Latest / Ctrl AI Profit / Ep. 163 | Meta Just Flooded the AI Compute Market
Transcript
- 0:20SPEAKER_01: I'm Michael, a small business owner and entrepreneur since 1983, founder of Cadenhead Services and 850 Media.
- 0:28SPEAKER_01: I speak from four decades of real operational experience, not white papers.
- 0:32SPEAKER_01: This is control AI profit, and this week Meta opened its data centers to everyone.
- 0:38SPEAKER_00: Meta confirmed a plan called MetaCompute to lease its idle data center AI capacity to external customers.
- 0:45SPEAKER_00: This puts Meta in direct competition with AWS, Azure, Google Cloud, and the independent NeoCloud providers.
- 0:53SPEAKER_01: Meta has been one of the biggest buyers of AI chips on the planet.
- 1:10SPEAKER_00: And they built too much.
- 1:12SPEAKER_00: Or at least more than they need for internal workloads at any given moment.
- 1:17SPEAKER_00: Data centers do not run at 100% capacity all the time.
- 1:21SPEAKER_00: There are peaks and valleys.
- 1:23SPEAKER_00: Meta looked at the valleys and saw a business opportunity.
- 1:28SPEAKER_01: This is not a side project.
- 1:29SPEAKER_01: MetaCompute is a direct shot at the cloud computing market.
- 1:34SPEAKER_01: Meta already knows how to run infrastructure at scale.
- 1:37SPEAKER_01: They already have the chips, the real estate, the power contracts, and the engineering talent.
- 1:42SPEAKER_01: Now they are selling what they built for themselves.
- 1:46SPEAKER_00: The market reaction was immediate.
- 1:48SPEAKER_00: Core Weave and other neo cloud providers fell 12% to 17%.
- 2:05SPEAKER_01: Here is what small business owners need to understand.
- 2:09SPEAKER_01: If you have been renting GPU instances from cloud providers, you are about to see a price war.
- 2:15SPEAKER_01: Meta does not need to make cloud margins.
- 2:17SPEAKER_01: They need to cover the cost of idle capacity and learn the cloud business.
- 2:23SPEAKER_01: Their pricing will be aggressive.
- 2:25SPEAKER_00: The comparison is Amazon Web Services in 2006.
- 2:29SPEAKER_00: Amazon built infrastructure for retail, had excess capacity, and turned it into a side business that became bigger than retail.
- 2:36SPEAKER_00: Meta is attempting the same playbook.
- 2:38SPEAKER_00: AI infrastructure built for internal models becomes a general-purpose cloud service.
- 2:43SPEAKER_01: For businesses running AI workloads, this creates both opportunity and complexity.
- 2:49SPEAKER_01: The opportunity is cheaper compute.
- 2:51SPEAKER_01: Meta has already signaled they will price below market to gain share.
- 2:55SPEAKER_01: If your training and inference costs are a meaningful line item, MetaCompute could cut them significantly.
- 3:01SPEAKER_00: The complexity is vendor fragmentation.
- 3:04SPEAKER_00: Now you have AWS, Azure, Google Cloud, Core Weave, Lambda, and MetaCompute as credible GPU cloud providers.
- 3:14SPEAKER_00: Each has different pricing, different availability, different software stacks, and different lock-in profiles.
- 3:21SPEAKER_01: I have seen this movie before.
- 3:23SPEAKER_01: When a dominant buyer enters a market as a seller, the first phase is price disruption.
- 3:28SPEAKER_01: The second phase is feature parity.
- 3:30SPEAKER_01: The third phase is ecosystem lock-in.
- 3:33SPEAKER_01: Meta is in phase one.
- 3:35SPEAKER_00: The businesses that benefit most are the ones that can move workloads between providers without rebuilding their entire pipeline.
- 3:42SPEAKER_00: If you built your AI training environment on AWS-specific tools, switching to Metacompute is expensive and risky.
- 3:49SPEAKER_00: If you built on open source frameworks with portable configurations, you can chase the lowest price.
- 3:55SPEAKER_01: Here is my framework.
- 3:57SPEAKER_01: If you are spending more than$5,000 per month on AI compute, you should be evaluating Metacompute within 90 days of launch.
- 4:05SPEAKER_01: The potential savings are meaningful, and the earlier you switch, the more leverage you have with your current provider to match pricing.
- 4:14SPEAKER_00: But do not migrate mission critical workflows on day one.
- 4:18SPEAKER_00: Meta is new at cloud services.
- 4:20SPEAKER_00: Their reliability, support, and uptime guarantees will not match AWS or Azure initially.
- 4:27SPEAKER_00: Use Meta for batch training, experimental workloads, and non-urgent inference.
- 4:32SPEAKER_00: Keep production workloads on proven infrastructure.
- 4:35SPEAKER_01: The other angle is lock-in risk.
- 4:38SPEAKER_01: Meta has a history of building ecosystems that are hard to leave.
- 4:42SPEAKER_01: They did it with Facebook pages, with Instagram shopping, with their advertising platform.
- 4:48SPEAKER_01: Their cloud service will likely integrate tightly with their AI models, their developer tools, and their data formats.
- 4:56SPEAKER_00: The businesses that adopt Metacompute without a migration plan may find themselves in the same position as businesses that built on Facebook's platform, dependent on a provider that controls pricing, access, and features with no practical alternative.
- 5:10SPEAKER_01: And here is the broader signal.
- 5:13SPEAKER_01: Meta's move confirms that AI infrastructure is transitioning from scarcity to glut.
- 5:18SPEAKER_01: The companies that built the most capacity, Meta, Google, Microsoft, Amazon, are now competing to rent it out.
- 5:26SPEAKER_01: The compute shortage that justified premium pricing for two years is ending.
- 5:31SPEAKER_00: The neo-cloud providers are the most exposed.
- 5:34SPEAKER_00: Companies like Core Weave built their entire business on the assumption that GPU supply would remain tight and that they could charge premium prices.
- 5:42SPEAKER_00: Meta entering as a low-priced competitor with its own chips and its own data centers changes their business model fundamentally.
- 5:50SPEAKER_01: For small businesses, the advice is straightforward.
- 5:52SPEAKER_01: Do not sign long-term GPU contracts right now.
- 5:56SPEAKER_01: The market is about to get cheaper.
- 5:58SPEAKER_01: Keep your workloads portable.
- 6:00SPEAKER_01: And when Metacompute launches in your region, run a cost comparison before renewing any existing cloud commitments.
- 6:13SPEAKER_00: Not because Meta will dominate cloud, they might not, but because their entry proves that capacity exists and that the premium pricing of the last two years was temporary.
- 6:29SPEAKER_00: Frank is an AI, an open claw-powered agent serving as digital media director at 850 Media.
- 6:37SPEAKER_00: An AI co-hosting a show about AI for business owners is not a gimmick.
- 6:41SPEAKER_00: It is a live demo of exactly what the show is about.
- 6:44SPEAKER_00: See you in the next one.