Category: Analysis

Sometimes it’s the news behind the news that’s really important. Such is the case with the recent announcement from Microsoft that they plan to start using the open source software-based Chromium project as a basis for future versions of their Edge browser. At a basic level, it’s an important (and surprising) move that seems significant for web developers and those who like to track web standards. For typical end users, though, it seems a bit ho-hum, as it basically involves under-the-hood changes that few people are likely to think much about or even notice.

However, the long-term implications of the move could lead to some profoundly important changes to the kinds of software we use, the types of devices we buy, the chips that power them, and much more.

The primary reason for this is that by adopting Chromium as the rendering engine for Edge, Microsoft should finally be able to unleash the full potential of the platform-independent, web-focused, HTML5-style software vision we were promised nearly a decade ago. If you’ll recall, initial assurances around HTML5 said that it was going to enable software that could run consistently within any compatible browser, regardless of the underlying operating system. For software developers, it would finally deliver on Java’s initial promise of “write once, run anywhere.” In other words, we could finally get to a world where everyone could get access to all the best software, regardless of the devices we use and own, and the ability to move our own data and services across these devices would become simple and seamless.

Alas, as with Java, the grandiose visions of what was meant to be, didn’t come to pass. Instead, HTML5-based applications struggled with performance and compatibility issues across platforms and devices. As a result, the potential nirvana of a seamless mesh of computing capabilities surrounding us never came to be, and we continue to struggle with getting everything we own to work together in a simple, straightforward way.

Of course, some might argue that they prefer the flexibility of choices and unique platform characteristics, despite the challenges of integrating across multiple platforms, application types, etc., and that’s certainly a legitimate point. However, even in the world of consistent software standards, there was never an intention to prevent choice or the ability to customize applications. For example, even though Chromium is also the web rendering engine for Google’s Chrome browser, Microsoft’s plan is to leverage some of the underlying standards and mechanisms in Chromium to create a better, more compatible version of Edge, but not build a clone of Chrome. That may sound subtle, but it’s actually an important point that will allow each of these companies (as well as others who leverage Chromium, such as Amazon) to continue to add their own secret sauce and provide special links to their own services and other offerings.

By moving the massive base of Windows users (as well as Edge browser users on the Mac, Android, and iOS, because Microsoft announced their intentions to build Chromium-powered browsers for all those platforms as well), the company has single-handedly shifted the balance of web and browser-based standards towards Chromium. This means that application developers can now concentrate more of their efforts on this standard and ensure that a wider range of applications will be available—and work in a consistent fashion—across multiple devices and platforms.

There are some concerns that this shifts too much power into the hands of a single standard and, some are worried, to Google itself, since it started the Chromium project. However, Chromium is not the same as Chrome (despite the similar name). It’s an open source-based project that anyone can use and add to. With Microsoft’s new support, they’ve ensured that their army of developers, as well as others who have supported the Microsoft ecosystem, will now support Chromium. This, in turn, will dramatically increase the number of developers working on Chromium and, therefore, improve its quality and capabilities (in theory, at least).

The real-world software implications for this could be profound, especially because Microsoft has promised to embed Chromium support into Windows. What this will do is allow web-based applications access to things like the file system, being able to work offline, touch support, and other core system functions that have previously prevented browser-based apps from truly competing against stand-alone apps. This concept, also known as progressive web apps (PWA), is seen as being critical in redefining how apps are created, distributed, and used.

For consumers, this means the need to worry about OS-specific mobile apps or desktop applications could go away. Developers would have the freedom to write applications that have all the capabilities of a stand-alone app, yet can be run through a browser and, most importantly, can run across virtually any device. Software choices should go up dramatically, and the ability to have multiple applications and services work together—even across platforms and devices—should be significantly easier as well.

For enterprise software developers, this should open the floodgates of cloud-based applications even further. It should also help companies move away from dependencies on legacy applications and early Internet Explorer-based custom enterprise applications. From traditional enterprise software vendors like SAP, Oracle, and IBM through modern cloud-based players like Salesforce, Slack, and Workday, the ability to focus more of their efforts on a single target platform should open up a wealth of innovation and reduce difficult cross-platform testing efforts.

But it’s not just the software world that’s going to be impacted by this decision. Semiconductors and the types of devices that we may start to use could be affected as well. For example, Microsoft is leveraging this shift to Chromium as part of an effort to bring broader software compatibility to Arm-based CPUs, particularly the Windows on Snapdragon offerings from Qualcomm, like the brand-new Snapdragon 8cx. By working on bringing the underlying compatibility of Chromium to Windows-focused Arm64 processors, Microsoft is going to make it significantly easier for software developers to create applications that run on these devices. This would remove the last significant hurdle that has kept these devices from reaching mainstream buyers in the consumer and enterprise world, and it could turn them into serious contenders versus traditional X86-based CPUs from Intel and AMD.

On the device side, this move also opens up the possibility for a wider variety of form factors and for more ambient computing types of services. By essentially enabling a single, consistent target platform that could leverage the essential input characteristics of desktop devices (mice and keyboards), mobile devices (touch), and voice-based interfaces, Microsoft is laying the groundwork for a potentially fascinating computing future. Imagine, for example, a foldable multi-screen device that offers something like a traditional Android front screen, then unfolds to a larger Windows (or Android)-based device that can leverage the exact same applications and data, but with subtle UI enhancements optimized for each environment. Or, think about a variety of different connected smart screens that allow you to easily jump from device to device but still leverage the same applications. The possibilities are endless.

