Lead Apple Cloud Infra & ML products, ex-VP/GM Google, IBM, NVIDIA for AI & GPUs, #DataCenter
Sunnyvale, CA
Joined July 2011
- Tweets17.7K
- Following888
- Followers3K
- Likes22.2K
Pinned Tweet
Super excited to share that my daughters (high school seniors) published this cookbook for teens (and adults!). Kudos to their mom on motivating them that cooking is a life skill. Please do support them by buying a copy (great gift!) & leaving a review
amazon.com/Essential-Teen-Co…
Sumit Gupta retweeted
We’re deepening our collaboration with @googlecloud to give customers more options for building and running AI at scale.
Google TPUs are coming to Oracle Cloud Infrastructure (OCI), marking the first time Google TPUs will be available on a cloud platform outside Google. Customers will have more flexibility to match AI infrastructure to demanding training and inference workloads, optimizing their environments for performance, throughput, and economics.
Learn more: social.ora.cl/6012BHcVbe #AIWorld
Sumit Gupta retweeted
Big news today from Apple: Keynote is getting an open, JSON-based file format (.keyx), plus a Python library to create Keynote files on any platform.
Pages, Numbers and Keynote files today are closed and hard to modify unless you use AppleScript (limited and old) or Shortcuts (a huge pain), so you can’t really use AI to make documents in those formats. This was a HUGE pain inside Apple as well as outside, and most tools worked around it by making a PowerPoint file and importing it into Keynote, which wasn’t perfect.
Creator Studio subscribers on the Mac also get MCP support in Keynote, Pages and Numbers, so you can bring your own AI agent to read, edit and automate all three. Both the new format and MCP support arrive in November.
Bringing an AI-friendly format to Keynote (and I assume Pages and Numbers soon after) is a big positive change.
Huge day for Creator Studio today.
Our new in-app assistants can help you get things done in record time. MCP allows you to bring your own agent if you want. And an open source format for Keynote (and Python library!) so you can create Keynote documents programmatically on any platform.
Super proud of the team. Hope you love it!
Sumit Gupta retweeted
The first iPhone prototype had a plastic screen. Keys in Jobs pocket kept scratching it and he went beserk.
So, Jobs asked Corning to make Gorilla Glass for iPhone. Corning CEO said timeline was impossible. Job told him, “Don’t be afraid, get your mind around it. You can do it.”
Sumit Gupta retweeted
The biggest mistake tourists make in Japan is dragging a fifty-pound suitcase up and down subway stairs.
Watch how Japanese travelers board the bullet train. They walk on with a cup of green tea, a small shoulder bag, and both hands free.
Where is their luggage? It is already in the next city, waiting for them.
In Japan, the system is called Takkyubin, recognizable everywhere by a logo of a mother black cat carrying her kitten gently in her mouth.
When you check out of your hotel in Tokyo at eight in the morning, you hand your heavy suitcase to the front desk, fill out a tiny slip, and pay about twenty dollars. You can even send it straight from the corner convenience store.
Then you spend the entire day wandering through mountain temples or eating street food with nothing on your back.
When you walk into your traditional inn in Kyoto that evening, your suitcase is already standing neatly in your room, wiped clean, with the handle turned toward you.
In fifty years of handling millions of bags a year, lost luggage is virtually unheard of.
A society runs smoothly when you can hand your belongings to a stranger at a convenience store and know they will beat you to your hotel.
Sumit Gupta retweeted
How Meta and Microsoft are working to wean their employees off Anthropic’s Claude
Sumit Gupta retweeted
The reason frontier LLMs will never be profitable is because the minute they stop spending, the free models catch up.
You're in a constant loop of increased spending to stay ahead, while everyone behind you catches up at a fraction of the cost by using your model.
LLMs are just commodity with no moat.
Sumit Gupta retweeted
Paper straws worse for the environment than plastic ones, says study trib.al/TEphuYk
Sumit Gupta retweeted
For decades, we called it the Indus Valley Civilisation.
But archaeology, palaeochannel mapping, geology and satellite imagery point to a wider, maybe clearer picture: the Sindhu–Sarasvati Civilisation.
