour couvrir les zones blanches en Très Haut Débit pour 200 M€ et en 12 mois seulement | cross pond high tech | Scoop.it

Communiqué de presse

Paris, le 16 juin 2017.

 

Nexedi offre à l'Etat de couvrir la France en très haut débit en moins de 12 mois pour 200 millions d'euros et sur la base de technologies exclusivement françaises dont l'ensemble du code source est libre ou souverain.

 

Comme le rapporte Numerama, un rapport de la Cour des comptes consacré au plan « France Très Haut Débit » a jugé très sévèrement la stratégie actuelle. Il considère notamment que les évaluations faites par le gouvernement Hollande pour raccorder la totalité des Français ne sont absolument pas réalistes. Ce n’est pas 20 milliards d’euros que le plan coûtera à la nation mais 34,9 milliards, soit 75 % de plus.

 

Le Président de la République considère que le plan Très Haut Débit actuel n'est pas réaliste en l'état. Si on veut raccorder tout le pays d'ici 2022, il faut réduire le rôle de la fibre optique et mettre davantage l'accent sur d'autres solutions, comme la 4G. Pour raccorder tout le territoire en fibre, à supposer que cela en vaille la peine, « il faudra des années, parfois des décennies », a-t-il prophétisé. Et comme le rapporte Nextinpact, le président de la République a déclaré qu'« il ne faut pas mentir aux gens ». La fibre ne sera pas rapidement déployée « jusqu'au dernier kilomètre, dans le dernier hameau », du moins pas avant des décennies.

 

Nexedi souscrit à cette analyse et propose de la mettre en pratique sous 12 mois avec du matériel et du logiciel "Made in France". Ce matériel déjà déployé par l'armée française et la SNCF permet d'obtenir un résultat meilleur pour un coût dix fois inférieur aux solutions d'infrastructure "Made in China" généralement utilisées par les opérateurs.

 

Il repose sur les produits d'un consortium de sociétés françaises (Air-Lynx, Amarisoft, AW2S, BJT Partners, Horizon Computing et Nexedi) qui constitue l'unique constructeur français d'infrastructures LTE (4G/future 5G) suite au rachat d'Alcatel et qui fait déjà l'objet d'un soutien de BPI France dans le cadre de son programme "Grands Défis du Numérique".

 

L'offre de Nexedi permet de réaliser un réseau 4G autonome à 300 Mbps ou 1 Gbps pour un investissement de 5000€ par site ou village et un coût récurrent annuel de moins de 500€ par site ou village. L'enveloppe de 200 M€ permet de couvrir de l'ordre de 20.000 sites et de les relier par une combinaison de liaisons radio ou satellitaires aux grandes infrastructures fibrées déjà en place ou à venir, tout en fournissant à l'armée une solution de communication résiliente (PMR).

 

Nexedi a démontré depuis 10 ans sa capacité à réaliser des infrastructures fiables à bas coût dans des domaines où les grands opérateurs nationaux échouent. Ainsi, Cloudwatt et Numergy ont bénéficié d'aides publiques considérables sans parvenir pour autant à livrer un cloud souverain compétitif. Au contraire, Nexedi a développé pour un coût 100 fois inférieur le logiciel de Edge Cloud "SlapOS" qui est désormais au coeur du succès du projet Teralab (hébergement Big Data souverain utilisé par de nombreux acteurs du CAC40). 

 

L'absence de couverture des zones blanches conduit à positionner la France entre l'Iran et le Kazakhstan en termes de pénétration de la 4G. L'absence de progrès dans le domaine du haut débit depuis 15 ans relève de blocages comportementaux des grands opérateurs qui préfèrent les infrastructures "Made in China" aux solutions sur base Amarisoft pour la couverture LTE, comme ils préfèrent pour leur business cloud les solutions "Made in China" à SlapOS, quitte à dépenser 10 à 100 fois plus qu'avec des technologies souveraines issues de PME françaises et déjà exportées.

 

L'offre de Nexedi est valable jusqu'au 14 juillet 2017.

Philippe J DEWOST's insight:

Une approche radicale, non-conventionnelle, très "deep tech" de la couverture Très Haut Débit et utilisant des solutions 100% françaises, à mon avis cela se regarde et se teste !