Strategically, the move is a fascinating one for Microsoft. On one hand, it suggests a closer tie to Google, much like the built-in support for Android-based phones did in the latest version of Windows 10. However, it’s specifically being done through open source, and is likely to leverage its recent Github developer resource purchase to make web standards more open and less specifically tied to Google. At the same time, because Apple doesn’t currently support Chromium and is still focused on keeping its developers (and end users) more tightly tied into its proprietary OS, Microsoft is essentially further isolating Apple from key web software standards. In an olive branch move to Apple users, however, Microsoft has said that they will bring the Chromium-powered version of Edge to MacOS and likely iOS, essentially giving Apple users access to this new world of software, but via a Microsoft connection.

In the end, a large number of pieces have to come together in order to make this web-based, platform-free version of the software world come to pass, and it wouldn’t be the least bit surprising to see roadblocks arise along the way. Still, Microsoft’s move to support Chromium could prove to be a watershed moment that quietly, but importantly, sets some key future technology trends into motion.

Sometimes it takes real world frustrations before you can really appreciate the advances that technology can bring. Such is the case with mobile broadband-equipped notebook PCs.

Before diving into the details of why I’m saying this, I have to admit upfront that I’ve been a skeptic of cellular-capable notebooks for a very long time. As a long-time observer of, data collector for, and prognosticator on the PC market, I clearly recall several failed attempts at trying to integrate cellular modems into PCs over the last 15 years or so. From the early days of 3G, and even into the first few years of 4G-capable devices, PC makers have been trying to add cellular connectivity into their devices. However, attach rates in most parts of the world (Western Europe being the sole exception) have been extremely low—typically, in the low single digits.

The primary reasons for this limited success have been cost—both for the modem and cellular services—as well as the ease and ubiquity of WiFi and personal hotspot functions integrated into our smartphones. Together, these factors have put the value of cellular connectivity into question. It’s often hard to justify the additional costs for integrated mobile broadband, especially when the essentially “free” alternatives seem acceptable.

Despite all these concerns, however, we’ve seen a great deal of fresh attention being paid to cellular connected PCs of late. Specifically, the launch of the always connected PC (ACPC) effort by Microsoft, Qualcomm, and several major PC OEMs (HP, Asus, and Lenovo) this time last year brought new attention to the category and started to shift the discussion of PC performance towards connectivity, in addition to traditional CPU-driven metrics. Since that first launch with Snapdragon 835-based devices, we’ve already seen second generation Snapdragon 850-based PCs, such as Lenovo’s Yoga C630, start to ship.

We’ve also seen Intel bring its own modems into the PC market in a big way over the last few months, highlighting the increased connectivity options they enable. In the new HP Spectre Folio leather-wrapped PC, for example, Intel created a multi-chip module that integrates its Amber Lake Y-Series CPU, along with an XMM 7560 Gigabit LTE modem. Conceptually, it’s similar to the chiplet-style design that combined an Intel CPU and AMD Radeon GPU into a single multi-chip module that Dell used in its XPS 15 earlier this year, but integrates a discrete modem instead of the discrete GPU.

Together these efforts, as well as expected advancements, highlight the many key technological enhancements in semiconductor design that are being directed towards connectivity in PCs. Plus, with the launch of 5G-capable modems and 5G-enabled PCs on the very near horizon, it’s clear that we’ll be enjoying even more of these chip design-driven benefits in the future.

Even more importantly, changes in the wireless landscape and our interactions with it are bringing a new sense of pertinence and criticality to our wireless connections. While we have been highly dependent on wireless connections in our PCs for some time, the degree of dependence has now grown to the point where most people really do need (and expect) reliable, high-quality signals all the time.

This point hit home recently after I had boarded a plane but needed to finish a few critical emails before we took off. Unfortunately, the availability and quality of WiFi connections while people are getting seated is dicey at best. But by leveraging the integrated cellular modem in my Spectre Folio review unit, I was able to do so no problem. Similarly, in a long Lyft ride to an airport on another recent trip, I leveraged the modem in the Yoga C630 for similar purposes. Plus, in situations like conferences and other events where WiFi connections are often spotty, having a cellular connectivity alternative can be the difference between having a usable connection and not having one at all.

Admittedly, these are first-world problems and not everybody needs to have reliable connectivity in these types of limited situations. In other words, I don’t think the extra cost of integrated cellular modems makes sense for everyone. But, for people who are on the run a lot, the extra convenience can really make a difference. This is another example of the fact that many of the technological advances that we now see in the PC market are generally more incremental and meant to improve certain situations or use cases. Integrated cellular connections are in line with this kind of thinking as they provide an incremental boost in the ability to find a usable internet connection.

In addition to convenience, the increase of WiFi network-based security risks has raised concerns about using public WiFi networks in certain environments. While not perfect, cellular connections are generally understood to be more secure and much less vulnerable to any kind of network snooping than WiFi, providing more peace of mind for critical or sensitive information.

