¡°One country, one company can¡¯t have complete monopoly over something, because look: it took something like COVID-19 for the whole world to realize, including myself, that even the pharmaceutical industry is dependent on China for some of the most basic things, which should not be the case,¡± according to Akhtar Ali.
He was responding to my question on why India hasn¡¯t matched up to China in terms of technology-related manufacturing, despite having some of the smartest people innovating in tech. Even on the back of the year we¡¯ve all lived through, Akhtar¡¯s assessment couldn¡¯t have been more logical. After all, we know monopolies are bad, they stifle innovation, ultimately hurting the customer and the ecosystem in the long run. But China didn¡¯t become a monopoly overnight.
¡°Why do you think all the multinational companies went into China,¡± Akhtar asked me, rhetorically. ¡°They had expected this huge, billion plus people market will open up and it hasn't opened up.¡±
But de-coupling from China isn¡¯t that easy and comes with its own challenges, money being the biggest one of them all. ¡°Getting businesses to agree to do that -- (look beyond China to make their stuff) -- is hard because as a first step your costs are going to go up. But with some of the geopolitical shifts happening, I think there is a big role for India to play in all of this,¡± Akhtar added.
In two years, this was my second conversation with Akhtar Ali, Vice President of India Systems Development Lab and Technical Computing at IBM. He¡¯s fast approaching his 30th anniversary working at IBM, overseeing development of a unique hardware and software solutions portfolio in the HPC space. His team of hardware-software architects works on the foundational building blocks, the nuts and bolts, bits and bytes, zeroes and ones of some of the most critical infrastructures powering our 21st century world.
Compared to other innovation labs out there, Akhtar Ali told me how IBM India¡¯s Systems Development Lab is unique in the world because it works on all three aspects of the holy trinity -- IBM¡¯s Z mainframe, POWER systems and Storage solutions. And India did a lot of heavy lifting in the latest POWER10 processor which was launched by IBM in August 2020.
¡°We are pretty much involved with all aspects of the processor,¡± Akhtar said, ¡°handling everything from chip development to physical design, the electronic design automation that goes into designing these complex chips, validation of the chips, and building the world¡¯s first enterprise-scale 7nm processor.¡±
For those who don¡¯t know, IBM designs and manufactures POWER series of microprocessors for high-performance computing workloads, especially servers and workstations used widely by the IT industry and beyond. IBM¡¯s POWER10 is the most recent processor in this series, manufactured in partnership with Samsung on a 7-nanometer node, offering world-class data security at multiple levels but also being flexible and open standards-based to enable hybrid cloud for memory-intensive and high-bandwidth workloads.
¡°POWER10 has been in the making for nearly four years and the India team has been involved right from the beginning of the blueprint stage,¡± said Rahul Rao, a Senior Technical Staff Member in Akthar Ali¡¯s team, who also leads the POWER Processor Design Team at IBM India. He joined IBM as a research staff member over 16 years ago, and is still going strong in a leadership role.
¡°Everything from what should be the dimensions of the chip, how much stuff do you want to put in it, figuring out how to put it together in different blocks and trying to make it as packed as possible so that you get the best performance in the smallest possible area,¡± Rahul told me, apart from also figuring out how to validate the POWER10 processor, measuring and testing it across different workload scenarios.
¡°We have a strong design automation team here in IBM India, using our own software to design these next-gen chips,¡± Rahul explained, ¡°and also work closely with worldwide organizations, in terms of how do we get better in terms of software and supported applications? What are the things we can do? How do we not use certain things computationally when they're not needed? How do we make them kick in when they are really needed? How do we put things to sleep and have them wake up without wasting energy? Basically, we asked and also tried to answer all of these questions for POWER10 right here from India.¡±
Adding to what Rahul mentioned, Akhtar further elaborated the role of IBM India engineers¡¯ achievement in POWER10¡¯s design and overall build, contributing techniques that are on the cutting-edge of hardware design and automation.
According to Akhtar Ali, IBM India team used a lot of pioneering AI techniques to make the whole design automation of POWER10 go faster -- right from physical design and timing to making things go faster by learning from simulation runs.
¡°The India team has actually improved the internal IBM design tools by bringing in AI feature functions,¡± mentioned Akhtar. ¡°They¡¯re not looking at AI as an end goal of what the customer will run an application on for a third-party use case, but they¡¯re using the same principle and practices for improving internal processes. That¡¯s one example of how people are being innovative here in IBM India¡¯s lab,¡± he added, underlining how that¡¯s possible only because the India team is involved in all aspects of the chip design process, from hardware blueprints to software stack and beyond.
