Apple 'Air' Product Teased, DeepSeek's Theoretical 545% Margin, and Massive Gold Hydrogen Reserves Located

We're experimenting and would love to hear from you!

In this episode of 'Discover Daily', we begin with a tease from Apple CEO Tim Cook. His message on X that  "there's something in the air" has sparked speculation about new MacBook Air models featuring the M4 chip. These potential upgrades include a 25% boost in multi-core CPU performance, enhanced AI capabilities, and improved features like a 12MP Center Stage camera and Wi-Fi 6E support. Apple's shift to a more subtle announcement strategy marks a departure from their traditional product launch approach.

We also delve into the world of AI economics with Chinese startup DeepSeek's claim of a theoretical 545% cost-profit margin for its AI models. While this figure is based on calculations involving their V3 and R1 inference systems, real-world factors significantly reduce actual revenue. DeepSeek's aggressive pricing strategy and low development costs have sparked debate within the tech community and impacted AI-related stocks.

The episode's main focus is the discovery of vast "gold hydrogen" reserves beneath 30 U.S. states, as revealed by a groundbreaking USGS map. This natural hydrogen, formed through a process called serpentinization in geological formations known as rift-inversion orogens, could revolutionize clean energy production. The abundance and widespread distribution of these reserves may accelerate the transition to sustainable energy sources, potentially reshaping the global energy landscape and creating new economic opportunities in regions with significant deposits.

From Perplexity's Discover Feed:

https://www.perplexity.ai/page/apple-air-product-teased-QhTieZlcTwWodiMLzGzP3g

https://www.perplexity.ai/page/deepseek-s-theoretical-545-mar-_vk4xxCjSt.tLxQJCoU2sg

https://www.perplexity.ai/page/massive-gold-hydrogen-reserves-kRgxDixrTJCI1W17S2zcbw

**Introducing Perplexity Deep Research:**

https://www.perplexity.ai/hub/blog/introducing-perplexity-deep-research 

Perplexity is the fastest and most powerful way to search the web. Perplexity crawls the web and curates the most relevant and up-to-date sources (from academic papers to Reddit threads) to create the perfect response to any question or topic you’re interested in.

Take the world's knowledge with you anywhere. Available on iOS and Android

Join our growing Discord community for the latest updates and exclusive content.

Follow us on:

• Instagram
• Threads
• X (Twitter)
• YouTube
• Linkedin

Episode Transcript

Speaker 1 (00:01):
Welcome to Discover Daily, by Perplexity, your
AI-curated digest ofbreakthroughs in tech, science
and culture.
I'm Alex.
Today, we'll look at thediscovery of massive gold
hydrogen reserves which couldhave significant implications
for clean energy production.
But first let's look at whatelse is happening across the
tech landscape, from productteasers to AI breakthroughs.

(00:25):
Let's start with some excitingnews from Apple.
Ceo Tim Cook has just dropped ahint about a new product launch
.
This week.
In a brief video message sharedon, x Cook stated there's
something in the air.
This message has set the techworld abuzz with speculation.

(00:45):
The phrase there's something inthe air is particularly
significant as it echoes Apple'stagline from the original
MacBook Air launch back in 2008.
This has led many to believethat we're about to see the
unveiling of new MacBook Airmodels featuring Apple's latest
M4 chip.
If the rumors are true, whatcan we expect from this new

(01:06):
MacBook Air models featuringApple's latest M4 chip?
If the rumors are true, whatcan we expect from this new
MacBook Air?
Reports suggest the M4 chipcould bring a 25% boost in
multi-core CPU performancecompared to its predecessor.
We're also looking at anenhanced 16-core neural engine,
which would significantlyimprove the device's AI and
machine learning capabilities.

(01:27):
Other potential upgradesinclude a 12-megapixel center
stage camera for better videocalls, support for Wi-Fi 6E and
Bluetooth 5.3, and a baseconfiguration starting with 16
gigabytes of RAM.
These improvements aim toposition the new MacBook Air as

(01:48):
a powerful and efficient devicefor both everyday users and
professionals working withAI-related tasks.
Apple's approach to thisannouncement marks a shift from
their traditional product launchstrategy.
Instead of a grand event,they've opted for a more subtle,
modern approach using socialmedia.
This aligns with Apple's recentmove away from rigid annual

(02:12):
product release cycles, adaptingto the slower pace of
technological advancements.
While the MacBook Air with M4chip seems the most likely
candidate for this announcement,we should keep in mind that
Apple has other Air products inthe pipeline.
These include a new iPad Airexpected in the coming months

(02:33):
and even rumors of an iPhone 17Air for later in the year.
Moving on to our second story,chinese startup DeepSeek has
made waves with its claim of atheoretical 545% cost-profit
margin for its AI models.
This figure is based oncalculations involving their V3

(02:55):
and R1 inference systems.
Let's break down these numbers.
Deepseq's calculation is basedon a 24-hour run of its V3 and
R1 inference models, with anaverage node occupancy of 226.75
.
They estimated a dailyoperational cost of just over
$87,000 for leasing NVIDIA H800GPUs, against a potential

(03:21):
revenue of $562,000 if allprocess tokens were billed at R1
model rates.
But this 545% figure is purelytheoretical and doesn't reflect
actual earnings.
Real-world factorssignificantly reduce revenue,
including free services,discounted rates and varying

(03:44):
usage patterns.
Despite these limitations,deepseek's disclosure provides
valuable insights into theeconomics of AI inference
services.
Deepseek's aggressive pricingstrategy plays a key role in its
market positioning.
The company offerssignificantly lower rates
compared to Western competitorslike OpenAI or Anthropic.