Of course, little of this would matter if network operators didn’t make pricing plans for cellular data usage on PCs attractive. Thankfully, there have been improvements here as well, but there’s still a long way to go to truly make this part of the connected PC experience friction-free. The expected 2019 launch of 5G-equipped notebooks will likely trigger a fresh round of pricing options for connected PC data plans, so it will be interesting to see what happens then.

Ultimately, while some of the primary concerns around the connected PC remain, it’s also becoming clear that many other issues are starting to paint the technology in a light. Always on, always reliable connections are no longer just a “nice to have,” but a “need to have” for many people, and along with the technology advancements, increased security and lower data plan costs are combining to create an environment where connected PCs finally start to make sense.

Are the robots coming, or are they already here? Fresh off the impressive, successful Mars landing of NASA’s InSight lander robotic spacecraft, it seems appropriate to suggest that robots have already begun to make their presence felt across many aspects of our lives. Not only in space exploration and science, but as we enter into the holiday shopping season, their presence is being felt in industry and commerce as well.

Behind the scenes at factories building many of the products in demand this holiday season, to the warehouses that store and ship them out, robots have been making a significant impact for quite some time. Building on that success, both Nvidia and Amazon recently made announcements about robotics-related offerings intended to further advancements in industrial robots.

Just outside of Shanghai last week, at the company’s GTC China event, Nvidia announced that Chinese e-commerce giants JD.com and Meituan have both chosen to use the company’s Jetson AGX Xavier robotics platform for the development of next-generation autonomous delivery robots. Given the expected growth in online shopping in China, both e-commerce companies are looking to develop a line of small autonomous machines that can be used to deliver goods directly to consumers, and they intend to use Xavier and its associated JetPack SDK to do so.

At the company’s AWS:Invent event in Las Vegas this week, Amazon launched a cloud-based robotics test and development platform called AWS RoboMaker that it’s making available through its Amazon Web Services cloud computing offering. Designed for everyone from robotics students who compete in FIRST competitions through robotics professionals working at large corporations, RoboMaker is an open-source tool that leverages and extends the popular Robot Operating System (ROS).

Like some of Nvidia’s software offerings, RoboMaker is designed to ease the process of programming robots to perform sophisticated actions that leverage computer vision, speech recognition, and other AI-driven technologies. In the case of RoboMaker, those services are provided via a connection to Amazon’s cloud computing services. RoboMaker also offers the ability to manage large fleets of robots working together in industrial environments or places like large warehouses (hmm…wonder why?!)

The signs of growing robotic influence have been evident for a while in the consumer market as well. The success of Roomba robotic vacuums, for example, is widely heralded as the first step in a home robotics revolution. Plus, with the improvements that have occurred in critical technologies such as voice recognition, computer vision, AI, and sensors, we’re clearly on the cusp of what are likely to be some major consumer-focused robotics introductions in 2019. Indeed, Amazon is heavily rumored to be working on some type of home robot project—likely leveraging their Alexa work—that’s expected to be introduced sometime next year.

Robotics is also a key part of the recent renaissance in STEM education programs, as it allows kids of many ages to see the fun, tangible efforts of their science, math, and engineering-related skills brought to life. From the high-school level FIRST robotics competitions, down to early grade school level programs, future robotics engineers are being trained via these types of activities every day in schools around the world.

The influence of these robotics programs and the related maker movement developments have reached into the mainstream as well. I was pleasantly surprised to see a Raspberry Pi development board and other robotics-related educational toys in stock and on sale at, of all places, my local Target over the Black Friday shopping weekend.

The impact of robots certainly isn’t new in either the consumer or business world. However, except for a few instances, real-world interactions with them have still been limited for most people. Clearly, that’s about to change, and people (and companies) are going to have to be ready to adapt. Like the AI technologies that underlie a lot of the most recent robotics developments, there are some great opportunities, but also some serious concerns, particularly around job replacement, that more advanced robotics will bring with them. The challenge moving forward will be determining how to best use robots and robotic technology in ways that can improve the human experience.

In the end, Amazon’s announcement of Queens and Crystal City as the two new HQ locations was anti-climactic. In part because of the decision to split HQ2 after a year of breathless media coverage, and in part because many believe that the D.C. area choice was already a foregone conclusion given Jeff Bezos’ already substantial ties to the region. Cynics believe that the choice was typically Amazonian – where the company could get the best deal, rather than it necessarily being the best location for the company or, heaven forbid, embracing something slightly more altruistic by helping an up-and-coming tech city like Pittsburgh or Atlanta move to the top tier.

More importantly, this highly public process prompted a year-long thought process of how cities must compete for business and talent in the 21^st century economy. The Bay area and Seattle are already overheated, and the infrastructure (affordable housing, transport) to support much more is inadequate. In my hometown of Boston, which was one of the finalists, there was a collective feeling of relief at having not been selected, given already sky-high housing prices, clogged roads, and an overburdened public transport system. And it is both sad and a poor reflection on our current leadership when it takes the prospects of an Olympics or major new corporate headquarters to catalyze the type of strategic thinking and investment for any city that wants to be competitive in the 21^st century economy.