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Softbank is buying robotics firms Boston Dynamics and Schaft from Alphabet

Softbank is buying robotics firms Boston Dynamics and Schaft from Alphabet | cross pond high tech | Scoop.it
Here’s a surprise turn of events: Softbank — maker of the friendly Pepper robot and a major M&A player in the tech world — has just announced that it is acquiring two more robotics companies from Google owner Alphabet as part of its own deeper move into the field: it is buying Big Dog developer Boston Dynamics and the secretive bipedal robotics firm Schaft.A spokesperson for Softbank has confirmed to us that the terms of the deal are not being disclosed, but we will try to find out. In case you are wondering (we were), we have double checked and these are the only two of Alphabet’s robotics companies getting acquired by Softbank.There had been lots of murmurs about Google (and then Alphabet) wanting to offload Boston Dynamics for over a year now. Most recently, Toyota was apparently eyeing up an acquisition of the two businesses.But robots are huge across all of Japan, with some of the biggest innovations, investments, and motivations for developing them coming out of the country; and so, in the end, it looks like another titan of Japanese industry beat Toyota to the punch.“Today, there are many issues we still cannot solve by ourselves with human capabilities,” said Masayoshi Son, Chairman & CEO of SoftBank Group Corp., in a statement. “Smart robotics are going to be a key driver of the next stage of the Information Revolution, and Marc and his team at Boston Dynamics are the clear technology leaders in advanced dynamic robots. I am thrilled to welcome them to the SoftBank family and look forward to supporting them as they continue to advance the field of robotics and explore applications that can help make life easier, safer and more fulfilling.”While Softbank has been hard at work doubling down on big bets in areas that it believes will be at the center of the future of computing — other acquisitions have included its $24 billion acquisition of chip maker ARM Holdings — for Alphabet, this is part of the company’s bigger efforts to rationalise some of its many acquisitions and strategic bets over the years that have not panned out as great fits with the wider business.It looks like the whole team is coming over with the deal. “We at Boston Dynamics are excited to be part of SoftBank’s bold vision and its position creating the next technology revolution, and we share SoftBank’s belief that advances in technology should be for the benefit of humanity,” said Marc Raibert, CEO and founder of Boston Dynamics, in a statement.“We look forward to working with SoftBank in our mission to push the boundaries of what advanced robots can do and to create useful applications in a smarter and more connected world.”When Alphabet (still called Google at the time) acquired Shaft and Boston Dynamics in 2013 (Shaft was purchased as one of a group of seven acquisitions), the company did not disclose the terms of the deals.While Boston Dynamics has been pretty public in posting (sometimes terrifying) videos and generally talking about of its advances in making animal-like robots that can trek across all terrains and get up instantly when knocked over, Schaft has been a fairly quiet presence.Schaft revealed its first big prototypes only about two years after the acquisition. The company has been around since 2012, after being incubated in the JSK Robotics Laboratory at the University of Tokyo by co-founders Yuto Nakanishi, Junichi Urata, Narito Suzuki and Koichi Nishiwaki. It remains a secretive company: an attempted visit to its website today was blocked.
Philippe J DEWOST's insight:
Impressed with Softbank acquisition spree ; now completing their consumer robotics stake (France's Aldebaran Robotics) with a military arm , after snapping (mobile) chip design behemoth ARM ($24Bn). Vertical integration in sight ?
more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Orange, VCs Commit $113M to Network Startups as 'Black Box' Frustration Mounts

Orange, VCs Commit $113M to Network Startups as 'Black Box' Frustration Mounts | cross pond high tech | Scoop.it

Orange and four venture capitalist partners have promised to invest up to €100 million ($113 million) in telecom infrastructure startups over the next three to four years.

 

The funding will support startups challenging equipment incumbents like Ericsson AB (Nasdaq: ERIC), Huawei Technologies Co. Ltd. and Nokia Corp. (NYSE: NOK) as telcos embrace disruptive network technologies based on software, virtualization and open source code.

 

The news is the latest sign of Orange's desire to play a more influential role in the development of network technologies and reflects growing frustration with the traditional vendors.

It came as the French service provider said it would begin nurturing network startups for the first time in partnership with social media giant Facebook, which has also become more active in the networks sector over the past year.

 

The two companies already collaborate through the Telecom Infra Project, a Facebook-led initiative that was launched in early 2016 with a goal of more rapidly commercializing low-cost and innovative network technologies. (See Facebook: TIP Will Open Telecom Hardware.)

 

Four startups, chosen from a pool of 22 applicants, will receive support from Orange(NYSE: FTE) and Facebook and be invited to a TIP summit in California in November, where they will be able to meet operator members.

