While it’s not quite robots building robots, we’re getting closer to it as iPhone-maker Foxconn teams up with Nvidia to build a global network of “AI factories.”
What’s interesting about these factories is they will build on the lights-out approach Apple’s partner has already adopted across iPhone production lines. These heavily automated factories make use of connected machinery and machine vision intelligence to augment human workers on the production lines.
The ultimate aim: replace production line workers as much as possible.
Automation + AI = ?
That’s the idea artificial intelligence (AI) factories build on. But as the partners explain it, the notion goes at least two steps further. You see, while AI may help augment production, the next level of IoT is to let AI guide production.
Nvidia founder and CEO Jensen Huang puts it this way: “A new type of manufacturing has emerged — the production of intelligence. And the data centers that produce it are AI factories.”
The plan is for Foxconn to use Nvidia’s tech to build “a new class of data center to power connected applications such as digitized manufacturing, inspection workflows, robotics, and vehicle development and more."
In some ways, it’s yet another evolution to the cloud-based post-capital model that increasingly drives global economies. Just as the move toward digital services and commerce relies heavily on data gathered in the cloud, so will this new era of industrial production.
The data created by the transaction, manufacture, distribution, use, and recycling of connected devices will perhaps become more valuable — and certainly more strategically important — than the devices themselves. To understand the notion, consider how Amazon’s sales data continues to drive that company’s growth. In the same way, the information generated by AI factories will feed into future gains.
For the data owners, at least.
What benefits will AI factories provide?
So, how do the partners explain this?
- Generative AI services on the production line and across the product lifecycle.
- The ability to use real data to drive simulations to accelerate the training of autonomous machines.
- Fast deployment of connected smart industrial equipment.
- An acceleration in the development of truly autonomous vehicles.
- In factories, improved processes, such as fault detection, and optimization of the current generation of connected machinery.
- Potentially highly automated systems capable of constant revision to handle supply, demand, or component availability in real time, across complex chains.
- Overall, better productivity and reduced costs.
Finally, data gathered during use of devices manufactured by these factories may be fed back to the production systems. This could accelerate fault reporting and diagnosis, optimize manufacturing, and minimize flaws.
Using a car as an example, Huang explained:
“This car would, of course, go through life experience and collect more data. The data would go to the AI factory. The AI factory would improve the software and update the entire AI fleet. In the future, every company, every industry, will have AI factories.”
The other facet to this we can surmise. Each time we see data surfaced that was previously hard to discern, the information gathered from across the product lifecycle is used to develop and ideate new product/service families that may have been impossible to visualize on the information held before. This will also happen with AI factory production.
Et tu, Siri?
Foxconn is Apple’s biggest manufacturing partner, and its iPhone factories are already highly advanced. With this in mind, it makes sense to expect the company to integrate these new technologies within its work for Cupertino.
Foxconn actually hints at this, saying: “Foxconn is eyeing its own [AI factories] that will tap into the NVIDIA Omniverse platform and Isaac and Metropolis frameworks to meet the strict production and quality standards of the electronics industry.”
If we accept that Apple already has some of the highest standards of quality control in tech manufacturing, and also consider its commitment to circular manufacturing, we can visualize how depersonalized hardware usage data might contribute to future product development as components are recycled and reused.
iPhones are quite smart devices, but in the context of globally connected intelligent manufacturing and lifecycle management, the sum of all these parts may well become greater than the whole.
And, of course, AI in consumer electronics manufacturing almost certainly hints at a future in which it becomes much easier to put factories anywhere, as manufacturing will become increasingly low-skilled and low-staffed.
In that context, AI might help Apple/tech’s future plan to decouple from China. But that, as they say, is another story.
Please follow me on Mastodon, or join me in the AppleHolic’s bar & grill and Apple Discussions groups on MeWe.