A thread on a civilisation, a river, a script and a narrative worth re-examining. 🧵1/21
Sumit Gupta retweeted
读 Anthropic 的招股书，我看到了音乐什么时候会停的信号。
它卖给客户的承诺按年到期，向供应商买算力的承诺却锁到2030+，而且大部分属于take or pay （不管用不用都要付）。
26年上半年，这些承诺从一千亿涨到近六千亿。自己还在招股书里写了：“增速会下降，可能比预期更快。”
于是我做了一个 scenario analysis，model 了一下在不同的增速下，Anthropic 未来几年的收入，对比它的这些承诺支出。哪怕是偏乐观的假设，到2027、2028年，光算力付款就会超过全部收入，缺口要靠融资填（借钱）。
填缺口的钱主要来自 Google、Amazon、Broadcom 等。这些公司既是供应商，又是股东，又是债主，其中的云厂商还是竞争对手。
问题在于，S-1 中披露放款条件和 Anthropic 收入的增速是绑定的：增长快的时候，钱可以随时提。增长慢的时候，提款资格同时失效。说简单一点就是钱在你最不需要的时候最多，在你最需要的时候反而最少。救命钱的提款条件和病因绑在一起。
这就暴露了循环融资的问题，预计最快在27 年下半年就会进入结构性问题（一旦增速低于某个门槛，这些条款会自动把问题放大）。Anthropic 自己不会先倒，它可以不续那些“,可取消”合同。
所以，音乐停止的顺序是：
1. Frontier Labs 增速放缓
2. 供应商融资收紧
3. Frontier Labs 不续约
4. 最后新云还不上债
招股书第140页有一句话："我们的算力承诺通常必须履行，无论收入或利用率如何；而我们获得额外资本的能力，取决于许多我们无法控制的条件。"
这句话的前半部分是确定的，而后半部分是不确定的。在我看来，整个 AI 产业链现在在赌的，就是这后半句话。
Sumit Gupta retweeted
Absolutely wild detail in Microsoft's 60,000-person buyout:
Copilot, Xbox and Windows teams are fully eligible. Data center staff are excluded.
So the people building the AI can take the money and leave. The people building the buildings for the AI can't.
27% of the company, two days before earnings, in a year Microsoft is guiding $175B in capex. Imho this is the clearest picture yet of where big tech's money is going.
$MSFT offers 60,000 employees a voluntary exit.
60K Y’all!!
First time this has ever been offered at Microsoft and it will affect employees in India, The United States, Canada, and the UK at level 70 or below…
Employees find out in a few hours if they are being targeted for this.
This just makes me think that H127 layoffs sound intense. Is this a pre-layoff round?
Also makes me wonder if they changed some scheduled layoffs due to be called out by @VP. Peace offering? 🤔 Timing is just interesting.
More to come.
Sumit Gupta retweeted
Worth a read $ANT
I always respect David's opinion on IPOs, even if I don't agree with them.
"There’s 40%+ Downside If Growth Slows, Even with Higher Margins"
"There’s 90%+ Downside If Anthropic Generates the Same Profit as Meta"
@NewConstructs ran a reverse DCF on the proposed valuation and a few other scenarios - results in our pre-IPO report:
newconstructs.com/anthropic-…
Sumit Gupta retweeted
Remembering some advice I received from a dotcom graybeard:
In a layoff tsunami you want to be first to look for a job. The guy who is "valuable" and hangs on to the very end finds himself looking for a job in a completely saturated market.
Go early.
Sumit Gupta retweeted
Ask yourself why Anthropic magically needs hundreds of billions of dollars a year for compute after IPO, when they were paying essentially nothing before.
Ask yourself why OpenAI was projecting $300 billion in losses on compute through 2030 when Anthropic was claiming to be profitable.
It’s the same business.
Ask yourself why the hell wouldn’t Anthropic just pay the $7.5 billion of original convertible notes with $120 billion cash on hand rather than letting it convert to $108 billion of equity.
Don’t be an idiot.
This quoted post is unavailable.
Sumit Gupta retweeted
This may seem like a lot until you realize that Microsoft employs 17,000 people just to run LinkedIn, which seems completely insane to me
BREAKING: Microsoft, $MSFT, is offering buyouts to 60,000 workers as part of the company’s efforts to cut costs amid AI adoption, per Bloomberg.
Personnel on select teams in the US, Canada, India, and the UK at a vice president level and below will be eligible.
This reflects ~27% of Microsoft's total workforce.
