Mad Hedge Technology Letter
January 10, 2025
Fiat Lux
Featured Trade:
(NVIDIA GETS PUT IN PLACE)
(NVDA)
It is uncommon when private tech companies lash out at the government like they are some kind of whipping boy.
Silicon Valley is so successful - they don’t need to target government policy.
Anger comes in many forms but openly criticizing the government could get you in some hot water in places like China.
Just look at Alibaba founder Jack Ma who was taken out to pasture by the Chinese communist party.
Criticism is usually reserved in Silicon Valley because subsidies and relationships are preserved to fight another day.
Nvidia finally felt it was time to let loose on the disastrous Biden Administration as the chip company gets dragged into politics just like almost everything else in American society.
Nvidia viciously criticized new chip export restrictions that are expected to be announced soon, saying the White House was trying to undercut the incoming Trump administration by imposing last-minute rules.
It’s is arguable that many strategic moves the current administration executes are to stymy the next administration.
Private tech companies are just collateral damage and Nvidia is finding that out the hard way.
The looming changes would cap the sale of US artificial intelligence chips on both a country and company basis — a move that would more tightly limit exports to most of the world.
The extreme ‘country cap’ policy will affect mainstream computers in countries around the world, doing nothing to promote national security but rather pushing the world to alternative technologies.
Nvidia has been the biggest beneficiary of a surge in AI spending over the past two years, helping turn the once-niche company into the world’s most valuable chipmaker. Its shares nearly tripled last year, following a 239% gain in 2023.
Speaking at the CES conference in Las Vegas this week, Huang said he expected Trump to bring less regulation.
I can now say with more certainty that tech stocks appear to be in a bubble and it doesn’t help that an obstructionist government is putting in limits to how much they can sell abroad.
Globalization has accelerated to some extreme that many people and businesses are still having a tough time wrapping their minds around what happened.
Putting a cap on the number of AI chips Nvidia can export will just gift the advantage to another competitor.
The Chinese have never played by the rules with their state subsidies and stealing of intellectual property.
These are several hallmarks of their national heavyweights.
Hamstringing Nvidia is the worst thing the US government could do minus shutting them down completely.
In general, the amount of bureaucratic nonsense, dysfunction, red tape, and needless saber-rattling is starting to hit the bottom line of Silicon Valley.
This could all bring forward a selloff from this tech bubble we are currently in.
Granted, I will acknowledge that the federal government isn’t only targeting the tech sector and the inefficiencies run across a wide swath of the U.S. economy system.
But that doesn’t make it better.
We are priced to the point where AI is guaranteed to become our savior and I would say to hold on because we are nowhere near certainty and there are very few use cases of all this AI data center investment.
We are trading at highs and the government going after Silicon Valley will hasten a sharp selloff in expensive tech stocks.
Don’t play with fire or you’ll be burned.
When John identifies a strategic exit point, he will send you an alert with specific trade information as to what security to sell, when to sell it, and at what price. Most often, it will be to TAKE PROFITS, but on rare occasions, it will be to exercise a STOP LOSS at a predetermined price to adhere to strict risk management discipline. Read more
(PORTFOLIO PERFORMANCE FOR 2024)
January 10, 2025
Hello everyone
Firstly, I’m deeply sorry for anyone who has been affected by the L.A fires. Losing everything in a fire is traumatic; I do hope the community comes together to give comfort and people support each other during this devastating time.
The market is expressing a topping pattern. So, we would be wise to take some funds off the table. As I said in my Monday newsletter, there is a real possibility that we could expect movement towards $5000 and under in the S&P500. So, let’s bank some profits.
Below I’m showing our portfolio and our end-of-year performance.
On the left, I show the date, ticker symbol, stock, and purchase price, and on the right, I show the price of the stock on December 31, 2024, + the $ gain/loss and the % gain/loss for the year.
Energy was our worst-performing sector. We can expect further lightning bolt movement in oil followed by a low and then a move up. The timing of these moves is hard to nail down.
I have cut and pasted from my Excel spreadsheet, instead of sending the whole thing out to you.
So, if you had bought one stock in each of the above companies when I suggested, you would have been ahead by $2,122.29 or 1480% for the year. (Two people have checked these numbers besides me).
Let’s take some profits now on the following stocks:
On November 8, 2023, we bought Digital Ocean (DOCN) at $26.30. On January 8, 2024, the stock sits at $34.48. Sell the stock and take profits.