 

TIP now counts about 450 members including "all of the most important mobile operators" in the world, claimed Steve Jarrett, Facebook's head of infrastructure partnerships for Europe and the Middle East.

 

The winning startups are

  • Amarisoft, a developer of virtual radio access network technology,
  • Athonet, which specializes in mobile core "softwarization,"
  • Adipsys, whose systems are already helping Orange to manage WiFi hotspots, and
  • Horizon Computing, which claims to have made breakthroughs on reducing the costs of running data centers.

Orange Fab, the division that looks after all startup activities for the French telco, has launched a new program called Telecom Track to look after the startups from the network and infrastructure sector.

 

Startups will also be eligible to receive financial support from Orange Digital Ventures, the French operator's investment fund, as well as venture capital partners Iris Capital, Innovacom, Cathay Innovation and Breega Capital, although funding for the Telecom Track players is not guaranteed.

 

"Those who grow fast and strike partnerships and scale internationally will get the money," said Julien-David Nitlech, a partner at Iris Capital.

 

While €100 million ($113 million) may seem like a relatively small amount in the context of the broader network equipment market, the sudden willingness of venture capitalists to support new infrastructure players may alarm the established vendors.

 

"The market has historically not had a lot of venture financing," said Jarrett. "We hope to change that."

Facebook colleague Min Jun added: "We thought the investor community would need more convincing and we have seen major traction. Investors are saying we believe in this and will commit funds."

Facebook will not make any direct equity investments but says it is "conceivable" that it could acquire startups in the accelerator program.

By helping to reduce network costs, and making it easier to deploy networks in areas that currently lack connectivity, the social networking giant hopes to get more people online and using its services, it is widely assumed.

 

For Orange, the ultimate goal is a complete overhaul of the way it has traditionally built networks.

 

"We cannot be dependent on long standardization processes anymore," said Etienne Moreau, an investment manager at Orange Digital Ventures. "If we want to have control of our technology and be a software company we need to get into new technologies like white boxes and get rid of the black boxes we have sourced from equipment vendors."

With white boxes, telcos would run network software on commercial, off-the-shelf servers, instead of relying on the "black boxes" that combine proprietary hardware and software.

Orange has previously flagged its interest in white boxes and noted the challenge they pose to equipment suppliers such as Cisco Systems Inc. (Nasdaq: CSCO), which has come under pressure to adapt its own technology and business model accordingly. (See Orange Plots Mass Network-as-a-Service Rollout and Cisco Takes Bold Software Step to Counter White Box Threat – Report.)

 

Bertrand Rojat, the deputy vice president of Orange's Technocentre research-and-development unit, told Light Reading that, as well as forming a "commercial relationship" with startups, Orange was eager to make their technology available to some of its telco partners.

 

Besides other service providers involved with TIP, that could include the members of Go Ignite.

"That is an initiative between Orange, Deutsche Telekom, Telefónica and SingTel," he said. "If we have a startup that is good for one of us then it might be good for all of us."

Such moves could help the startup technologies to gain the scale they would need to be commercially viable in wide area networks.

Much like AT&T Inc. (NYSE: T) in the US, Orange appears to be stepping into roles that vendors have traditionally performed as it tries to seize control of network development and sever the ties that have previously bound it to a small number of big players.

 

Philippe J DEWOST's insight:

Extremely interesting post following Orange Fab's latest press release, as it delivers several key hints :

1/ It took Facebook and OCP (and then more recently TIP) for Telcos to finally realize that becoming a giant purchasing department and outsourcing infrastructure (and knowledge) to a handful of equipment vendors was not the only way. We are still very early but still, a few giants wake up.

2/ Open Source will redefine Hardware the same way it flipped Software ; only pending question is "Who will be the Red Hat of Open Hardware ?"

3/ #HardwareIsNotDead and Deep Tech is back with more and more VC's looking (back) into it !

Congratulations to Orange and Iris Capital for paving the way !

more...
Scooped by Philippe J DEWOST
Scoop.it!

An IBM Breakthrough Ensures Silicon Will Keep Shrinking

An IBM Breakthrough Ensures Silicon Will Keep Shrinking | cross pond high tech | Scoop.it

The limits of silicon have not been reached quite yet.

Today, an IBM-led group of researchers have detailed a breakthrough transistor design, one that will enable processors to continue their Moore’s Law march toward smaller, more affordable iterations. Better still? They achieved it not with carbon nanotubes or some other theoretical solution, but with an inventive new process that actually works, and should scale up to the demands of mass manufacturing within several years.

