Skip to Content

Challenger Deep

Challenger Deep

It was the best of times, but mostly it was the worst of times, especially if you were Nvidia this week following the meteoric rise of AI model DeepSeek and the massacre of the tech company’s market value.

After releasing its platform less than two years after forming, DeepSeek quickly made waves in the past month gaining a following of techno-faithful singing the praises of the little chat bot that could. Moreover, they did it for a lot less money than competitors like OpenAI or Google. If you want a sneak peak into the efficacy of DeepSeek may I recommend the fantastic article at Tom’s Guide by Amanda Caswell?

This kind of disturbance in the tech world does wonders for healthy competition, and I’m looking forward to the kind of innovation it will breed. I’m also interested in seeing the model put to use in security applications moving forward. Analytics, detection, and cybersecurity have come a long way in the past few years, but there’s always room for improvement.

At bargain prices no less. DeepSeek reportedly only spent roughly $5.6 million training its chat bot, where competitors have spent upwards of $100 million or more to achieve the same or inferior results according to NPR. The Chinese company was able to do that with restricted hardware stockpiled prior to the ban imposed on advanced microchips imposed by the U.S. a few years ago.

What that means now is that while U.S. companies may have been caught with their pants down with regard to the impressive software on display in DeepSeek’s model, it retains a hardware advantage and access to a great supply of available chips courtesy of the CHIPS & Science Act. What we have now is the rare opportunity for the U.S. to copy China’s homework for a change.

That is, if the current administration’s sledgehammering of federal funding doesn’t cost us the opportunity.

Shallow actions, deep implications.

Comments

To comment on this post, please log in to your account or set up an account now.