Profit = $8.18 OR 31.10%
On November 27, 2023, we bought Dell (DELL) at $75.00.
On January 6, 2024, we scaled in again at $77.50.
Again, on January 17, 2024, we scaled in at $85.00.
On January 8, 2024, the stock sits at $119.31. Sell the stock and take profits.
Profit = $75.00 -$119.31 = $44.31 OR 59.08%
Profit =$77.50 - $119.31 = $41.81 OR 53.94%
Profit = $85.00 -$119.31 = $34.31 OR 40.36%
If you bought any of the Home Builders: Lennar, Pulte Group, D. R. Horton, Toll Brothers, I advise you to sell out of them. Interest rates will stay on the high side.
On October 10, 2024, I presented a list of stocks where you could add weight. The Home Builders were part of this list and looked promising with the prospect of many more interest rate cuts. Now, however, that does not look likely to happen, so we need to cut this sector from our portfolio. On October 10, the stocks were at the following prices. On January 8, 2025, the stocks listed these prices. I advise to scale out on days when the market and these stocks show an uptick.
D.R. Horton $183.39 - $139.90
Lennar $178.20 - $133.54
Toll Brothers $149.07 - $127.03
Pulte Group $138.66 - $110.46
On April 3, 2024, we bought Taiwan Semiconductor (TSM) at $140.22. On January 8, 2024, it’s at $207.12
Profit = $66.90 OR 47.71
QI CORNER
Cheers
Jacquie
Global Market Comments
January 10, 2025
Fiat Lux
Featured Trades:
(A CHEAP HEDGE FOR THIS MARKET)
The S&P 500 is now 3.2% of its all-time high at 6,100 and is showing definientia signs of rolling over.
So financial advisors, pension fund managers, and cautious individuals have been ringing me up asking what is the best way to hedge the heady 23% gain in 2024.
You can forget about buying the Volatility Index (VIX), (VXX). The huge contango, the discount front month futures contracts have too far month ones, almost guarantees that your hedge will be enormously expensive and expire worthless before it has the chance to do any good.
No, there’s a much better way to do this.
Buy deep out-of-the-money, long-dated S&P 500 (SPY) put options. You want to go deep out-of-the-money so your insurance policy is cheap.
You also want to go long-dated, so time decay doesn’t kill you and your option position lives long enough to do some good. By long-dated, I’m thinking five months out, like the June 20, 2025 option expiration date.
All of this logic points to the (SPY) June 2025 $530 puts today priced at $8.00, which as of today are 10% out-of-the-money.
Here is the beauty of this position. A put option rises in value in falling markets. But so does option-implied volatility, creating a leveraged hockey stick effect on the value of your put position. And deep out-of-the-money options always see implied volatilities rise much faster than near-money ones.
You don’t need the market to drop the full 10% to make enough profit in this position to offset losses elsewhere in your portfolio.
A much more likely 5% market correction would cause the value of the June 2025 $530 puts to jump from $8.00 to $14.00, a gain of 75%, as long as that drop happens soon. However, add in an expected pop in implied options volatility and the profit could be as much as 100%.
So how many June 2025 $250 puts should you buy?
Let’s say you have a $100,000 portfolio. Only two put option contracts would provide enough coverage for your entire exposure ($100,000/100 shares per contract/$530 (SPY) strike price) = 1.88 contracts rounded up to two. Two contracts of the June 2025 $530 puts will cost you $1,600 (2 X 100 shares per contract X $8.00).
In other words, $1,600 buys you an insurance policy on $100,000 portfolio exposure for six months. Sounds like a deal to me.
There are endless variations of this strategy. For example, it is a good idea to long-date your longs and short-date your shorts to maximize accelerated time decay in your favor.
In such a scenario you would stay long the six-month put option described above, but sell short one-month options against it, but with a strike 15% out of the money instead of 10%. I could go on and on. That cuts the cost of this hedge by two-thirds.
There are a few qualifications with such a simple hedge. Let’s say that you read the Diary of a Mad Hedge Fund Trader and have a highly concentrated portfolio focused on technology and financial stocks.
In such case, the tracking error between the (SPY) and your portfolio will be large (after all, that is the point), and you may not get all the downside protection you want.
On the other hand, what if we really get the 10% correction? What if the black swans suddenly land in flocks? In that case, the value of your June 2025 $530 puts soar to at least $29, and more likely $32 when you add in the expected effects of rocketing implied volatilities. The value of your hedge rises to $3,200.