That should also, conveniently enough, be just in time to power the self-driving cars, on-board artificial intelligence, and 5G sensors that comprise the ambitions of nearly every major tech player today—which was no sure thing.

5nm Or Bust

For decades, the semiconductor industry has obsessed over smallness, and for good reason. The more transistors you can squeeze into a chip, the more speed and power efficiency gains you reap, at lower cost. The famed Moore’s Law is simply the observation made by Intel co-founder Gordon Moore, in 1965, that the number of transistors had doubled every year. In 1975, Moore revised that estimate to every two years. While the industry has fallen off of that pace, it still regularly finds ways to shrink.

Doing so has required no shortage of inventiveness. The last major breakthrough came in 2009, when researchers detailed a new type of transistor design called FinFET. The first manufacturing of a FinFET transistor design in 2012 gave the industry a much-needed boost, enabling processors made on a 22-nanometer process. FinFET was a revolutionary step in its own right, and the first major shift in transistor structure in decades. Its key insight was to use a 3-D structure to control electric current, rather than the 2-D “planar” system of years past.

”Fundamentally, FinFET structure is a single rectangle, with the three sides of the structure covered in gates,” says Mukesh Khare, vice president of semiconductor research for IBM Research. Think of the transistor as a switch; applying different voltages to the gate turns the transistor “on” or “off.” Having three sides surrounded by gates maximizes the amount of current flowing in the “on” state, for performance gains, and minimizes the amount of leakage in the “off” state, which improves efficiency.

But just five years later, those gains already threaten to run dry. “The problem with FinFET is it’s running out of steam,” says Dan Hutcheson, CEO of VLSI Research, which focuses on semiconductor manufacturing. While FinFET underpins today’s bleeding-edge 10nm process chips, and should be sufficient for 7nm as well, the fun stops there. “Around 5nm, in order to keep the scaling and transistor working, we need to move to a different structure,” Hutcheson says.

Enter IBM. Rather than FinFET’s vertical fin structure, the company—along with research partners GlobalFoundries and Samsung—has gone horizontal, layering silicon nanosheets in a way that effectively results in a fourth gate.

 

“You can imagine that FinFET is now turned sideways, and stacked on top of each other,” says Khare. For a sense of scale, in this architecture electrical signals pass through a switch that’s the width of two or three strands of DNA.

“It’s a big development,” says Hutcheson. “If I can make the transistor smaller, I get more transistors in the same area, which means I get more compute power in the same area.” In this case, that number leaps from 20 billion transistors in a 7nm process to 30 billion on a 5nm process, fingernail-sized chip. IBM pegs the gains at either 40 percent better performance at the same power, or 75 percent reduction in power at the same efficiency.

Just in Time

The timing couldn’t be better.

Actual processors built off of this new structure aren’t expected to hit the market until 2019 at the earliest. But that roughly lines up with industry estimates for broader adoption of everything from self-driving cars to 5G, innovations that can’t scale without a functional 5nm process in place.

“The world’s sitting on this stuff, artificial intelligence, self-driving cars. They’re all highly dependent on more efficient computing power. That only comes from this type of technology,” says Hutcheson. “Without this, we stop.”

Take self-driving cars as a specific example. They may work well enough today, but they also require tens of thousands of dollars worth of chips to function, an impractical added cost for a mainstream product. A 5nm process drives those expenses way down. Think, too, of always-on IoT sensors that will collect constant streams of data in a 5G world. Or more practically, think of smartphones that can last two or three days on a charge rather than one, with roughly the same-sized battery. And that’s before you hit the categories that no one’s even thought of yet.

“The economic value that Moore’s Law generates is unquestionable. That’s where innovations such as this one come into play, to extend scaling not by traditional ways but coming up with innovative structures,” says Khare.

Widespread adoption of many of those technologies is still years away. And success in all of them will require a confluence of both technological and regulatory progress. At least when they get there, though, the tiny chips that make it all work will be right there waiting for them.