Sumit Gupta retweeted
This is staggering. Every employee who receives the offer will have to think, “If I turn this down, do I get a pink slip on much worse terms next year?”
BREAKING: Microsoft, $MSFT, is offering buyouts to 60,000 workers as part of the company’s efforts to cut costs amid AI adoption, per Bloomberg.
Personnel on select teams in the US, Canada, India, and the UK at a vice president level and below will be eligible.
This reflects ~27% of Microsoft's total workforce.
Sumit Gupta retweeted
🚨 BOMBSHELL COURT EVIDENCE EXPOSES INSIDER INFLUENCE OVER MIT'S TENURE-REVIEW PROCESS IN THE DAVID M. SABATINI ORDEAL.
🔱 NEW REVELATION — Court evidence exposes MIT Professor Sangeeta Bhatia's involvement in the selection of Sabatini's tenure-review committee
On 27 September 2021, Kristin Knouse exchanged messages with Izabella Pena following a conversation with @MIT Professor Sangeeta Bhatia.
The court-disclosed messages reveal Kristin Knouse telling Izabella Pena:
🔴 “Sangeeta just called”
🔴 “they even asked for insight into who to put on the committee”
🔴 “AND she said she will go to Rafael Reif’s office and protest the decision if and when needed.”
Pena responded:
“At least they are asking her insight”
Knouse replied:
“Exactly”
“This is good”
⚠️ The evidence exposes a troubling channel of insider influence surrounding MIT's tenure-review process.
These contemporaneous communications document that Knouse understood Bhatia to have been consulted about the committee's composition and to be prepared to challenge its eventual decision directly with MIT President Rafael Reif.
Rather than an independent review insulated from interested parties, the messages reveal access to the committee-selection process by an academic aligned with the complainant, raising serious concerns about institutional impartiality and conflicts of interest.
This disclosure adds another significant piece to the growing documentary record surrounding the institutional handling of Professor @DMSabatini's case.
📑 REVIEW THW ORIGINAL COURT EVIDENCE BELOW.
🔱 ScienceGuardians™ Legal and Investigative Team
MORE COURT EVIDENCE. MORE REVELATIONS. STAY TUNED. ⚠️
@SciGuardians 🔱
🔱 Part 5 — Joint Appendix JA004741: Court evidence reveals reported support and pressure involving senior MIT faculty
🚨 Court documents reveal that Kristin Knouse described support from influential MIT faculty members during the proceedings concerning Professor David M. Sabatini.
⚠️ In messages dated 13 October 2021, Knouse identified support attributed to senior MIT figures and described their willingness to approach MIT President Rafael Reif to influence the institutional outcome.
These communications document efforts to mobilize influential individuals within MIT during Sabatini's tenure-review proceedings.
📑 THE ORIGINAL COURT EVIDENCE IS PRESENTED BELOW.
Sumit Gupta retweeted
A king travelled to another world for what seemed like moments. When he returned, 116 million years had passed.
There is a super nuts story preserved in ancient Indian religious literature that describes something similar to what physics calls time dilation.
The story appears in the Vishnu Purana and the Bhagavata Purana.
According to the story, King Kakudmi took his daughter Revati to Brahmaloka, the realm of the creator deity Brahma, to seek advice about a suitable husband.
When they arrived, Brahma was listening to celestial musicians.
They waited for the performance to finish.
Nothing particularly unusual happened during their visit. They listened to the music, approached Brahma and explained why they had come.
Then Brahma laughed.
The men Kakudmi had been considering for his daughter were dead. Their children were dead. Their descendants had disappeared.
The Bhagavata Purana, Book 9, Chapter 3, Verse 32, gives a crazy specific number.
Twenty seven complete cycles of the four yugas had passed on Earth.
In traditional Hindu cosmology, one cycle represents 4,320,000 years.Multiply that by 27.
116,640,000 years.
A father and daughter leave Earth, visit another realm, experience what seems to them like a brief interval, and return to find that more than 116 million years have elapsed.
The Vishnu Purana, Book 4, Chapter 1, is interesting as it explicitly states that Kakudmi perceived the passing of entire ages as though they were a moment.
This was being described in Sanskrit literature long way way humanity developed modern theories of spacetime.
For context, the Bhagavata Purana is commonly dated to approximately the eighth through tenth centuries CE, although its traditions may be older. The Vishnu Purana also preserves the account in a different literary form.