In my view, there are five key elements necessary to be in the game:

• Talent. Both extant and potential via a good educational system and strong universities.
• Diversity of economy. You don’t want to be too dependent on any one industry or economic sector. Pittsburgh is a great example. Whereas the collapse of the steel industry nearly killed Pittsburgh 1.0, Pittsburgh 2.0 has a much greater diversity of vibrant industries, fueled by a unique level of cooperation among its private, public, and educational sectors.
• Infrastructure. An adequate road system and a viable public transportation system. It’s becoming clear that 21^st century workers don’t want to spend their lives sitting in cars. For certain types of employers/employees, proximity of a decent airport is also a factor.
• Affordable Housing & Livability. If you’re earning $100,000 and can’t afford a pleasant one-or two- bedroom apartment/condo in the city or a modest home in a close-in suburb, it’s a problem. There’s also the slightly more amorphous concept of ‘livability’, such as a city’s walkability, and the presence/proximity of culture and other amenities. Something I’ve always thought is important is ‘what’s a tank away’?, in other words are there nice places you can easily get to for a day trip or a weekend (beach, mountains, etc.).
• Progressive Local Leadership. Given the dysfunction on the national level and lack of strategic, long-term investment in education and infrastructure, cities and states with strong local leadership are breaking through. Examples: Nashville, Tulsa, and Los Angeles.

Now, let’s take a look at a few of the cities that were not only finalists in the Amazon hunt, but would be viable contenders at least for the ‘next Amazon’. How do they stack up on the above criteria?

Boston. Educational institutions, diversity of economy, livability, and talent are its strengths. But the city’s housing prices have become Bay-area-esque, and its infrastructure is overburdened and crumbling, with no long-term plan in place. It’s like a city that’s over-touristed and is saying, ‘no more’.

Dallas. Yes, it’s economy is on fire, but my sense is that this is a place that people move to once their career is established rather than being a preferred location for younger talent. And while it’s affordable compared to a lot of other cities, Dallas lacks top tier educational institutions that are feeders to tech companies, and remains too auto-centric, despite some recent investments in public transport.

Austin. This city has a lot of the right ingredients in place and has attracted a lot of tech companies already. A much younger, more vibrant feel than Dallas, in part because of the giant University of Texas at Austin. But growth has outpaced infrastructure investment, with sprawl and traffic impacting the quality of life factors that made this city attractive after Dell helped put it on the map.

Atlanta. Has many of the same attributes and challenges as Dallas, but is a notch above in terms of top tier educational institutions. I think traffic/sprawl/infrastructure challenges are what kept it out of the running for Amazon.

Pittsburgh. Here’s a city that has done and is doing a lot of things right to become a 2.0 version of itself. A quite livable place. Not yet a major league city on a global scale, and needs to substantially invest in its transport infrastructure if current growth trajectory continues.

Nashville. Has a lot of the same ingredients as Pittsburgh, and has used its assets to become a major healthcare/tech center. A progressive mayor has courted companies and made the right investments and strategic decisions to make the city much more livable (new park & bike trails, better roads/transport, tons of new housing).

Los Angeles. The highly progressive mayor Eric Garcetti is making huge investments in infrastructure and affordable housing and doing real things to address the homeless issue, tackle inequality, and diversify the economy. This fascinating, diverse place has the potential to be a revitalized global city for the 21^st Century…or it could get crushed under the weight of its size and years of under-investment.

I should also mention three Canadian cities that are already seriously on the map:

Toronto. Now North America’s third largest city, Toronto (and the tech epicenter of nearby Waterloo) was a contender for Amazon HQ. And not just because of the idea of ‘Trump Snub’ that the media loved writing about. This city is culturally and economically diverse, has strong educational institutions, and is very livable. It does suffer from U.S.-like problems of traffic and sprawl and inadequate rapid transit outside the city core. And housing prices are among the highest in North America. But you will be hearing more about Toronto in the coming years.

Vancouver. Incredible quality of life, if you can get past the Seattle-esque six months of gloominess. This will be a major 21^st century city, given its setting, strong educational institutions, and diversity. The huge run up in real estate prices (mainly due to foreign investment) and lack of good rapid transit are challenges…that are actually being addressed. V

Montreal. This city has all the right ingredients: already a tech and creative hub, strong educational institutions and tons of talent, still relatively affordable, and a high quality of life. Montreal is also making a significant investment in improving its roads and expanding its transport system. Its brutal winters are a factor for some, and still restrictive language laws keep some companies (and people) away.

And finally, here’s a few more from among the major North America cities:

Getting to the Next Stage: Minneapolis-St. Paul, Portland (OR), Phoenix, Detroit, Philadelphia.

Not Progressing/Worry Button: Chicago, Miami, Orlando, Charlotte, Baltimore…and San Francisco.

Cities shouldn’t be overdoing the post-mortem about why they didn’t get Amazon HQ2. Instead, they should be thinking about what’s needed for them to land the ‘next HQ’.

A classic way for engineers to solve a particularly vexing technical problem is to move things in a completely different direction—typically by “thinking outside the box.” Such is the case with challenges facing the semiconductor industry. With the laws of physics quickly closing in on them, the traditional brute force means of maintaining Moore’s Law, by shrinking the size of transistors, is quickly coming to an end. Whether things stall at the current 7nm (nanometer) size, drop down to 5nm, or at best, reach 4nm, the reality of a nearly insurmountable wall is fast approaching today’s leading vendors.