Philippe J DEWOST's insight:

"I'm not dead. I'm getting better" - Moore's Law & the Holy Grail according to IBM

more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

L'attractivité de la France reprend du poil de la bête grâce à la French Tech

L'attractivité de la France reprend du poil de la bête grâce à la French Tech | cross pond high tech | Scoop.it
Start-up nation ?Deuxième atout, les décideurs internationaux classent la capacité de recherche et d’innovation au premier rang des atouts de la France, loin devant ses capacités touristiques et ses infrastructures. Un vrai changement d’appréciation."Ce n’est pas encore une image de start-up nation, mais c’est une perspective très positive", se félicite Marc Lhermitte.Un coup de chapeau à Fleur Pellerin, secrétaire d'Etat chargée de l'Economie numérique, qui a lancé la French Tech, ce label qui regroupe toutes les jeunes entreprises françaises technologiques. Flambeau repris par Axelle Lemaire et Emmanuel Macron, avec les encouragements de François Hollande. Ils se sont efforcés de promouvoir les start-up, mais aussi de rendre la fiscalité et les réglementations plus hospitalières.
Philippe J DEWOST's insight:
La French Tech en 2016: +30% d'investissements étrangers , +24% de créations d'emplois !Ces superbes résultats ne doivent pas nous faire oublier la prochaine étape qui d'ici environ 2 ans deviendra décisive : un marché d'exits (corporate et mises en bourse) suffisant pour générer de la liquidité, valoriser correctement les meilleures de nos startups, générer des retours satisfaisants pour tous les investisseurs (y compris les Fonds du PIA) et éviter incidemment que la French Tech ne deviennent l'open bar de nos amis américains et chinois.
more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

Stanley Robotics lève 3,6M€ pour déployer son robot voiturier à l'international

Stanley Robotics lève 3,6M€ pour déployer son robot voiturier à l'international | cross pond high tech | Scoop.it
La start-up parisienne Stanley Robotics vient d'officialiser une levée de fonds de 3,6 millions d'euros auprès d'Elaia Partners, de Bpifrance (via le fonds Ville de demain) et d'Idinvest Partners. Fondée en 2015, Stanley Robotics a développé un robot voiturier baptisé Stan. L'automate est le fruit d'une dizaine d'années de R&D, les fondateurs ayant travaillé précédemment dans un laboratoire de recherche dédié aux technologies des véhicules autonomes. L'objectif de cette augmentation de capital est de déployer sa solution à l'international.Testé à l'aéroport de Roissy-Charles-de-GaulleDepuis le mois de février dernier, Stan est expérimenté dans deux parkings de l'aéroport de Roissy. "Il suffit au conducteur de garer sa voiture dans un box dédié à l'entrée du parking, puis de s'enregistrer sur un automate à proximité. Une fois le véhicule isolé du public, le robot se met en position, repère les pneus, s'ajuste à la longueur de la voiture, la soulève, puis la transfère", expliquent nos confrères de L'Usine Nouvelle. Le système est par ailleurs connecté au billet d'avion du conducteur et connaît donc sa date de retour. De cette façon, le robot ramène le véhicule dans le box situé à l'entrée du parking en temps voulu.Stocker 50% de véhicules supplémentairesSi ce service de voiturier autonome facilite l'arrivée et le départ des voyageurs stationnant leur véhicule sur de longues périodes dans les aéroports, il propose surtout des avantages significatifs aux gestionnaires des parkings. Selon les créateurs de Stan, ce service autonome permettrait d'optimiser le stationnement des véhicules et ainsi de stocker 50% de voitures en plus.D'après la start-up, les avantages de sa solution ne se limitent pas au seul gain d'espace. Elle permettrait de repenser entièrement la conception de nouveaux parkings où la présence du public serait interdite. Les contraintes réglementaires seraient alors significativement allégées, ce qui permettrait de diviser par deux le coût d'une place de parking, détaille L'Usine Nouvelle.Cap sur l'internationalLa start-up, qui emploie actuellement 25 collaborateurs, entend déployer sa solution dans d'autres régions du monde. "D'autres projets sont en cours de développement en France et à l'étranger. Le premier d'entre eux sera lancé d'ici la fin de l'année", indique-t-elle. "Europe, Amérique du Nord, Asie, Golfe… Nous regardons attentivement toutes les zones où le trafic aérien va exploser dans les prochaines années", précise Stéphane Evanno, COO et cofondateur de Stanley Robotics. La pépite parisienne n'est toutefois pas la seule à développer ce genre de service. En Allemagne, par exemple, le robot Ray, mis au point par l'entreprise Serva Transport Systems, a été testé à l'aéroport de Düsseldorf dès juillet 2014.
Philippe J DEWOST's insight:
Un très bel investissement du Fonds "Ville de Demain" souscrit par la Caisse des Dépôts et géré par sa filiale Bpifrance dans le cadre du PIA.
more...
No comment yet.
Scooped by Philippe J DEWOST
Scoop.it!

