Help Amnesty International save the chips. What happens if it turns out to be a bust?

But what if artificial intelligence did not succeed?

This is not just a virtual question. Some signs indicate that the growth of artificial intelligence is StallOr at least slowdown. New models no longer show significant improvements from increasing the size or amount of training data. Nobel Prize winner Dimis Hasabis Notice recently “We no longer get the same progress” in developing artificial intelligence. Andrink Horwitz, one of the most prominent investors in artificial intelligence, is the same Common concerns The capabilities of the artificial intelligence model seem to be the plateau.

One of the reasons for slowing AI may be that the models have already consumed most of the available digital data, leaving a little left for more improvement. The developers instead resort to artificial data, but it may be less effective – and they may even make Worse models.

Artificial intelligence development is also a dense capital. Training the most advanced models requires that calculation groups cost billions of dollars. Even one training can cost tens of millions of dollars. However, while the development costs continue to rise, cash rewards are limited. Regardless of artificial intelligence assistants, there is there Little examples It is artificial intelligence to generate returns that justify these enormous capital investments.

Some companies are already expanding the investment of Amnesty International’s infrastructure due to the cost. Microsoft, for example, is “Slowness or stopping Some early stage projects “and equipment orders were canceled for many international data center projects. It is said Cut their GPU orders. Chips, energy deficiency, and public concerns are also obstacle to the adoption of collective artificial intelligence.

If AI Boom Peters is out, this is bad news for the chips industry, which used this new technology to avoid dangerous recession.

The chips are more expensive to make them. The development of new manufacturing operations costs billions of dollars; Building new factories can cost tens of billions of dollars. All of these costs are transferred to consumers, but outside artificial intelligence, customers are not keen to buy more expensive chips. Luxury techniques in artificial intelligence treatments today are not useful for other purposes.

Artificial intelligence is late for industry account: manufacturing has become more expensive, while performance gains are shrinking. The economic promise of artificial intelligence justifies high chips, but if it goes, the chips industry needs to find something else to persuade people to maintain investment in the manufacture of advanced chips. Otherwise, the advanced chips industry will become unnecessary: it will cost new technologies more and more, with less and less.

The stagnation of the chips industry will increase many geopolitical and economic goals. Point governments Billion dollars In building local chips industries. US President Donald Trump routinely threatens Use definitions To bring the manufacture of semiconductors to the homeland.

The supposed US leadership in the development of chips may prove that it is a mirage, especially like China dominates the production of old chips. Amnesty International’s reversal would shake the technology sector in the world, forcing large technology to rethink its bets.

Looking at these risks, policy makers need to encourage more innovation in artificial intelligence by facilitating access to data, chips, energy and cooling. This includes pragmatic policies on copyright and data protection, a balanced approach to the manufacture of ground and external chips, and the removal of organizational barriers in front of energy use and generation. Governments should not necessarily apply the precautionary principle to artificial intelligence; The benefits are very large so that their development cannot be obstructed, at least in these early stages. Artificial intelligence applications should not face widely, such as independent compounds or home robots, unreasonable requirements for implementation.

Investors must also explore alternative AI methods that do not require much data and infrastructure, and may open the growth of new artificial intelligence. The industry must also explore nonI applications for chips, even if only to manage its risks.

To ensure that the chips can survive, it should reduce the cost of making advanced chips. Companies should work together to search and develop, as well as work with universities, to reduce development costs. More investment in numbers, advanced packaging and re -configuration devices. The industry should support the inter -operating standards, open source tools, and the development of graceful devices. The joint infrastructure supported for design and manufacturing can help smaller companies to finish ideas before manufacturing. However, more importantly, the motivation for prominent manufacturing may be reverse results: to do so indifferently will increase the costs of chips.

The future of the chips and the AI are deeply intertwined. If the chips will flourish, artificial intelligence should grow. If not, the entire chips sector may now be in danger.

The opinions expressed in cutting comments Fortune.com are only the opinions of their authors and do not necessarily reflect opinions and beliefs luck.