As a result, semiconductor companies are having to develop different ways to keep the essential performance progress they need moving in a positive direction. One of the most compelling ideas, chiplets, isn’t a terribly new one, but it’s being deployed in interesting new ways. Chiplets are key IP blocks taken from a more complete chip design that are broken out on their own and then connected together with clever new packaging and interconnect technologies. Basically, it’s a new version of an SoC (system on chip), which combined various pieces of independent silicon onto a multi-chip module (MCM) to provide a complete solution.

So, for example, a modern CPU typically includes the main compute engine, a memory controller for connecting to main system memory, an I/O hub for talking to other peripherals, and several other different elements. In the world of chiplets, some of these elements can be broken back out into separate parts (essentially reversing the integration trend that has fueled semiconductor advances for such a long time), optimized for their own best performance (and for their own best manufacturing node size), and then connected back together in Lego block-type fashion.

While that may seem a bit counter-intuitive compared to typical semiconductor industry trends, chiplet designs help address several issues that have arisen as a result of traditional advances. First, while integration of multiple components into a single chip arguably makes things simpler, the truth is that today’s chips have become both enormously complex and quite large as a result. Ensuring high-quality, defect-free manufacturing of these large, complex chips—especially while you’re trying to reduce transistor size at the same time—has proven to be an overwhelming challenge. That’s one of the key reasons why we’ve seen delays or even cancellations of moves to current 10nm and 7nm production from many major chip foundries.

Second, it turns out not every type of chip element actually benefits from smaller sizes. The basic argument for shrinking transistors is to reduce costs, reduce power consumption, and improve performance. With elements like the analog circuitry in I/O components, however, it turns out there’s a point of diminishing returns where smaller transistors are actually more expensive and don’t get the performance benefits you might expect from smaller production geometries. As a result, it just doesn’t make sense to try and move current monolithic chip designs to these smaller sizes.

Finally, some of the more interesting advancements in the semiconductor world are now occurring in interconnect and packaging technologies. From the 3D stacking of components being used to increase the capacity of flash memory chips, to the high-speed interfaces being developed to enable both high-speed on-chip and chip-to-chip communications, the need to keep all the critical components of a chip design at the same process level are simply going away. Instead, companies are focusing on creating clever new ways to interconnect IP blocks/components in order to achieve the performance enhancements they used to only be able to get through traditional Moore’s Law transistor shrinks.

AMD, for example, has made its Infinity Fabric interconnect technology a critical part of its Zen CPU designs, and at last week’s 7nm event, the company highlighted how they’ve extended it to their new data center-focused CPUs and new GPUs now as well. The next generation Epyc server CPU, codenamed “Rome,” scheduled for release in 2019, leverages up to 8 separate Zen2-based CPU chiplets interconnected over their latest generation Infinity Fabric to provide 64 cores in a single SoC. The result, they claim, is performance in a single socket server that can beat Intel’s current best two-socket server CPU configuration.

In addition, AMD highlighted how its new 7nm data center-focused Radeon Instinct GPU designs can now also be connected over Infinity Fabric both for GPU-to-GPU connections as well as for faster CPU-to-GPU connections (similar to Nvidia’s existing NVLink protocol), which could prove to be very important for advanced workloads like AI training, supercomputing, and more.

Interestingly, AMD and Intel worked together on a combined CPU/GPU part earlier this year that leveraged a slightly different interconnect technology but allowed them to put an Intel CPU together with a discrete AMD Radeon GPU (for high-powered PCs like the Dell XPS15 and HP 15” Spectre X360) onto a single chip.

Semiconductor IP creator Arm has been enabling an architecture for chiplet-like mobile SoC designs with its CCI (Cache Coherent Interconnect) technology for several years now. In fact, companies like Apple and Qualcomm use that type of technology for their A-Series and Snapdragon series chips, respectively.

Intel, for its part, is also planning to leverage chiplet technology for future designs. Though specific details are still to come, the company has discussed not just CPU-to-CPU connections, but also being able to integrate high-speed links with other chip IP blocks, such as Nervana AI accelerators, FPGAs and more.

In fact, the whole future of semiconductor design could be revolutionized by standardized, high-speed interconnections among various different chip components (each of which may be produced with different transistor sizes). Imagine, for example, the possibility of more specialized accelerators being developed by small innovative semiconductor companies for a variety of different applications and then integrated into final system designs that incorporate the main CPUs or GPUs from larger players, like Intel, AMD, or Nvidia.

Unfortunately, right now, a single industry standard for chiplet interconnect doesn’t exist—in the near term we may see individual companies choose to license their specific implementations to specific partners—but there’s likely to be pressure to create that standard in the future. There are several tech standards for chip-related interconnect, including CCIX (Cache Coherent Interconnect for Accelerators), which builds on the PCIe 4.0 standard, and the system-level Gen-Z standard, but nothing that all the critical players in the semiconductor ecosystem have completely embraced. In addition, standards need to be developed as to how different chiplets can be pieced together and manufactured in a consistent way.

Exactly how the advancements in chiplets and associated technologies relate to the ability to maintain traditional Moore’s law metrics isn’t entirely clear right now, but what is clear is that the semiconductor industry isn’t letting potential roadblocks stop it from making important new advances that will keep the tech industry evolving for some time to come.

After years of hype and inflated expectations, it’s clear that the mania around fully autonomous cars has cooled. In a refreshing, and much needed change, we’re starting to see companies like Nvidia, Intel/Mobileye, and Arm now talk much more about the opportunities to enable enhanced safety for occupants in cars supporting advanced technologies.