Yes, you don’t get complete 1:1 coverage. But it’s better than going into such a route naked, with no downside protection at all.
Let’s say you’re a cheapskate and you want your insurance policy for free. Yes, this can be done.
There is another hedging strategy that is far easier to execute. Just take a long cruise around the world. That way, corrections will come and go and you might not even know about it, unless your butler brings you an online copy of the Wall Street Journal every morning, as mine does.
This is the hedging strategy most of you have pursued for the past nine years and it has worked really well. At least you end up with a nice tan and some pleasant photos.
As for the June 2025 $530 puts, they’re most likely end up expiring worthless, but you’ll sleep better at night. Such is the price of peace.
(NVDA), (MSFT), (GOOGL), (AMD), (META)
You know what keeps catching my attention these days? The way artificial intelligence is sneaking into every corner of the market faster than my daughter can fix those broken calculators in her UC computer science class (more on that later).
Let me throw a number at you that made me sit up during my morning coffee: the global AI market, currently parked at half a trillion dollars, is barreling toward three trillion by 2032.
We're talking about a 20% compound annual growth rate, the kind of numbers that make even jaded old traders like me take notice.
I recently found myself poring over a 273-page congressional task force report on AI (yes, the things I do for you, dear reader).
Between coffee stains and margin notes that started looking like modern art, I discovered 66 findings and 89 recommendations that actually make sense. Imagine that – Congress getting something right about technology!
Speaking of getting things right, the FDA just pulled off something remarkable. They've created what they call "Predetermined Change Control Plans," which is bureaucrat-speak for letting AI medical companies update their algorithms without filing enough paperwork to deforest the Amazon.
This could help save the healthcare industry $13 billion in 2025 alone – not bad for a government initiative.
As expected, the usual names like Nvidia (NVDA), Microsoft (MSFT), Google (GOOGL), AMD (AMD), and Meta (META) have their hands all over this. They're building data centers that use enough power to light up Lake Tahoe twice over, and the numbers justify their aggressive expansion.
Corporate AI adoption shot up 270% between 2015 and 2019, with nearly one hundred million new AI-related jobs expected by 2025.
Even the Pentagon is getting into the game, pouring billions into AI development. Which makes sense – when you're looking at technology that could reshape global competition, you don't want to be left behind.
And the impact is spreading far beyond Silicon Valley. Take healthcare, for instance. Some 90% of U.S. hospitals are planning to implement AI solutions by 2025, driving the healthcare AI market from $12 billion to over $100 billion by 2030.
Remember when hospitals were still using fax machines? Some still are - probably the same ones with waiting room magazines from 2010. But now they're racing from those paper jams straight into AI diagnostics.
The congressional report flags what you'd expect - privacy concerns around AI handling sensitive medical data. After all, we're talking about systems that can access everything from your blood pressure readings to your insurance claims. But unlike those old fax machines, this technology is moving too fast for traditional regulations to keep up.
The task force gets it right: success with AI requires a delicate balance. Just like I tell my daughter about her computer science projects, this isn't just about the technology. It's about how we use it.
That means clear government guidelines, aggressive but responsible innovation from industry, and serious intellectual firepower from academia. Get this formula wrong, and we'll have bigger problems than misrouted faxes.
For those watching this digital gold rush (and I know you are), here's my take: AI isn't just another tech bubble filled with hot air and PowerPoint presentations.
The projected $15 trillion in global economic value by 2030 isn't just a number pulled out of thin air – it's the kind of growth that creates generational wealth opportunities.
Just remember what I always say about transformative technologies: there's a time to go all in (like buying tech stocks in 2009), and there's a time to be strategic.
Right now, we're in that sweet spot where the technology is real, but the market hasn't fully priced in the implications.
Speaking of implications, my daughter just texted me that her next computer science project involves teaching AI to recognize broken circuit boards. Given how fast this technology is moving, I wouldn't be surprised if next semester she's programming AI to fix the circuits itself.
And that's exactly why I'm keeping a close eye on this sector. When college sophomores are doing what billion-dollar companies were struggling with just a few years ago, you know you're onto something big.
Mad Hedge Technology Letter
January 8, 2025
Fiat Lux
Featured Trade:
(BUY THE MICROSOFT DIP)
(MSFT)
How will Microsoft grow their stock in 2025?