How my research in AI put my dad out of a job – Snips Blog

How my research in AI put my dad out of a job – Snips Blog | cross pond high tech | Scoop.it
Back in 2007, when London was booming as the financial capital of the world, a new field called “algorithmic trading” was emerging. In essence, it is about leveraging Artificial Intelligence to place…
Philippe J DEWOST's insight:
Absolute must read. Y compris par le nouveau Ministre.
more...
No comment yet.
Rescooped by Philippe J DEWOST from Consensus Décentralisé - Blockchains - Smart Contracts - Decentralized Consensus
Scoop.it!

Bitcoin Energy Consumption Index

Bitcoin Energy Consumption Index | cross pond high tech | Scoop.it

Key Network Statistics

Bitcoin's current estimated annual electricity consumption* (TWh)= 12.79
Annualized global mining revenues = $1,425,461,447
Annualized estimated global mining costs = $639,742,471
Country closest to Bitcoin in terms of electricity consumption = Mozambique
Estimated electricity used over the previous day (KWh) = 35,054,382
Implied Watts per GH/s = 0.332
Electricity consumed per transaction (KWh)= 110
Number of U.S. households that could be powered by Bitcoin = 1,184,708.00
Number of U.S. households powered for 1 day by the electricity consumed for a single transaction = 3.71
Bitcoin's electricity consumption as a percentage of the world's electricity consumption = 0.06%

*The assumptions underlying this estimate can be found here.

Did you know?

Ever since its inception Bitcoin’s trust-minimizing consensus has been enabled by its proof-of-work algorithm. New sets of transactions (blocks) are added to Bitcoin’s blockchain roughly every 10 minutes by so-called miners. While working on the blockchain these miners aren’t required to trust each other. The only thing miners have to trust is the code that runs Bitcoin. The code includes several rules to validate new transactions. For example, a transaction can only be valid if the sender actually owns the sent amount. Every miner individually confirms whether transactions adhere to these rules, eliminating the need to trust other miners.

The trick is to get all miners to agree on the same history of transactions. Every miner in the network is constantly tasked with preparing the next batch of transactions for the blockchain. Only one of these blocks will be randomly selected to become the latest block on the chain. Random selection in a distributed network isn’t easy, so this is where proof-of-work comes in. In proof-of-work, the next block comes from the first miner that produces a valid one. This is easier said than done, as the Bitcoin protocol makes it very difficult for miners to do so. In fact, the difficulty is regularly adjusted by the protocol to ensure that all miners in the network will only produce one valid bock every 10 minutes on average. Once one of the miners finally manages to produce a valid block, it will inform the rest of the network. Other miners will accept this block once they confirm it adheres to all rules, and then discard whatever block they had been working on themselves. The lucky miner gets rewarded with a fixed amount of coins, along with the transaction fees belonging to the processed transactions in the new block. The cycle then starts again.

What kind of work are miners performing?

The process of producing a valid block is largely based on trial and error, where miners are making numerous attempts every second trying to find the right value for a block component called the “nonce“, and hoping the resulting completed block will match the requirements (as there is no way to predict the outcome). For this reason, mining is sometimes compared to a lottery where you can pick your own numbers. The number of attempts (hashes) per second is given by your mining equipment’s hashrate. This will typically be expressed in Gigahash per second (1 billion hashes per second).

 

Philippe J DEWOST's insight:

Bitcoin as a country would rank #82 between Mozambique and Slovenia in terms of energy consumption !

 

Fascinating and uptodate index with a trove of data for any of those who want a better insight on "the energy issue" posed by Bitcoin Mining, with several comparisons helping you to seize and frame notions such as TeraWatthour (TWh) per year.

 

And as a bonus, a very clear explanation of what mining is about and why proof of work has been designed as such.

more...
Philippe J DEWOST's curator insight, May 12, 12:25 PM

Bitcoin as a country would rank #82 between Mozambique and Slovenia in terms of energy consumption !

 

Fascinating and uptodate index with a trove of data for any of those who want a better insight on "the energy issue" posed by Bitcoin Mining, with several comparisons helping you to seize and frame notions such as TeraWatthour (TWh) per year.

 

And as a bonus, a very clear explanation of what mining is about and why proof of work has been designed as such.