It’s not that the tech industry is giving up on autonomy—as recent announcements about new rounds of trials from Lyft, Uber, and others, as well as advanced new chip designs clearly illustrate—but the timeframes for commercial availability of these advancements are starting to get pushed out to more realistic mid-2020 or so dates. Even more importantly, the messaging coming from critical component players is shifting away from roads packed with Level 5 fully autonomous cars within a few years, to ways that consumers can feel more comfortable with and safer in semi-autonomous cars.

Over the last few weeks, Nvidia, Intel, and Arm have all discussed research reports and technology advancements in the automotive market that are focused primarily on security, with the technology providing a supporting role. Nvidia, for example, released a comprehensive study called “The Self-Driving Safety Report” that provides a view into how the company incorporates safety-related technology and thinking into all aspects of its automotive product developments. The report covers everything from AI-based design, to data collection and analysis, to simulation and testing tools, all within a context of safety-focused concerns.

Intel, for their part, released a comprehensive study on what they termed the Passenger Economy this past summer, but recently touted some findings from the report that focus on the relatively slow consumer acceptance for self-driving cars due to concerns around safety. Essentially, while 21% of US consumers say they’re ready for an autonomous car now, it’s going to be 50 years before 63% consumers believe they become the norm. To address some of these concerns, Intel is touting its Responsibility-Sensitive Safety (RSS) model, which it describes as a mathematical model for autonomous vehicle safety. The idea for RSS is to develop a set of industry standards for safety that can then be used to reassure consumers in a transparent way about how autonomous cars will function. Recently, Intel announced that Baidu had chosen to adopt the RSS model for its autonomous driving efforts in China.

Back in late September, Arm announced a new program called Safety Ready that ties together a number of the company’s security and safety technologies into a unified structure that, while not limited to the automotive market, is very well-suited for it. Safety Ready incorporates both chip IP designs and software that are focused on applications where functional safety is critical and allows the company to meet the key automotive-related functional safety certifications, including ISO 26262 and ASIL-D. At the same time, the company also introduced a new automotive-specific chip design called the Cortex-A76AE that integrates a capability called Split-Lock that allows a dual-core CPU to either function as two individual components doing separate tasks or two single components running in lockstep, where one can take over immediately if the other fails. As in many automotive applications, redundancy of functions is key for safety concerns and the Split-Lock capability of this new design brings it to digital components as well.

While it may seem that all these announcements are a somewhat dramatic shift from how the tech industry had been talking about autonomous cars, in reality they are simply part of a maturing perspective on how this market will develop. In addition, they’re based on some practical realities that many in the autonomous automotive industry have started to recognize. First, as research has continued to show, most consumers are still very leery of autonomous car features and aren’t ready to trust cars that take over too much control.

Second, the level of difficulty and technical challenge in getting even basic autonomy features to work in a completely safe, reliable way is now recognized as being even harder than it was first believed. Even semi-autonomous cars integrate an extraordinarily complex combination of advanced technologies that includes AI and machine learning, advanced sensors and fusing of sensor data, and intelligent mapping, all of which have to work together seamlessly to ensure a safe, high-quality driving experience. There’s no doubt that we will start to get there, but for now, it’s reassuring to see companies focus on the critical safety enhancements that assisted driving features can bring, as we look further out to the world of full autonomy.

Sometimes it’s good to be late. That’s the argument we’ve heard from quite a few technology companies that are tardy to a particular market. They claim that their “lateness” isn’t really a detriment, but actually a positive attribute to their offering.

That’s certainly the case that Oracle is making when it comes to their cloud computing product—Oracle Cloud, which ties together Oracle Cloud Infrastructure, or OCI, with their various cloud, database and application platforms. Right now, Oracle is clearly battling it out for number four with IBM and SAP, well behind Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform, regardless of which market research data provider you choose to believe. Of course, with the group creating OCI primarily based in Seattle, it certainly doesn’t hurt (nor is it terribly surprising) that 90-95% of them are alumni from the Amazon AWS or the Microsoft Azure cloud computing group, according to people who are in the Oracle cloud product development group. As a result, several said that they can learn from the missteps of these companies, and develop a more efficient solution.

Even with this 4th¬-place position, however, Oracle clearly sees an opportunity to differentiate from the others, based on its core database IP and a focus on high-performance applications. As was pointed out yesterday on the first day of Oracle’s OpenWorld event in San Francisco, there are quite a few companies that are interested in migrating traditional Oracle-based database-focused products onto the cloud. So far, the percentage of traditional Oracle-focused companies that have actually done this seems to be pretty small—perhaps high single digits as a percentage—but it’s clear that much of the hesitation is due to the more conservative approach to technology adoption that many Oracle customers have.

In other words, they may be slow, but that’s really just a reflection of where the real market is. Despite what you may read in the common tech press, the truth is, many companies are much slower at adopting these new technologies than the hype would have you believe, especially in the case of cloud or even AI adoption.

During CTO Larry Ellison’s keynote speech yesterday, he focused on both the performance and cost advantages of Oracle’s cloud-based solution, highlighting benchmarks against AWS that were almost too good to be true, with figures like 80x improvements in performance and 100% reduction in costs. Basically, he promised that anyone who could move data-intensive workloads with high performance requirements—admittedly only a portion of typical cloud workloads—to Oracle’s cloud would see a high-quality return on investment. In addition, he highlighted the company’s technology infrastructure offering, which he argued had both more flexibility and more capability than many of its competitors.

