Intel Plus Altera

Why this deal is worth watching with a powerful magnifying glass.

popularity

Since the Intel-Altera deal reached the handshake phase earlier this month, there have been a lot of theories being forwarded and a lot of questions being raised. And there have been very few answers, in part because the deal isn’t finalized yet and in part because this marriage will take time to play out in the market—maybe years.

But along with the GlobalFoundries-IBM pairing, this is the other really industry-shaking deal. Both will have permanent repercussions that will be felt on a global scale.

The Intel-Altera matchup has implications for Intel’s foundry business, for the U.S. defense industry, for Intel’s penetration into markets where it has had only limited success, and for Intel’s stronghold inside of corporate data centers.

Semiconductor Engineering has been posing questions ever since rumors of this deal began surfacing to scores of executives and engineers about what the deal means for them and how they see it playing out. And from almost every angle, this deal makes enormous sense—strategically and tactically.

Intel’s track record with acquisitions is mixed, at best, and past history does raise some legitimate questions. Throughout the 1990s and early 2000s the company attempted to get into the communications server business in fits and starts, buying companies such as Dialogic and Voice Technologies and Xircom, but never with a clear plan about how it would win in that market. Intel jumped into the white box PC business for a while, too, but dumped all of that when it came back to the realization that, at heart, it’s a processor company.

Intel’s success has been in big processors—the kind that can process billions of transactions inside of data centers, whether they are internal clouds, public clouds, or any other configuration or name you might want to slap on them—and the smaller kind that can power desktop or laptop computers. The company completely missed the mobile phone revolution, missed many of the vertical opportunities that have come along in the home, automotive and industrial markets, and there are continuing questions about exactly how successful it will be in the IoT device market.

But when it comes to standard products—focusing on cutting costs, cutting margin, shrinking features, developing new process technology and getting chips to market that are powerful, low-energy and market-proven—Intel wins. What makes the Altera deal so interesting is that, like Intel processors, FPGAs are standard products. They adhere to Moore’s Law. But they also offer flexibility to be able to configure servers in the cloud so they can be used as heterogeneous multi-core machines, which means they can be highly power efficient or extremely powerful—or both. And they can be configured to any configuration and reconfigured later.

These chips also can be stacked together, using interposers, bridges, or through-silicon vias, depending upon how Intel decides to package these chips, to improve both performance and lower power. In fact, the combination of these technologies will offer the kinds of power/performance benefits unseen since the days of classical scaling, pre-90nm, and maybe even more. Intel has publicly said that stacking architectures will be essential to the continuation of Moore’s Law. What it didn’t say was how it would get there. Altera is a key piece to that puzzle.

On the tactical side, the Intel-Altera deal has other benefits. For one thing, it secures a big customer for Intel Custom Foundry and a steady revenue stream that can help pay for the enormous costs of maintaining a lead-edge fab. The cost of developing advanced process technology and filling it with new equipment is so high that even Intel has to open its foundry up to other chipmakers. It’s like opening up a large house to boarders in order to subsidize the upkeep. For another, it provides Intel an entry into markets where FPGAs are strong—communications, networking, automotive, test and measurement, set-top boxes and TVs, and defense.

The defense aspect is interesting in light of the sale of IBM’s fabs to GlobalFoundries. The U.S. defense industry has been scrambling to keep it supply chain secure, but with IBM gone it has only one other option for an advanced digital foundry in the United States—Intel. (See related story here.) This could prove to be a new and lucrative market for Intel’s foundry services.

So what becomes of the marriage between these companies? Assuming the deal is approved by regulatory agencies, it could have big implications in many markets. Unlike previous deals, where Intel either consumed companies like some giant black hole, or more recent deals like McAfee and Wind River, where it kept those companies as semi-independent operations, the Altera deal is one that holds enormous promise if managed properly. Time will tell if this one is managed properly, of course, and part of that will depend upon how well the cultures of the two companies will blend. But this certainly is an industry-defining deal, and one worth watching with a powerful magnifying glass.