We just lately talked about Tesla’s $16.5 billion funding to reshape its chip provide chain.
Now it looks as if Amazon could possibly be testing comparable waters.
However not with a brand new chip manufacturing facility. And never with a giant announcement both. Actually, the transfer Amazon just lately made was about as under-the-radar as you may get.
It just lately purchased $84 million price of AMD inventory.
Clearly, this wasn’t precisely a headline-grabbing transfer for the $2.4 trillion firm.
In any case, $84 million is a rounding error for billionaire Jeff Bezos. He spent extra on his current nuptials in Italy!
So this might simply be routine portfolio diversification.
However Amazon sometimes doesn’t make large public investments in different tech giants.
And while you join the dots, this modest buy may point out one thing a lot greater.
I imagine it might symbolize step one in a deeper strategic realignment round Amazon’s cloud and AI ambitions…
And a recognition that its homegrown efforts haven’t stored tempo with the AI race that’s driving the 4th industrial revolution.
Betting on a Rival
Let me stroll you thru my considering.
Amazon’s first quarter 13F submitting, launched final month, revealed that the corporate had acquired 822,234 shares of AMD someday earlier than March 31, 2025.
This place was valued at round $84.4 million on the time.
Clearly, a stake this measurement isn’t large enough to affect AMD’s route. But it surely might signify the beginning of a deeper connection between the 2 firms.
As I’ve written about earlier than, AMD is more and more central to the AI {hardware} dialog.
Its MI300X chips are wanting like the primary severe challenger to Nvidia’s H100 dominance.
Supply: AMD
However not like Nvidia, which is chronically supply-constrained, AMD is raring to increase adoption of its knowledge middle chips by means of cloud suppliers like AWS.
Actually, a few of AWS’s digital servers already run on AMD’s EPYC chips, because of an ongoing partnership between the 2 firms.
So this funding doesn’t come out of nowhere. But it surely does increase an even bigger query:
Why now?
One attainable motive is that Amazon’s personal chip efforts haven’t gone as deliberate.
Amazon launched its first customized AI chip, Inferentia, in 2018 and launched it publicly in 2019. Then in late 2020, it unveiled Trainium, a training-focused processor designed to compete with GPUs.
Each chips had been positioned as cost-effective, high-performance alternate options to Nvidia’s dominance.
However adoption has been restricted, and the broader market response has been lukewarm.
Regardless of early partnerships with startups and a few inner use circumstances, Trainium hasn’t gained significant adoption amongst large-scale clients.
Inferentia has discovered some traction for inference duties, however neither chip has damaged out past a handful of AWS workloads.
One motive is Nvidia’s stranglehold on the AI software program stack.
Nvidia doesn’t simply promote GPUs — it constructed a whole ecosystem round them.
Its proprietary software program platform generally known as CUDA has turn into the trade normal. Most machine studying instruments are constructed round it, and retraining these techniques to work on non-CUDA chips may be expensive and time-consuming.
Naturally, this makes it a troublesome promote to alter to a competitor’s chips, even when they promise extra pace for much less cash.
And that sort of friction has stalled Amazon’s capability to show Trainium and Inferentia into severe contenders, whilst demand for AI {hardware} is reaching new heights.
So what do you do when your customized chips can’t compete at scale?
One resolution is to show to your suppliers.
In Amazon’s case, it may also imply turning to a possible accomplice.
In different phrases, a more in-depth partnership with AMD looks as if the plain play.
As a result of, proper now, each main tech firm is at present locking in its AI {hardware} technique.
Microsoft is deepening its Nvidia dependency. The corporate is even leasing total knowledge facilities to safe GPU entry.
Google continues to refine its in-house TPU chips whereas nonetheless shopping for Nvidia models in bulk.
Meta is each constructing its personal chips and shopping for them off-the-shelf to hedge its bets.
Whereas Oracle is leaning into Ampere-based ARM chips for its cloud development.
I’ve been pounding the desk about this main tech shift for some time.
If you wish to be aggressive in as we speak’s market, you’ll be able to’t simply lease compute anymore. It’s a must to personal — or at least affect — all the stack, from the chips themselves to the software program layers that run on prime.
That’s what made Tesla’s Samsung deal so fascinating. By guaranteeing its next-gen AI chips are inbuilt Texas, it’s decreasing geopolitical threat and cementing provide.
Amazon’s AMD funding is perhaps its personal model of that play.
Right here’s My Take
If Amazon’s current inventory buy is an indication of a creating partnership with AMD, it might have actual penalties for the following part of AI infrastructure.
For one, it could give AWS a viable different to Nvidia’s lengthy wait instances and premium pricing.
Extra importantly, it might open the door for AMD to construct customized chips particularly for Amazon’s AI wants, and with tighter integration into its software program stack.
That’s the sort of vertical optimization — proudly owning all the stack —- we’ve already seen from Apple, Google and Tesla.
And if Amazon can pull it off with AMD, it might give the corporate each a provide and a efficiency edge within the AI cloud wars.
However none of that is assured.
In any case, Amazon’s present stake in AMD is comparatively small, and there’s been no official announcement of a strategic partnership between the 2 firms.
It’s fairly attainable that what’s occurring right here is all smoke and no hearth.
However in an trade the place each firm is scrambling to regulate its AI future, it’s not loopy to imagine {that a} small transfer like this might reveal greater ambitions for Amazon.
To me, it alerts a possible pivot towards a tighter integration with one in every of Nvidia’s fiercest rivals.
And if it does flip right into a deeper collaboration between the 2 firms, don’t be shocked to see customized AMD chips powering AWS AI companies by this time subsequent yr.
Regards,
Ian KingChief Strategist, Banyan Hill Publishing
Editor’s Notice: We’d love to listen to from you!
If you wish to share your ideas or solutions in regards to the Day by day Disruptor, or if there are any particular subjects you’d like us to cowl, simply ship an e-mail to dailydisruptor@banyanhill.com.
Don’t fear, we gained’t reveal your full identify within the occasion we publish a response. So be happy to remark away!