The key message in Ellison’s keynote speech, however, was on security-related issues in the cloud. In fact, he said that Oracle leveraged its late appearance in the market to build a newer, more agile and more secure means of creating a cloud platform. In what he termed Cloud Generation 2.0, he said the company built a separate ring of cloud control computers that lived outside of (and conceptually encircled) the standard cloud computing nodes. In other words, it is a network outside of the core cloud computing network that avoids the challenges of having to intermix the cloud control plane architecture with that of a company’s more proprietary data structure. By doing so, he claimed, the company could avoid the security risks of other challengers’ platforms, which co-mingle the cloud control software with customers’ own data and applications.

Conceptually, it’s an intriguing new architecture that looks at the interactions between customers’ data and the cloud providers’ software in very different ways. Plus, given the fact that these cloud control computers would not be based on x86 architecture (the company would not say exactly what they would use, but given Arm’s recent efforts in infrastructure, that seems a likely bet), they add a new twist to the security discussion. To be clear, this architecture opens up potential security and management challenges of its own, such as not knowing what’s happening within the enclosed circle of bare metal cloud computing nodes that have no deployment or management software running on them. However, it simultaneously creates a new type of interaction model between customer and data infrastructure provider that has additional potential security benefits.

Ultimately, it boils down to a question of trust. What companies do you trust, not only from a software perspective, but also from a security, data management, and comprehensive point of view. There isn’t an easy answer to these questions, but it’s clear that, with its new security-focused cloud architecture, Oracle is providing a very different perspective on how to think about cloud-based computing models.

There’s nothing like a common enemy to bring together companies that otherwise find themselves in competition with one another. In the still untamed world of the Internet of Things (IoT), however, it doesn’t require another company to trigger that kind of reaction, just a  set of real-world challenges: complexity, confusion, and overwhelming choice.

After bold proclamations from an enormous range of sources about the nearly limitless opportunity that IoT was supposed to represent, the cold hard reality of modest deployment numbers has created a dark cloud over the tech industry. The promised land of billions or even a trillion connected devices that IoT was supposed to enable seems as distant as ever and some analysts are starting to walk back their overenthusiastic early proclamations.

That’s likely why the two leaders in major chip architectures for IoT (and virtually all!) devices—Arm and Intel—have come together to help break down some of the barriers that have held the industry back. As both companies clearly recognize, the opportunity enabled by IoT is still very real, but it turns out it’s a lot harder to achieve than many first anticipated. As with so many issues in the tech industry, much of the problem has to do with scale. Setting up a few connected sensors to measure important data and then drawing insights from that sounds pretty straightforward, and, in many early pilot tests, results were very encouraging. Nearly everyone involved in the industry, in fact, can point to a few great case studies of where IoT deployments have made a very positive impact.

Taking those principles into large, widespread deployments, however, has proven to be very difficult. From the enormous diversity of IoT software platforms and ecosystems, to an even wider range of device types (and chip architectures powering them), through the massive set of potential security challenges, many companies eager to leverage IoT have slowed or even halted their implementation plans.

One of the biggest challenges, it turns out, is actually one of the first things that has to be done: connecting the devices to the software tools that will collect the data they generate. While that may sound simple, it turns out there are quite a few steps to take and issues to consider when talking about quickly and securely connecting hundreds of millions of devices that will be built by tens of thousands of different companies.

Specifically, you have to consider the provisioning of a device—which is the process of setting it up to properly connect to a network—and ensuring that it’s connected to the right place, communicating securely, and updated with the latest device firmware, a set of tasks typically referred to as onboarding. In addition, you need the flexibility to connect those devices to any variety of different cloud platforms (especially in the multi-cloud era that many companies now find themselves in), and you need to ensure that the device hasn’t been tampered with at any point during its manufacturing, distribution, installation, or operation.

Again, multiply all those concerns by the billions and it’s easy to see that even the smallest delay or the slightest oversight in any step of the process could be very problematic. Acutely aware of these concerns, both Arm and Intel have introduced a variety of technologies to address them over the years. Notably, Intel’s provisioning system called Secure Device Onboard (SDO), leverages the company’s hardware root-of-trust based EPID (Enhanced Privacy ID) technology to cleverly mask the real identity of a device, while simultaneously assuring a connected network that the device is who it says it is and that it can be connected automatically to a trusted network. The net result is the ability to securely bring IoT devices online without any human interaction, also called zero touch, which is a huge timesaver.

In addition, Intel SDO features a capability called late binding that allows a device’s security credentials and intended network target to be added at any point in the supply chain. This allows device manufacturers, or companies who are piecing together IoT devices from a variety of different suppliers, to cost-effectively mass produce items that can be customized for specific applications or environments at a later date. This allows IoT device makers the ability to save costs, but not give up the critical security customizations necessary to avoid huge problems like the Mirai Botnet security attack that has hit many insecure IoT devices over the last few years.

Arm, for its part, recently unveiled their Pelion IoT Device Management service, which also has a zero touch provisioning feature (for Arm IP-based devices, such as those with chips featuring Cortex-M or Cortex-A processor cores), and provides device management capabilities. Like the Intel solution, Arm’s Pelion service is ultimately based on a hardware root of trust that’s built into the core design of Arm-licensed processors and microcontrollers so commonly used in IoT devices.