(04:07):
Openai or Anthropic, forinstance, their Premium R1 model
is priced at $0.14 per millioninput tokens for cash hits and
$0.55 per million for cashmisses, with output tokens
costing $2.00 and $0.19 permillion.
But several factors impactactual revenue, including lower
pricing for the V3 model, freeaccess to web and app services,

(04:31):
off-peak discounts and varyingusage patterns.
As a result, the actual profitmargin of the DeepSeek R1 model
is estimated to be closer to 85%.
It's still impressive, but farfrom DeepSeek's theoretical
percentage.
The 545% cost-profit margin hassparked debate within the tech

(04:53):
community.
Critics argue that the termprofit margin is misleading as
it implies realized earningsrather than a theoretical
calculation.
A more accurate descriptionwould be gross margin or
contribution margin.
The controversy highlights theneed for standardized reporting
metrics in the rapidly evolvingAI industry.

(05:15):
It underscores the importanceof clear communication when
presenting financial data to thepublic and investors.
This debate may lead toincreased scrutiny of AI
companies' financial claims andpush for more transparent and
consistent reporting practicesacross the sector.
Deepseek's claims have hadsignificant impacts on the AI

(05:39):
industry.
Ai-related stocks experiencevolatility, with Nvidia losing
nearly 17% of its value and over$588 billion in market
capitalization.
The company's low developmentcosts of $5 to $6 million have
challenged assumptions aboutnecessary investments in AI

(06:01):
infrastructure.
Now let's dive into our mainstory of the day.
Scientists have unveiled a maprevealing vast reserves of
natural hydrogen, dubbed goldhydrogen, buried beneath 30 US

(06:27):
states.
This discovery challengesprevious assumptions about
geologic hydrogen availabilityand opens new possibilities for
clean energy exploration.
The primary mechanism forlarge-scale natural H2
generation is a process calledserpentinization of exhumed

(06:47):
mantle material.
This occurs when waterinteracts with certain rock
types in the Earth's crust.
Geological formations known asrift inversion origins have been
identified as potentialhotspots for natural H2
generation.
These are geological formationscreated by the folding of rift

(07:07):
basins.
They offer several advantagesover traditional rift
environments for hydrogenproduction, including colder
temperatures that provideoptimal conditions for
serpentinization and H2production, the presence of
suitable reservoirs and sealsnecessary for economic hydrogen
accumulations and the formationof an orogenic hydrogen system

(07:29):
analogous to petroleum systems.
Research suggests that riftinversion origins could produce
up to 20 times more hydrogenannually compared to rifting
environments.
This makes them valuabletargets for future exploration
and development.
The discovery presents asignificant opportunity for

(07:50):
clean energy production.
Unlike other forms of hydrogenproduction, gold hydrogen
requires no energy input forextraction, making it
potentially one of the cleanestand most cost-effective energy
sources available.
The US Geological Survey hasreleased an interactive map
showcasing potential locationsfor naturally occurring hydrogen

(08:13):
gas reserves across thecontiguous United States.
This innovative tool identifiesregions where hydrogen may
exist in large extractablequantities.
Key findings from the mapinclude highly prospective areas
encompassing large portions ofMichigan, eastern Kentucky,
southern North Dakota and partsof Kansas, colorado, wyoming,

(08:38):
iowa and Oklahoma.
Additional promising regionshave been identified along the
California coast and easternseaboard.
The USGS map is based on threecritical geological factors
suitable hydrogen sources,porous reservoir rocks and
natural seals to trap the gas.
This comprehensive approachprovides valuable insights for

(09:01):
prioritizing future research andexploration efforts in regions
with the highest likelihood ofsuccess.
Before we conclude, a brief noteon our own evolution at

(09:22):
Perplexity.
Our new deep research feature,launched earlier this month, now
analyses hundreds of sources inminutes.
Think of it as deploying apersonal research team through
our web and mobile platforms.
Through our web and mobileplatforms, this cutting-edge
tool combines autonomousreasoning with rapid processing
to deliver exhaustive reports onspecialized topics.

(09:45):
Deep research excels atexpert-level tasks across
various domains, from financeand marketing to product
research, and is available onour desktop and mobile apps.
That's it for today.
Thanks for listening and thankyou for your fan mail.
We read and work to incorporateevery bit of feedback we get,

(10:07):
so thank you.
This is Alex, reminding youthat every change begins with a
simple question why Stay curious?