These are not twentieth century stories retrofitted with scientific language.
Modern physics established that elapsed time is not necessarily identical for every observer. Motion and gravity affect how time is measured.
Does that mean the people who preserved this account understood relativistic physics? Of course that has not been established.
But they were describing an extraordinary difference in experienced time between two realms, with numerical periods that extend into millions of years.
We can interpret Brahmaloka as a divine realm, another plane of existence, or a place entirely outside the ordinary human experience of time.
What inspired people more than a thousand years ago to describe two individuals leaving our world, experiencing a brief visit elsewhere, and returning after entire geological ages had passed?
Was this simply a sophisticated religious illustration of cosmic time?
Or were they preserving an idea about the nature of reality that we are only beginning to understand?
I don't think we should confuse an ancient account with scientific evidence.
But I also don't think we should overlook how extraordinary the story is.
Sumit Gupta retweeted
If I had to start option selling from scratch… this is what I’d do:
(as someone who made $254,000 last month)
First thing you need to understand is that option selling isn’t “trading”...
Day trading is trading.
Staring at charts for 14 hours hoping a candle does something is trading.
Option selling is a system.
• Set your rules
• Follow them
• Collect premium
• Go live your life
The sooner you stop treating it like gambling, the sooner it starts paying you like a business.
So here’s exactly what I’d do from day one:
Step 1: Learn ONE strategy.
Start with cash-secured puts.
You are selling insurance on stocks you’d want to own anyway.
You get paid premium up front, and if the stock drops to your price…
Congratulations, you just bought it at a discount.
That’s the concept. Don’t overcomplicate it.
Step 2: Pick 3-5 tickers. MAX.
Not the entire S&P.
3-5 quality names you’d happily hold for a year.
Think Pepsi. UNH. Verizon.
Companies that aren’t going to zero overnight.
“But Matt what about that one AI stock my buddy-”
NO.
Quality tickers only. Boring ones.
That’s where the money is.
Step 3: Tape these rules to your monitor
• 5% return on capital
• 30-45 DTE
• Close at 50% profit
These 3 rules alone will keep you out of more trouble than any course on the internet.
Combine that with only trading quality tickers?
You're already ahead of 99% of traders.
Step 4: Know WHEN to enter.
My rules are simple:
#1. Enter on red days.
The market is pulling back, premiums are inflated, and you’re selling fear to people who are panicking.
#2. Confirm that daily AND weekly charts are green on the bigger trend.
A red day inside a green trend is a gift…
A red day inside a red trend is a trap.
#3. IV at 70%+.
That’s your sweet spot where you’re getting PAID for the risk.
Below 50% the juice isn’t worth the squeeze.
Stack all three and you’ve got an almost perfect entry.
Step 5: Get comfortable with assignment.
This is where 90% of beginners panic.
The stock hits your strike, you get assigned, and you think something went wrong.
Nothing went wrong.
You picked a quality ticker remember?
You WANTED to own this company.
Now you sell covered calls against those shares and collect premium on the other side.
That’s the wheel.
CSPs on the way in… covered calls on the way out… premium collected the entire time.
Step 6: Stop checking your account
I’m serious.
Set your trades. Set your rules. Walk away.
your P&L doesn’t move faster because you’re staring at it.
Go to the gym.
Take your kids to the park.
Do literally anything else.
The worst trading days of my life happened when I sat at my laptop all day.
Why?
Because then you start to second-guess yourself...
"maybe just one more trade today"
Before you know it, your account's bleeding red.
Step 7: Once your CSP income is consistent… THEN look at LEAPs.
This is where people get it backwards.
They want the 200-300% plays before they’ve made a single dollar in premium.
Your puts are your paycheck.
Steady and predictable every month.
LEAPs are your lottery ticket...
But you fund them with HOUSE MONEY from the puts.
That’s two gears working together.
Income and upside.
Don’t skip to gear two before gear one is running.
That’s it. That’s the entire system.
• CSPs on quality tickers.
• Covered calls when assigned.
• Premium reinvested.
• LEAPs when you’re ready.
I’ve taught this to thousands of students and the ones who make it all have one thing in common…
They kept it simple.
Send me a DM with the word “SYSTEM” and I’ll show you how it works in depth.