In short, MSFT are building what the market wants, and what the market wants are AI data centers.
The stock price should be rewarded if they can deliver these new AI data centers to the market.
The data center increase shows no signs of slowing down and I do believe this puts a floor under tech stocks.
To be honest, there has been a lot of bad energy surrounding the current tech business models because many of them are getting stale.
Why upgrade to the next iPhone when there isn’t much of an upgrade?
The refresh cycle data shows people are standing pat and using their own tech longer and that is bad news for tech software and hardware companies.
So instead of trying to squeeze the remaining juice out of a stale model, beefy balance sheet tech companies are driving full force into AI investment even though this investment doesn’t reciprocate with any sort of revenue stream.
It’s a little bit of a build it and it will come mentality which I do believe is quite risky and at some point, we are due for a heavy selloff.
That selloff could get triggered if the US 10-year interest rate blows past 5.5%, then all bets are off.
Microsoft says it plans to spend $80 billion on building AI data centers this year.
Microsoft has poured billions of dollars over the last two years into Anthropic, as well as Elon Musk’s startup xAI.
Advances by these firms would not have been possible without new partnerships founded on large-scale infrastructure investments that serve as the essential foundation of AI innovation and use.
The $80 billion would reflect a significant increase on the $53 billion capex spend Microsoft made in 2023.
Documents leaked last April revealed it had more than 5GW of capacity at its disposal, with plans to add an additional 1.5GW in the first half of 2025. It is possible this has since been revised upwards as it looks to provide compute power to OpenAI to run ChatGPT and its other AI services, as well as supporting its own Azure public cloud platform.
Part of this is definitely the management at OpenAI namely CEO Sam Altman. He is seen as the avant-garde of AI and the leader of the whole movement. He is demanding a massive build out and investors have largely taken him at this word. Nobody has really questioned him and that stems partly from no one really knows where this AI thing is headed in the future, but we are convinced that buckets of data space are needed for whatever comes next.
My issue is what if the thing that comes next is a cataclysmic letdown, then where do tech stocks head?
Most likely they would head for the gutter.
So we give the benefit of the doubt to this gargantuan AI infrastructure build-out and it feels like we are flying blind in a snowstorm, but that is what the market is telling us and the market is always right until it is not.
Sometimes tech does figure it out, and we are really hoping there is something of great value at the end of the build-out.
Buy the dip in MSFT until the AI infrastructure story is killed off.
(PLAYING GAMES AT THE END OF EACH ADMINISTRATION)
January 8, 2025
Hello everyone
(This is a brief look into the work of Lawrence McDonald’s text How to Listen When Markets Speak: Risk, Myths, and Investment Opportunities in a Radically Reshaped Economy. We learn what some of his research shows, and this piece here is via his blog, The Bear Traps Report.
Over the last 250 years or so, all American political parties have played “games” at the closing stage of each administration. Take 2000, for instance, when the incoming George W. Bush team discovered that every keyboard in the White House and other administrative officers was missing the “W” key.
The outgoing Clinton staff had removed all the “W” keys to annoy the new administration after an extremely contentious election.
The damage was no big deal – around $15,000.
The outgoing Biden administration will be no different. But their mischievous acts may be a little more unpalatable. Under the existing budget, Bien is opening the floodgates with spending.
Spending for 2025 is expected to exceed $2 Tr by the time Biden leaves DC on January 20th. This is over 30% of the annual budget, and Trump will have to cut spending for the rest of the year to stay within the limits of the allocated budget. This could mean a notable slowdown in GDP growth in the first quarters of 2025.
The government’s fiscal year started on October 1st, and Biden could be on course to spend almost $2TR by the end of December and a deficit that may exceed $800bl (+60%y/y).
So, when Trump steps in on January 20th, he has three-quarters left of the government’s fiscal year; by then Biden has possibly spent more than 30% of the total allocated budget.
Lunatics – as Usual – on Capitol Hill
Congress, in its usual fashion, has failed to agree on the next budget, so the government is currently operating under a “continuing resolution” (CR). This continuing resolution means the government is allowed to spend the same amount of money they spent last year, which is $6.75TR. The government’s fiscal year started on October 1st, and Biden is on a run rate to spend almost $2TR by the end of December and a deficit that may exceed $800bl (+60% y/y). So, when Trump comes in on January 20th, he has three quarters left of the government’s fiscal year, but by then, Biden has spent more than 30% of the total allocated budget. This forces Trump to cut spending right off the bat. We estimate spending could drop by $500bl quarter over quarter, or 25% from Q4 to Q1. This is an estimate, and the timing of spending can change. But the fact is that Biden is emptying the coffers before Trump gets in. Every week, more money and weapons are sent to Ukraine, more subsidies are given to semiconductor makers to build plants in the US, and more government employees are hired.