This also had a remarkable effect on the development team, according to Rahul Rao. ¡°By using the same algorithms in our design classes itself, to some extent as we are, we're running into some kind of bottleneck issues, which gives us an insight into the challenges our customers will face and reveal what are some of the common problems they have, which allows us to course correct accordingly right at the chip design level,¡± Rahul mentioned, while emphasizing how IBM India has put in a lot of innovation into POWER10, which included contributing to ¡°hundreds of patents¡± under processing and review right now.
Time will tell how many of those patents get granted, of course, but patents remain one of the litmus tests of true-blue cutting edge innovation, which involves pushing the needle of progress forward -- another area where Indians seem to be doing remarkably well on the global stage. In fact, IBM inventors from India received over 900 patents in 2019 alone, the second-highest contributor to the global tally after the US.
There¡¯s no doubt India has a lot of incredible talent, which is evident from Akhtar Ali and Rahul Rao¡¯s Indian roots. So why don¡¯t we make these POWER10 processors, or any other cutting-edge processors of any kind, here in India?
¡°We don¡¯t make them in India because we don¡¯t make them anywhere else in the world either,¡± Akhtar responded with a wry smile, explaining how semiconductor fabrication plants (also known as ¡°fabs¡±) are an extremely expensive business, something that IBM doesn¡¯t do by themselves anymore -- they partner with Samsung to manufacture the POWER10 chips, for instance.
¡°We realized that being in the fab business only to build chips for yourself isn¡¯t a good business model because the fixed costs of owning and running a fab are extremely high, and that¡¯s why there are only a few of them in the world,¡± Akhtar emphasized, while listing Intel, TSMC, Samsung and GlobalFoundries as the big chip-manufacturing companies that are still doing it.
Semiconductor manufacturing plants or fabs need billions of dollars to set up, they have very high day-to-day operating costs and need upgrading every few years -- how else will it be at the cutting-edge of microprocessor production, if the fab itself isn¡¯t modernized constantly? Needless to say, all of this sucks in a lot of money.
¡°You not only need the volume but also the ability to put in a lot of fixed capital which is really tough, which is why not just IBM but even Intel hasn¡¯t put a chip-manufacturing or fabrication site here in India,¡± Akhtar pointed out. ¡°No one¡¯s putting a lot of different fabs all over the globe, other than countries where these companies already started building out fabs in to begin with.¡±
Ultimately, it¡¯s all business-driven, Akthar emphasized. ¡°In terms of high-tech manufacturing, the principal determinant of where you invest your money is a function of how cheap it is to produce at scale, because the pieces that you need to put it all together can be gotten from anywhere, including India.¡±
¡°I definitely think it is not a question of technical ability or technical merit as to why we don¡¯t have some of these high-tech manufacturing infrastructure here in India, because it¡¯s all business driven,¡± Rahul Rao added to the discussion, while speaking about the lack of a high-end fab in India. He pointed out how there are certain pockets in India where high-level chip design and manufacturing does happen, while referring to IIT Madras¡¯ Shakti processor in particular.
¡°IBM does encourage people like me to strongly collaborate and help in these initiatives, which in a way provides more momentum to indigenous technology development and intellectual capital build-up which is absolutely critical, ¡± Rahul mentioned. ¡°I go to IIT Madras and teach there myself at least six-seven times a year, thanks to IBM¡¯s CSR initiatives that enable people like me to give back to the local community and make an impact where it matters the most.¡±
Having a lot of Indians in positions of power in technology companies also helps, according to Akhtar Ali, and he underscored this with an interesting little story about current IBM CEO, Arvind Krishna.
¡°Arvind and I both hail from IIT Kanpur, and IBM in the US has something called Employee Charitable Contribution where employees can contribute through their paycheck into their favourite charity and IBM matches that contribution,¡± Akhtar mentioned. ¡°Many years ago, it was Arvind Krishna who managed to put IIT Kanpur as one of the charitable contributions for furthering technical education in India, making it not only easier for all of us at IBM to contribute to IIT Kanpur but also have IBM match our contributions to be able to make an even bigger impact.¡±
¡°In India, IBM¡¯s doubling down on furthering STEM (science, technology, engineering and mathematics) education for girls and women. We are heavily pushing for it, we believe we have some very good women role models within IBM, both at the technical and business leadership level and the executive team, and that¡¯s another thing that¡¯s extremely important for IBM,¡± according to Akhtar Ali.
It doesn't take a genius to figure out India¡¯s high-tech manufacturing process won¡¯t kick into gear any time soon and China will continue to be the biggest game in town. Of course, India needs to stay the course and be in it for the long haul -- not just as an alternative to China for the world, but simply to pursue its own growth story, which promises to be an eventful journey. When India does finally reach the promised land in the future, it will be thanks to all the behind-the-scenes work done by industry veterans like Akhtar Ali and Rahul Rao at IBM along the way.