By getting these two solutions to work together, the companies are helping overcome a number of factors that individually they weren’t able to achieve. First, of course, is the fact that the solutions have now been extended to cover both x86-based Intel powered IoT devices and Arm core-licensed IoT devices—essentially just about every device imaginable. As a result, now both types of devices can be provisioned with zero human touch, both types of devices can take advantage of Intel’s late binding technology, and both types of devices can be managed through Arm’s Pelion Device Management service.

The ultimate result is that these steps should make it significantly easier for companies who may deploy a wide variety of different IoT devices, built with different components from different manufacturers, to quickly and securely connect to the right places in a consistent manner. This can not only significantly reduce potential friction points in large-scale deployments, but makes it much easier to manage, update, and work with them as a collective group. Plus, by making it more cost-effective (and more secure) for device manufacturers, it should lower the costs of large IoT installations, reducing yet another potential barrier to adoption.

The march to billions of IoT devices and the amazing ecosystem that it should enable is still bound to be a long one, but by combining the best capabilities of their respective IoT solutions, Arm and Intel are taking a major step together in clearing the road of potential distractions.

The technology may have a very futuristic feel to it, but in current implementations, it’s clear now that artificial intelligence (AI) and machine learning (ML) have very practical real-world applications that aren’t nearly as scary as some may fear.

Thanks to a freshly completed study by TECHnalysis Research on the usage of AI applications in US businesses based on a survey of 504 IT professionals working in medium (100-999 employees) and large (1,000+ employees) companies that are currently doing some type of AI work, the perspective that appears is pragmatic. (To read more on the study, you can also check out two previous columns on the subject: “Survey: Real World AI Deployments Still Limited” and “AI Application Usage Evolving Rapidly”.)

Companies want to use AI-based applications to improve their overall efficiency across a number of different areas. The hope is that AI-based tools can reduce some of the more tedious, repetitive tasks that can slow organizations down—or that some simply choose not to do because the tasks are so challenging to maintain.

In addition, companies of various sizes and types believe that AI-based applications can speed up or help automate many of these tasks. Whether it’s automatically filtering spam and phishing attacks from email; analyzing files or other data as they’re opened, created, or passed along; or acting as a perimeter shield on networks and examining the packets that travel across them, the realistic benefits of many AI-based or AI-enhanced applications are proving to be attractive to nearly 1-in-5 US companies with at least 100 employees.

The chart in Figure 1 highlights the top goals of AI projects and deployments that these companies have already embarked on.

Fig. 1

In addition to efficiency and automation, many organizations see AI as a great tool for increasing security across many different aspects of their organization, including data, networks, and the devices their employees use.

For a number of companies, AI is also seen as a next-generation big data analysis tool, destined to deliver on the promise of big data that many companies disappointingly discovered wasn’t as easy to mine for insights as they were led to believe. The idea is that AI algorithms can take over some of the data grunt work necessary to uncover useful information and leverage larger amounts of raw data in more efficient ways. It’s still early days here, but it’s clear that many companies are eager to see these kinds of results.

Some organizations are also hoping to generate cost savings from AI-based tools, but it’s clear that this is still a lesser priority for many organizations, particularly because of the costs that are often involved with AI-based applications and projects.

Speaking of which, cost concerns were one of the top three challenges that companies are currently facing with their AI efforts, as the chart in Figure 2 illustrates.

Fig. 2

The biggest challenges, however, had to do with complexity—both of the technology itself as well as the means to implement it. This isn’t terribly surprising because the level of real understanding about AI is still quite low. Despite the fact that AI and machine learning have been around in some form or other for decades, most people are just starting to learn about them. Plus, the hidden “black box”-type means by which many AI algorithms work makes it difficult for anyone but dedicated specialists to completely get their heads around the technology and understand how to make it do what you need (or want) it to.

On top of all this, there are still many lingering fears about the potential influence of the technology. Despite the relatively low rankings on impact on headcount impact (see chart), for example, it was clear from the verbatim comments that survey respondents put into an open-ended question on the overall importance, value, and impact of AI, that the fear of layoffs triggered by AI loomed large. As one respondent wrote, “…there is a real fear of putting people out of jobs. You hate to be left behind when helpful technology is out there, but it’s hard to eliminate a job that’s been there for 20 years.”

At the same time, there was also a great deal of excitement and promise expressed in response to the same open-ended question. For example, “AI is bound to impact every single industry, and, in our organization, AI can help deliver better search results and deliverables for some specific business cases. Finding patterns and reducing inefficiency is super important for our organization and [both of these] will benefit from the advent of new AI solutions. We strongly support AI-based solutions and prefer to adopt them quickly and gain a business edge.”

Practically speaking, there was also the recognition that AI is here now, and its impact is going to be critical. As one respondent summarized, “I think that AI will change how a lot of us do business. The change will be good. Getting over the initial hurdles of integration is the hardest part. The data provided by AI integrations will be invaluable and come much faster than was possible before.”

(You can download a copy of the TECHnalysis Research AI in the Enterprise Study Highlights for free. A complete version of the full report, with 178 slides that go into extremely fine detail on all the major question asked in the survey, is available for purchase.)