US Yields Surge While Others Languish
Since September, US Yields have surged over 20% on Biden’s sugar high, while Canadian and German yields are down since then, Chinese yields have collapsed, and UK yields are only modestly above the September level.
Government Job Growth Twice the Rate of the Private Sector
Private sector job growth has lagged government job growth significantly in the last year as the government keeps hiring people.
Why is this so Bad?
We believe that this spending deluge by Biden on his way out is partially to blame for the surge in bond yields in Q4. Some may say it’s because of Trump and his promised tax cuts, but the Republican House majority is so slim that it’s unclear how much of a fiscal stimulus Trump is actually able to get through Congress. Also, the incoming Senate majority leader Thune (R, SD) has said he will only get one bill through reconciliation in FY 2025 and another one in FY2026. His priority is on immigration and energy legislation, so a fiscal spending bill might not come until late 2025 or early 2026 if anything. But if yields are being pushed up by all this spending in Q4, then what will happen if spending falls back in early 2025? And what will happen to GDP growth? A $500bl drop in government spending from Q4 to Q1 is the equivalent of 1.7ppt of growth. So, if Q4 nominal growth comes in at 5.7% annualized, this could drop to 4% in Q1 if government spending slows down accordingly.
Treasury’s Reliance on Short-Term Debt Exploded in Recent Years
Election Rigging? We are witnessing a Covid era like spending in 2024 without a pandemic. The Treasury Department has come to rely on short-term bills to fund the government. But with $36Tr of debt, the Treasury has to issue bills almost every day to keep funding the government and to refund maturing debt.
Interest Payments on the Federal Debt Load
2026: $2.1T?
2025: $1.5T?
2024: $910B
2023: $658B
2022: $475B
2021: $352B
2020: $224B
*CBO data, Bloomberg. The average weighted coupon on the U.S. debt load is about 2.7% vs. over 4.5% for 10-year U.S. Treasuries. As bonds mature, they get refinanced at much higher yields.
$10Tr of Debt Refinancing Next Year
In 2024 Treasury faced around $10Tr of maturing debt. To refinance this debt, it issued a whopping $26Tr of bills and bonds. More than 84% of that paper was short-term bills with a maturity of 6 months or less. Treasury keeps re-issuing bills with a maturity of 4 to 8 weeks or 3,4 to 6 months, which are the most popular maturities in a continuing, ever-increasing roll down of the debt, day after day, month after month.
Apple Long-Term Bonds and Interest Rates
ALERT – By issuing nearly a colossal load of extremely short-term bills, Janet Yellen succeeded in suppressing bond volatility in an election year and, in our view, strategically placing that bond market volatility into 2025 after the election. You can “why” see above, she wanted LESS long-term paper in circulation markets in the election year. Now, in 2025 – this paper has to be rolled over and termed out into longer-dated bonds. The USA is behaving like a financially trapped emerging market country. Living on the “front end” of the yield curve is a VERY dangerous game. The Apple AAPL 2.55% bonds due 2060 are trading down at 57 cents on the dollar. If long-term bond yields go to 6%, take a guess where this bond will trade. Near 47 cents on the dollar? Now think of the trillions of USD loans issued in 2017-2021 on bank balance (commercial real estate, mortgages, corporate debt outstanding). Losses are in the trillions of dollars with higher incoming interest rates.
Interest Rates UP – Bond Prices DOWN
Never, ever forget that 6% today is equivalent to the destructive capacity of 10% twenty years ago. Interest rates up, mean bond prices down. A 1% move in interest rates higher today is an entirely different, far more lethal equation.
Incoming Stress Points
In 2025 the U.S. Treasury faces $9.6Tr of maturities in their so-called publicly held debt. In Q1 alone — the government faces $5.58Tr of maturities (bonds coming due, redemption), but 86% of those are short-term bills that the Treasury department rolls over into new 4-week, 8-week, 3,4, or 6-month bills, among others. As a result, almost daily bill auctions are coming to a theater near you, as the Treasury Department mindlessly keeps pushing new paper into the market to pay back the colossal amount of maturing debt.
Is There Any Reason to Buy Treasuries?
The new Treasury department under Scott Bessent may reduce bill issuance a bit and increase coupon paying issuance, just to alleviate some of the pressure on the bills market and extend the duration of outstanding US debt. Now that the big slush fund that bought all these bills, the so-called Reverse Repo Facility (RRP), is close to being depleted, it will be harder to sell all that short-term paper. In addition, Goldman Sachs expects that the Federal Reserve will stop the run-off of treasuries from its balance sheet by the end of January and begin buying treasuries again with the proceeds of the maturing MBS on its balance sheet. As such, the Fed becomes a modest buyer of treasuries next year, which allows the Treasury to increase coupon issuance without disrupting the long end.
One big bullish catalyst for treasuries would be a regulatory change to exempt treasuries from the Supplemental Leverage Ratio (SLR). It is unclear if and when this would be implemented, although Bessent was hinting at regulatory relief for banks to boost banks’ treasury holdings. Exempting treasuries allows banks to hold more Treasuries on their balance sheets without needing to hold additional capital against them, freeing up the capacity for banks to participate more actively in the Treasury market. It’s unclear how much treasury demand that would create, but in 2021 when the temporary SLR exemption was reinstated after COVID, prime dealers reduced their Treasury holdings from $250bl to $125bl in 2 months. A change in the SLR ratio may come but is going to take months before the rules are changed. A phase-out of QT for treasuries would be a more immediate, albeit more modest, relief for the bond market. According to this timeline, the Fed will end up buying $100bl of treasuries in 2025, a big change from the $500bl of treasury sales in 2024.
¹https://www.latimes.com/archives/la-xpm-2002-jun-12-na-clinton12-story.html
The Fed has been Politicized.
We have been very critical of Yellen’s term at the Treasury, but upon some further reflection, we think it’s really the case that Yellen’s only real issue was acting in the short-term interests of her boss and her party as opposed to thinking longer-term about how the government finances itself on a sustainable basis.
Her decision to fund the government with T-bills over duration securities and violate long-standing Treasury Department “norms” was incredibly short-sighted, but as someone who works for the President, ORDERS to follow.
Many have been super critical of her for these decisions because she should know what they would lead to and how really what she (and Powell together) has done is favor asset owners and the wealthy over everyone else in America, exacerbating wealth inequality to precarious levels in this country while still not bringing inflation back down to target. So ultimately, her decisions got her team knocked out of office anyway.
Looking forward, though, the issue is that there is no one in the government who is really thinking about and acting on behalf of the longer-term interests of the country when it comes to how much debt we are raising and how we are financing the government. The myopia about these decisions to get the existing political party in control through the next election is incredibly concerning.
The Fed has said this is not their lane; however, they are elected to 14-year terms and are supposed to be above politics. There are things they could have done to offset the politicization of the Treasury. They chose not to, they continue to protect asset holders and the Treasury market, decisions that really just make them become political as well. They could have better neutralized Treasury’s political decisions through more active QT, actually selling securities instead of just rolling them off, not adding to their duration holdings such that the weighted average maturity (WAM) of their positions is longer than Treasury’s own WAM. Powell’s Fed needs to be getting way more criticism than they are currently about these decisions which have made it harder to bring inflation down for the average American.
So, if the Treasury is not going to think long-term and the Fed is not going to either (the Fed actually is complicit because they don’t allow any real treasury market dysfunction to exist, which would be the way to deal with these long-term issues by having the market/bond vigilantes do their thing), then who will? This is a problem, the bond market is starting to figure it out, term premiums are starting to normalize, and the new administration will have to make some big decisions early on in their term.
Maybe @elonmusk and @DOGE can look into this as well. Someone has to!
QI CORNER
NVIDIA just dropped Project DIGITS, a $3,000 personal AI supercomputer that looks like a Mac Mini but packs 1,000x the power of your average laptop
Powered by the Nvidia GB10 Superchip and based on the NVIDIA Grace Blackwell architecture, this supercomputer can run AI models with up to 200 billion parameters.
The crazy part? You can link two units to handle 405B parameter models.
From this perspective, OpenAI's GPT-4o is around 200B parameters while Grok-1 by xAI has 314B parameters.
Jensen Huang is single-handedly bringing data center-class AI computing to individual users.
Reported by Linas Beliunas
Cheers
Jacquie